Refundable tax credits can play an important role in the financial lives of low-income families. The two that provide the most support are the earned income tax credit (EITC) and the child tax credit (CTC). Refundable tax credits differ from most programs that provide income support to low-income families in two significant ways: eligibility is determined on a calendar year basis and benefits are largely received as a single payment between February and May, after the year has ended and a tax return has been filed. Many benefit programs that operate outside the tax system consider periods that are much shorter when determining eligibility (and are not tied to the calendar year) and they typically pay benefits monthly.

In 2021, Congress experimented with a new mechanism to deliver the CTC, mandating that it be delivered as monthly payments paid in advance of filing a tax return from July through December. Monthly payments of the CTC were correlated with a drop in food insecurity among families with children relative to families without children. They also coincided with a decrease of 3 percentage points in the monthly poverty gap between Black families and white families. After reporting receipt of the monthly payments, many adults living with children favored advanced monthly payments, especially those with lower incomes.

How well the IRS can accurately predict credits will be key to the success of advance payment programs. Advancing too much credit puts families at risk of having to repay the credits at tax time unless a robust “hold harmless” provision is put in place that would limit how much errantly advanced credit must be repaid. Advancing too little credit means families miss out on needed support.

Using data from the 2018 Survey of Income and Program Participation (SIPP, which has monthly income amounts for 2017) and the National Bureau of Economic Research TAXSIM model, we estimate how accurately data from the first quarter of the year can predict credit amounts a person will ultimately qualify for based on their annual characteristics. Low-income people are more likely than higher-income people to experience financial difficulties, so we focus our analysis on families with children at some point in the year with incomes below 200 percent of (twice) the federal poverty level (FPL). If credits can be accurately predicted using data from the first quarter of the year, it may present a path forward toward advance payments of credits.

Accuracy of Credit Estimation

For most families, one quarter of data is sufficient to accurately estimate refundable tax credits. If number of dependents, filing status, and household income for families are stable enough throughout the year, the IRS could use information from the first quarter of the year to accurately estimate tax benefits a family will qualify for over the course of the year. Among all families with a dependent at some point in the year we find the following:

• 81 percent will be stable enough that first-quarter data will accurately predict their EITC (which we define as within 10 percent of the actual credit for which a family will be eligible)
• 75 percent will be stable enough that first-quarter data will accurately predict their CTC under the 2017 Tax Cuts and Jobs Act, which changed the CTC rules from 2018 through 2025 (referred to as 2018 law in this analysis)
• 77 percent will be stable enough that first-quarter data will accurately predict their CTC under the 2021 American Rescue Plan (2021 law)

Among low-income families with a dependent at some point in the year, we find the following under the same assumptions:

• 65 percent will be stable enough that first-quarter data will accurately predict their EITC
• 69 percent will be stable enough that first-quarter data will accurately predict their CTC under 2018 law
• 79 percent will be stable enough that first-quarter data will accurately predict their CTC under 2021 law

In general, EITCs are easier to predict than CTCs because, among all families with dependents, only about 40 percent are eligible for the EITC. First-quarter data will predict most families will be ineligible for the EITC. Around 90 percent of all families with children are eligible for the CTC, and predicting some credit is more difficult than predicting no credit.

However, overestimating credits will still happen. Among families with children at some point in the year, for 10 percent of all families and 18 percent of low-income families the first quarter of data leads to predicted EITCs higher than the credit they will qualify for at the end of the year. Similarly, the first quarter of data leads to CTC predictions that are too high for 12 percent of all families under the 2018 version of the credit and 13 percent under the 2021 version. Among low-income families, those numbers are 12 percent for the 2018 CTC and 11 percent for the 2021 CTC.

Underestimating credits will also happen. Among all families with children at some point in the year, we estimate first-quarter data would underestimate the EITC for 9 percent of all families and 17 percent of low-income families. First-quarter data would underestimate the CTC under 2018 law for 13 percent of families with a child at some point in the year. These data would underestimate the CTC under 2021 law for 10 percent of these families. Among low-income families, those shares are 19 percent for the 2018 CTC law and 9 percent for the 2021 CTC law.

Designing Credits for Advanced Payments

Phase-in and phase-out regions can complicate predictions. For example, most low-income families with children would be eligible for the maximum 2021 CTC whereas eligibility for the 2018 CTC phases in with income—some low-income families are eligible for no credit and many are eligible for a credit smaller than the maximum credit. Eliminating the phase-in of the 2021 CTC makes it much easier to predict compared to the EITC or the 2018 version of the CTC because a drop in earnings does not affect the credit. The relatively high phase-out range for both versions of the CTC makes it unlikely that a family will have an income jump big enough to make them ineligible under either 2018 or 2021 law.

Having very different credit amounts for families with children and those without children also complicates predictions. If a child benefit could be split across multiple households or workers without children could qualify for EITCs similar to workers with children, a child moving out of a household would have less impact on reducing a family’s credit.

Even absent legislation to change the design of the EITC or CTC, it might still be possible to deliver advanced credits without creating undue risk for low-income families who may be unable to pay back errantly advanced credits. We propose collecting data from the first quarter of the year when families file their tax returns. Those data would be used to predict a family’s EITC and CTC, and for families that did not opt out, payments would begin in July. In some cases, families would know about upcoming changes in the number of children they have (e.g., birth of a child) and could report that information to improve credit predictions.

Working toward an advanced payment would better match credit timing to household needs, particularly for low-income families. It could also reduce income volatility for these same families that exacts a negative toll on families and children.