TY2021 and TY2022 Education Tax Credit Underclaims for Filers and

Agency decision

Ask Donna

What actually matters in this document.

Text

Nonﬁlers

Lori Stuntz

Internal Revenue Service

Mark Payne

Internal Revenue Service *

The views expressed in this paper are those of the authors and do not necessarily represent the views of

the IRS.

We are grateful to Deana Ackerman, Kim Castelin, Adam Cole, Julie Anne Cronin, Joseph Gray-Hancuch, Alicia

Miller, Mark Robinson, and Mike Udell for their helpful comments on this working paper. We also thank Corryn

Stephens Manning for assistance working with the data. We thank Brenda Schafer and the Taxpayer Behavior Lab

for creating state level geographic heatmaps.

1. Introduction

Over the course of the last several decades, Congress has given the Internal Revenue Service (IRS)

responsibility for administering a range of credits and deductions for individuals and businesses. The IRS

conducts numerous education and outreach programs as well as releases other resources designed to

help taxpayers understand and claim these tax benefits. However, many individuals and families remain

unaware of particular credits and deductions or face barriers that deter them from claiming the full

amount for which they are eligible. As part of its strategic operating plan, the IRS has pledged to help

taxpayers understand and claim appropriate credits and deductions. 1

Credits and deductions are provisions that reduce the amount of taxes a taxpayer needs to pay or

otherwise increase the amount that can be refunded based on certain eligibility criteria. Taxpayers can

receive three general types of credits and deductions:

•

A tax deduction reduces the amount of income subject to tax.

A non-refundable tax credit is a credit that reduces the amount of tax liability, but the amount

of the credit may not exceed the taxpayer’s tax liability. (For example, if a taxpayer is eligible for

a non-refundable tax credit of $1,500 and has $1,000 in tax liability, the credit reduces the

taxpayer’s liability by only $1,000.)

A refundable tax credit is a credit that reduces the amount of tax liability and, if the amount of

the credit exceeds the taxpayer’s tax liability, the amount of the credit that exceeds the

taxpayer’s tax liability can be given as a refund to the taxpayer, even in cases where the taxpayer

does not have any tax liability. (For example, if a taxpayer is eligible for a refundable tax credit of

$1,500 and has $1,000 in tax liability, the credit reduces the taxpayer’s liability by $1,000, and

the taxpayer will have an additional $500 refunded.)

To design better strategies to support eligible taxpayers in claiming credits and deductions, the IRS and

decision makers ﬁrst need to understand where taxpayers are not fully claiming credits and deductions

and the barriers that may impact claiming behavior. To ground these discussions, the IRS has begun

developing approaches to measure uptake by eligible taxpayers and to estimate the share of eligible

taxpayers who do not claim credits and deductions for which they are eligible. In the inaugural year of

releasing aggregate estimates of the tax credits and deductions gap— the amount of credits and

deductions for which taxpayers are eligible but did not claim, irrespective of whether those taxpayers

ﬁled tax returns— the IRS focused on a subset of fully or partially refundable credits for individuals. This

working paper is part of that series of working papers and focuses on education credits, the American

Opportunity Tax Credit (AOTC) and the Lifetime Learning Credit (LLC). The IRS invites comments and

recommendations on ways to improve the methodology used to measure the credits and deductions gap

for these credits.

As this work continues in the coming years, the IRS plans to expand the set of credits and deductions

examined and continue to reﬁne the methodologies used to estimate the credits and deductions gap.

This work will feed into other eﬀorts by the IRS to increase awareness of credits and deductions and

remove barriers to claiming. Overall, this will inform and complement other IRS’s eﬀorts to prevent

inadvertent errors, fraud, and abuse.

Internal Revenue Service. “Internal Revenue Service Inflation Reduction Act Strategic Operating Plan, FY2023 –

2031.” Page 36 https://www.irs.gov/pub/irs-pdf/p3744.pdf

We estimate the tax beneﬁt gap for the education tax credits in terms of the number of students for

whom there is an education credit underclaim, number of tax returns for which there is an education

credit underclaim, and the amount of the underclaim. Education credit underclaims may arise from three

broad groups of students: 1) eligible students that appear on a return with an education credit claim for

another student but there is no credit claimed for them (“partial claimants”); 2) eligible students that

appear on return without an education credit (“non-claimants”); and 3) eligible students that do not

appear on a return (“non-ﬁlers”). We do not estimate underclaims for students who have credit claims,

rather we assume that existing education credit claims are for the correct education credit and for the

correct amount. Because tax return data does not provide complete information to identify AOTC

eligibility, we consider a range of underclaim estimates with diﬀering AOTC eligibility assumptions. As a

preview of our results, we ﬁnd that 47% of students claim any education credit and when considering

only students who are eligible for an education tax credit the claim rate increases to a range of 62% to

67%. This is in line with Herlache et al. (2021) who estimate an AOTC take-up rate of around 47% for all

students and Cronin & Gray-Hancuch (2024) who ﬁnd 45% of undergraduate students have any

education credit. Several papers estimate education-credit take up rates on smaller subsets of taxpayers.

Guyton et al. (2017) ﬁnd that take-up rates range from 65% for independent students to 77% for highschool seniors and take-up rates increase with income. Cronin & Gray-Hancuch estimate a 60% AOTC

take-up rate for students that look AOTC eligible, while Bulman & Hoxby (2015) consider a range of

eligibility assumptions and note that take-up rates increase with income.

This paper adds to the existing literature by estimating claim participation rates and credit underclaims

for Tax Years 2021 and 2022 for both AOTC and LLC using administrative data. Our analysis is based upon

the population of students and the tax returns on which they appear rather than a sample. We also

proﬁle non-claimants and partial claimants along dimensions including geography, preparation method,

ﬁling status, student age, income, and past credit claims, and look at reasons why a tax return without

any claim or only a partial claim is not eligible for an education tax credit. In line with Cronin & GrayHancuch (2024), we further demonstrate that as many as one-third of non-claimants may not claim a

credit due to a lack of understanding of how to treat scholarships as income in order to maximize the

amount of education tax credits the tax unit is eligible to claim.

For TY2022, in dollar terms we ﬁnd $12.7 billion in total education credit claims and total education

credit underclaims that range from $3.2 billion to $8.9 billion, with a middle of the road underclaims

estimate of $6.3 billion. 2 The estimated credit participation rate among tax credit eligible tax return ﬁlers

ranges from 62% to 67% depending upon AOTC eligibility assumptions. Likewise, for TY2021 we ﬁnd

$13.5 billion in total education credit claims and a middle underclaims estimate for ﬁlers of $6.6 billion,

and an estimated participation rate for eligible ﬁlers that ranges from 62% to 66%. Nonﬁlers represent

15.5% of students with any estimated underclaim in TY2021 and 20.1% in TY2022. 3

This paper proceeds as follows. Section 2 discusses credit eligibility rules for both the AOTC and LLC and

presents historical data on credit claims. We discuss how we identify and construct a dataset of students

from administrative tax return data and present data on existing education credit claims in Section 3. We

identify potential apparently-eligible partial claimants and nonclaimants and estimate dollars of credit

underclaims in Section 4. In Section 5, we combine estimates on claims and underclaims to generate

Estimated education credit claims are “as reported” on tax returns, pre-audit, and do not include any adjustments

for incorrect claims.

Some of this diﬀerence across tax years could be due to timing. At the time of analysis, less time had passed for

TY2022 ﬁling and some of the F1098-T students who do not match to a tax return may be eventual late ﬁlers.

participation rates for students that appear on ﬁled tax returns. Section 6 examines potential barriers to

claiming an education credit looking at tax return preparation methods as well as reasons why a

nonclaimant may not be eligible for any education credit. Non-ﬁler underclaim methodology estimates

are presented in Section 7. Finally, Section 8 combines the ﬁler and non-ﬁler estimates and concludes.

2. Education credit eligibility rules

The Internal Revenue Code contains two tax credits for higher education tuition expenses, the American

Opportunity Tax Credit (“AOTC”) and the Lifetime Learning Credit (“LLC”). Table 1 compares key

diﬀerences for the two credits in credit amounts, qualiﬁed expenses, and eligibility requirements. The

AOTC is available for 100% of the ﬁrst $2,000 of qualiﬁed expenses and 25% of the next $2,000 of

qualiﬁed expenses, for a maximum of $2,500. Forty percent of the otherwise available AOTC is

refundable up to a maximum of $1,000. The LLC is available for 20% of the ﬁrst $10,000 of qualiﬁed

expenses and no portion of this credit is refundable. As of tax year 2021, both credits have the same

income limitations. Taxpayers who are married ﬁling joint returns may claim the full credit with modiﬁed

adjusted gross income (“AGI”) up to $160,000, then the credit begins to phase-out and reaches $0 for

taxpayers with $180,000 modiﬁed AGI 4. These amounts are cut in half to $80,000 and $90,000 for

taxpayers who do not ﬁle Married Filing Jointly.

Table 1. Tax Credits for Higher Education

American Opportunity Tax

Credit (AOTC)

Credit Amount

100% of ﬁrst $2,000; 25% of

next $2,000

Refundability

40% refundable

Income limitation

$160k - $180k for joint ﬁlers;

$80k - $90k else

Qualiﬁed expenses

Tuition, fees, and course

materials

Calculation

Applied on a per student basis

with a maximum of $2,500 for

any eligible student

Years Eligible

First 4 years of postsecondary

education

Enrollment Status

At least half-time; degree

seeking

Felony for controlled substances Not eligible

Lifetime Learning Credit (LLC)

20% of $10,000

Not refundable

$160k - $180k for joint ﬁlers;

