How do firms exploit their business network for tax evasion? Evidence from Mongolia

我将纳税申报表数据集的覆盖期延长至2020年，目前它们覆盖的时间范围为2014年至2020年。此外，我还获得并更新了2020年公司的所有权信息，以及2016年的地理信息系统位置。我清理了数据集，使用掩蔽的公司税务ID合并了它们，并检查了数据集的一致性。此外，我使用运营审计和增值税发票数据集识别了与逃税公司处于同一贸易网络中的公司。我开始使用机器学习算法分析识别公司集群，该算法基于税务审计结果与公司各种纳税申报表（例如企业所得税和增值税申报表）信息之间的过去关系。

I extended the coverage period of the tax return datasets until 2020, and currently they cover the period between 2014 and 2020. Additionally, I obtained and updated firms’ ownership information for 2020, as well as their GIS location for 2016. I cleaned the datasets, merged them using the masked firm tax IDs, and checked the consistency of the datasets.Furthermore, I identified the firms in the same trade network with tax evading firms using the operational audit and VAT invoice datasets. I started the analysis of identifying the firm clusters using machine learning algorithms based on the past relation between tax audit results and information on firms’ various tax returns such as corporate income tax and VAT returns.