Government agencies that develop taxation strategies have a tough job – aside from being a highly complex project with numerous moving variables, it’s challenging to find supporters of the work. It’s a rare person who enjoys the process of being taxed. That doesn’t mean that developing tax policy can’t be improved in ways that benefit the citizens who are subject to taxation.

Researchers from the University of Nebraska-Omaha (UNO) used their ACCESS-allocated time on Purdue’s Rosen Center for Advanced Computing (RCAC) Anvil supercomputer to study how to better model the outcome of different taxation strategies. Zhigang Feng, a professor in the Department of Economics at UNO, worked with collaborators at multiple institutions to refine taxation models using machine learning.

The problem with traditional models is they fail to account for the way households allocate their spending. You’re in a tax bracket based on your income, but how much of your income needs to go to housing depends on where you live. What you spend on groceries may depend on special dietary needs. Perhaps you’ve set aside extra money in savings, anticipating college expenses. These differences are referred to as household heterogeneity, meaning that each household has a unique set of circumstances that can significantly impact how taxation strategies affect them.

That’s a lot of variables to consider, and it creates an almost infinite number of possibilities. Using artificial intelligence allows researchers to finally include household heterogeneity in their taxation strategy models.

“This problem isn’t something traditional numerical methods in the standard economist’s toolbox can handle – even with a handful of CPUs using MPI, let alone an average computer,” said Feng. “We needed multiple GPUs running in parallel to harness the optimization power of modern AI techniques, and we needed them on demand. We also required a machine with massive memory to store the state of every simulated individual. Thankfully, Anvil was able to provide us with both.”

To solve these models, we needed Anvil. There’s no question – without it, this is not something we would have been able to achieve.

–Zhigang Feng, University of Nebraska-Omaha

You can read more about this research in the original article posted here: Anvil and AI used to solve for best taxation strategies.

If you think your research would benefit from compute resources, you can find out how to get started with ACCESS here. ACCESS has allocations for all types of research, and U.S.-based educators can even get an allocation to bring high-performance computing into the classroom, all at no additional cost to you.

Resource Provider Institution(s): Purdue’s Rosen Center for Advanced Computing (RCAC)
Resources Used: Anvil
Affiliations: University of Nebraska-Omaha
Funding Agency: NSF 
Grant or Allocation Number(s): SOC250017

The science story featured here was enabled by the U.S. National Science Foundation’s ACCESS program, which is supported by National Science Foundation grants #2138259, #2138286, #2138307, #2137603, and #2138296.