Methodology

Everyone agrees that our tax code is too complex. Few agree on how it should be simplified. It's the product of many years of political and bureaucratic processes, and as such, it is full of handouts to special interests, loopholes, and inconsistencies that make the whole things difficult to understand and even more difficult to fix. Simplifying or fundamentally changing how we tax individuals is a very worthwhile project, but beyond the scope of this web application (for now).

Think of this tax plan creator as a back-of-the-napkin sort of calculation. It's a great place to start thinking about who and how much we tax, but there are much more precise methods for calculating the complex effects of changes to the tax code, and many more things that could (and should) be done to the tax code besides changing tax rates.

For example: to calculate the tax burden on the average taxpayer who earns $20,000 per year, we don't calculate specific deductions or credits, we simply calculate the average amount of credits and deductions for that income level, and calculate taxes from there. One should not understand this calculator to say that under such-and-such a tax plan, a taxpayer will certainly owe X. Rather, on the whole, taxpayers with a certain income will pay an average of X.

The way the IRS and the Congressional budget office actually simulate tax code changes is by using a microsimulation of the effects of any changes. This means they actually calculate taxes for approximately 144,000 representative taxpayers, and use that specific data to extrapolate the effects on the whole tax base. This gives a far more accurate picture of both individual effects as well as aggregate effects on the entire tax base.

In the future, I would love to update this calculator to run on-demand microsimulations, but for the time being, this simulation is based on broader generalizations of the tax base synthesized from the IRS's Statistics of Individual Income.

If you would like to learn more about microsimulating the tax code, and check out projects doing much more detailed analysis: check out these resources: