A Data-driven Approach towards Combatting Social E-commerce Transaction-based Tax Evasion

With the advent of the digital economy, the opportunity and feasibility of hidden economy activities have expanded to the digital realm. Like their traditional forms, digitalised hidden economy transactions also led to serious tax evasion, here facilitated by online communication on social media platforms, enabling social e-commerce. Social media platforms are web-based technologies that enable the development and deployment of social media solutions and services. The functions of social media platforms make it easy for unregistered market participants to conduct hidden business activities without paying tax. The Daigou Industry is a representative example of the social e-commerce business model. This thesis is motivated by the phenomenon of non-payment of taxes by cross-border e-commerce Daigous, it illustrates the significance of transaction-based social e-commerce tax evasion and defines that tax evasion arising from social e-commerce activities are the results of the digitalisation of the hidden economy. Transaction-based tax evasion in the hidden economy, a well-known and hard to detect activity continues to plague frontline tax administrators. These activities have now moved online, hidden in social e-commerce facilitated by social media platforms, such as WeChat and Instagram. Navigating the digital economy is now well and truly part of the tax administration program in most jurisdictions. The tax authorities, however, have raised the alarm and admitted that they have found no solution to detecting the increasingly rampant tax evasion activities on social e-commerce platforms, leaving a large gap in the literature. This thesis explores why those digitalised hidden transactions at present cannot be detected by tax authorities. It concludes that in order to combat tax evasion arising from online cross-border goods transactions, a data-driven approach brings big data to the rescue. This thesis presents big data analytics as a realistic and realisable research technique for tax policy and administration research, by an illustration of adopting machine learning to detect potential tax evasion audit or investigation targets from social e-commerce, where Instagram is as an example in the social e-commerce marketplace