Researchers from Five Element Analytics and Hofstra University are preparing to present their findings on the development of a simulation to simulate United States personal income tax returns. The findings are to be presented at the Northeast Decision Sciences Conference in Philadelphia on April 4th – 6th, 2019.
The purpose of the simulation was to provide a reasonable representation of the US income tax returns for learning accounting methods and examine the impact of policy changes on the tax code. The need for simulation data across many domains has been researched extensively, and the intersection between data science and business presents a unique opportunity to create data, where only summary information, such as percent of returns by state, has been provided.
The purpose of any simulation is to provide researchers, policy makers and students data, by which they can conduct tests and experiment with uncertain conditions. The simulation developed, after over a year of detailed research and trial, draws upon economic theory and current practices as well as guidelines from the IRS Statistics of Income Division. Multiple measures of error were calculated, including the number of returns within each AGI (Adjusted Gross Income) bucket, the mean and median of the simulated returns within each bucket, as well as the sum amount of income within each bucket.