Tax Evasion Simulation Dataset
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/iassael/tax-evasion-dqn
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是由一个模拟模型生成的，该模型运用Q-learning方法，探索在政府控制的税收环境下，自利且风险规避的企业的逃税行为。该数据集可作为评估税收政策的测试平台，帮助理解风险规避对逃税策略的影响。该数据集涵盖了5万次企业演变模拟，每次模拟持续250个时间步长，其任务是在不同税收政策情景下模拟和预测企业的逃税行为。
This dataset is generated by a simulation model that adopts the Q-learning method to explore tax evasion behaviors of self-interested and risk-averse firms under a government-controlled tax environment. It can serve as a testbed for tax policy evaluation, aiding in understanding the impact of risk aversion on tax evasion strategies. The dataset includes 50,000 firm evolution simulations, each lasting 250 time steps, with the task of simulating and predicting firms' tax evasion behaviors across different tax policy scenarios.
提供机构：
Authors of the paper
