This paper presents an innovative approach to enhancing the Current Population Survey (CPS) with tax information from the IRS Public Use File (PUF). The methodology is technically sophisticated and addresses a real need in the US tax policy microsimulation community. However, I have several concerns that should be addressed before publication.
The paper would benefit from a more thorough comparison with existing tax microsimulation approaches. As someone who has developed comprehensive federal and state tax calculators, I'm particularly interested in understanding:
- How does the enhanced dataset perform when used with different tax calculators (e.g., TAXSIM, Tax-Calculator, my own models)?
- What are the implications for state tax modeling, given that the PUF lacks state identifiers?
- How do the imputed itemized deductions compare to those derived from state tax return data?
The paper uses a 2015 PUF to impute values onto a 2024 CPS. This 9-year gap raises concerns:
- Tax behavior has changed significantly post-TCJA (2017). The predictive relationships learned from 2015 data may not hold in 2024.
- The paper should discuss how structural tax reforms affect the validity of the imputation approach.
- Consider validating against intermediate years where both CPS and more recent administrative data are available.
The methodology focuses heavily on federal tax variables, but state and local taxes are crucial for comprehensive policy analysis:
- How well does the QRF approach preserve state-level variation in tax variables?
- The SALT deduction calibration is mentioned, but more detail on state tax modeling capabilities would strengthen the paper.
- Consider adding state-specific validation metrics beyond population totals.
Capital gains realization is notoriously difficult to model and highly concentrated. The paper should:
- Provide more detail on how capital gains distributions are preserved through the imputation process
- Discuss the treatment of carried-over losses and basis
- Validate specifically against capital gains realizations by AGI class
- Table formatting in the results section could be improved for readability.
- The discussion of behavioral responses (or lack thereof) should be expanded.
- Consider releasing a subset of the enhanced data for replication purposes.
- The bibliography should include more recent work on tax microsimulation post-TCJA.
This is solid work that makes a valuable contribution to the field. With revisions addressing the temporal consistency issues and expanding the state tax discussion, this paper would be suitable for publication in IJM. EOF < /dev/null