For many Americans, doing your taxes isn’t all that complicated. It’s just data entry.
The actual work of doing your taxes mostly involves rifling through various Internal Revenue Service forms you get in the mail. There are W-2s listing your wages, 1099s showing miscellaneous income like from one-off gigs, 1098s showing mortgage interest or tuition payments, etc.
But here’s the thing about those forms: The IRS has them, too. For many people, the IRS has all the information it needs to calculate their taxes, send taxpayers a filled-out return, and have them sign it and send it right back to the IRS if everything looks in order.
This isn’t a purely hypothetical proposal. Countries like Denmark, Belgium, Estonia, Chile, and Spain already offer such ”pre-populated returns” to their citizens. And a new paper estimates that at least 41 percent of American households — some 62 million tax filing units — could have their entire tax returns handled this way with no further intervention necessary.
Tens of millions of unnecessary returns
The paper is by four economists: Lucas Goodman and Andrew Whitten at the Department of the Treasury’s Office of Tax Analysis, Bruce Sacerdote of Dartmouth, and Katherine Lim of the Minneapolis Fed. Half the authors working at the Treasury helps explain the dataset the paper uses: a randomized, representative sample of actual tax returns filed in 2019. The IRS strictly regulates who gets to use this kind of granular tax data (it must be for tax policy work), but it’s a goldmine for those researchers.
In this case, the IRS data let the authors actually generate “pre-populated returns” for taxpayers, based on information the IRS already knew, and then compare those returns to the ones actually filed by taxpayers. If they match, that means a pre-populated return policy could work for that person.
“A pre-populated return is deemed successful if its calculated tax liability is approximately equal to the tax liability actually reported on the 2019 tax return,” the authors explain. This was one of two methods they used; the second sorted through the IRS returns looking for complications that would prevent a pre-populated return from being correctly compiled. That approach tended to produce higher estimates of how many returns could be put together automatically.
The former, more conservative approach found that 41 percent of returns, representing 62 million tax units, could have accurate returns pre-prepared by the IRS in this fashion. (A tax unit could be a single person, a single parent-headed family, a married couple and their offspring, etc. — whoever’s represented by the tax return.) The less conservative approach, counting everyone without complications that might prevent an automatic return, puts the number at 73 million returns, or 48 percent.
Pre-populated returns could also help people who aren’t currently filing taxes. In the US, many people are not required to file an income tax return, usually because they earn too little money to trigger that requirement or because the money they do get is from a partially exempt source like Social Security. But those people often would benefit from filing a return because of benefits like the earned income and child tax credits. Those credits are refundable, meaning that you don’t have to have a positive income tax burden to receive them; the earned income tax credit (EITC) in particular is designed to mostly go to low-income people who don’t earn enough to owe income taxes.
Despite those benefits, some 22 percent of eligible taxpayers don’t claim the EITC in a typical year; by one estimate, two-thirds of those not receiving the benefit didn’t get it because they didn’t file a tax return. The bundling of social assistance programs with a complex tax code places significant burdens on less-wealthy Americans trying to access those programs.
So the authors of the automatic filing paper estimated how many non-filers could get tax benefits under an automatic filing system. They estimate that 7.2 million tax units who aren’t required to file are owed refunds, averaging some $411 each. Those units would be likelier to get their refunds under a pre-populated filing system.
Ending tax returns … for everybody?
For the tens of millions of households for whom pre-populated filing works, it could be a huge leap forward. But 41-47 percent of households is not a majority, and in an ideal world, the other 53-59 percent of tax units would be able to benefit from a system like this too. So what are the barriers preventing them?
The paper’s appendix table A2 estimates the share of returns with different attributes that prevent a pre-populated return from working. The most common, affecting 16.2 percent of returns, is Schedule C or self-employment income: People have a different estimate for their earnings from self-employment or odd jobs than the 1099 forms sent to the IRS indicate. They might have significant business expenses or jobs that didn’t trigger a 1099 form that alter their actual taxes due.
The next most common, affecting 10.9 percent of returns, is itemized deductions. These have become much less frequently done since the standard deduction was increased by the Trump tax bill in 2017, but almost everyone who itemizes claims the charitable deduction or the state property tax deduction. Both of those rely on information that isn’t consistently reported to the IRS, so they can’t be included on pre-populated returns.
Both of those are tricky issues to get around. Especially with the rise of “gig economy” employers like Uber, Lyft, and DoorDash that issue 1099s and treat employees as contractors, more and more low-income people are relying on self-employment income where discrepancies can arise that make auto-filing impossible. You could resolve the itemized deduction issue by eliminating itemized deductions, but I somehow doubt the people whose taxes you’d simplify in the process would thank you for it.
Other problems, though, might be easier to fix. A significant share of taxpayers had wage income that was different from what their W-2 forms indicated; better wage reporting requirements for businesses might get around that. Difficulties determining what share of pension income is taxable also came up a fair amount, which a simpler pension taxation system might address. As a volunteer tax preparer, I’ve had pension issues come up a lot and our current system is mindbogglingly complex. I love thinking about taxes and, nonetheless, learning the “simplified method” for pension taxation made me want to die.
But even if “only” two out of every five returns can be done by the IRS automatically, it’s worth asking: why aren’t they? Even if “only” 62 million households would benefit, that would still save a huge amount of time and angst every year, and make tax season run much more smoothly.
The IRS estimates that the average non-business filer spends nine hours a year filing their 1040. Even if we assume returns capable of being auto-filled are less complex and only take half as long, that adds up to 279 million hours of life, or nearly 32,000 years of life, not wasted if 62 million filers were able to auto-file their taxes. Sounds nice!