$80k - $90k else

Tuition and fees

Applied on a per return basis

with a maximum of $2,000 per

tax return

No limit on number of years

No minimum enrollment or

degree seeking requirement

Eligible

Generally, qualiﬁed expenses for the AOTC include tuition, fees, and course materials while LLC qualiﬁed

expenses only include tuition and fees. The AOTC is calculated on a per student basis with a maximum of

$2,500 for any eligible student on a return. The LLC is applied on a per return basis with a maximum of

$2,000 per tax return. AOTC eligible students are in their ﬁrst 4 years of postsecondary education,

enrolled at least half-time, in a degree seeking program, and have no felony conviction for controlled

substances. The LLC has no such limitations on student eligibility. All AOTC eligible students are also LLC

Modiﬁed AGI for education credits is AGI plus any foreign earned income exclusion, foreign housing exclusion,

foreign housing deduction, and any exclusion of income by bona ﬁde residents of American Samoa or Puerto Rico.

eligible. An individual student may only claim one tax credit for higher education in a tax year. A tax

return may claim both education tax credits if they are claimed for diﬀerent students on the return.

Both education tax credits are calculated on Form 8863, where the AOTC is separated into a refundable

and nonrefundable portion. Figure 1 depicts estimated dollar amounts for the refundable and

nonrefundable portions of the AOTC as well as the LLC over time from the IRS Statistics of Income (SOI)

as well as student enrollment. 5 SOI line count estimates are based upon a sample of tax returns and are

released annually in Publication 4801. 6 Education credit claims have steadily declined since tax year

2013. Student enrollment has also declined over this time period from 28.2 million students in academic

year 2011-2012 to 25.5 million students in academic year 2019-2020.

Figure 1. Tentative Education Tax Credit Amounts and Annual Student Enrollment, Tax Years 2012-2021,

(Billions of Dollars Claimed and Enrollment in Millions of Students)

28.2

27.1

26.7

$ Billions

26.3

26.1

25.8

25.7

25.5

24.7

Millions of Students

27.5

2012

2013

2014

2015

Refundable AOTC

2016

2017

2018

2019

Nonrefundable AOTC

LLC

2020

2021

Enrollment

Source: SOI Individual Line Count Estimates and Department of Education, NCES 12-month academic year

enrollment at degree granting institutions.

3. Identifying Students in Administrative Tax Data

While education credits are claimed on a tax return, eligibility is dependent on each student’s

circumstances. We use administrative tax data to identify students and calculate education credit claims

(U.S. Department of Education, National Center for Education Statistics)

(Statistics of Income, Internal Revenue Service, Department of Treasury, 2012 - 2021) available at

https://www.irs.gov/statistics/soi-tax-stats-individual-income-tax-returns-line-item-estimates-publications-4801and-5385 Each education credit component represents a speciﬁc line on F8863: Refundable AOTC is Line 8,

Nonrefundable AOTC is Line 9, and LLC is Line 18.

and underclaims by joining data from three separate tax forms: F1040, F8863, and F1098-T. We calculate

education credit claims and underclaims at the student level, as the taxpayer is instructed to do on

F8863, and aggregate the student level claims and underclaims back to the tax return. We can identify

students in tax data in two ways: 1) students who receive a F1098-T tuition statement from an education

institution; and 2) a taxpayer reported student listed on the F8863 with an education credit claim. Note

that students with an education credit claim on F8863 may not always have a F1098-T (more on this

later) and students with a F1098-T may not have any education credit claim. We construct a data set of

students in a series of steps outlined in Table 2 and discussed in turn.

Table 2. Identifying Students on Individual Tax Returns (Counts in Thousands)

TY2021

TY2022

F1098-T Tuition Statements

23,390

Unique Students

21,571

Match to a person on a F1040

20,043

Step 2: Tax Returns with F8863 Education Credits

F8863 Filings

9,104

And has at least one student with a F1098-T

8,247

And has no students with a F1098-T

857

Step 3: Identify Persons on Individual Returns and link to F1098-Ts and F8863s

F1040 Filings

160,059

Number of Persons on those F1040 Returns

301,761

Persons who match to F1098-T

24,009

And the person appears on a return with a F8863

9,080

And the person appears on a return with no F8863

14,929

Step 4: Apportion F8863 return level credits to students on a return

Number of Returns with F8863

9,104

Number of Students on those Returns

10,048

Student has a 1098t

9,080

Student has no 1098t

969

22,944

21,053

19,248

Number of Students with a tentative credit

Student has a 1098t

Student has no 1098t

9,600

8,631

969

9,032

7,956

1,076

Number of Returns with total education credit claims > $0

Number of Students with total education credit claims > $0

Step 5: Students on a Return with no F8863

Number of students

Number of unique students

9,036

9,545

8,462

8,946

14,929

12,737

14,776

12,482

Step 1: Students with F1098-T Tuition Statements

8,609

7,647

962

156,104

295,520

23,214

8,439

14,776

8,609

9,514

8,439

1,076

Step 1: Students with F1098-T Tuition Statements

Education institutions provide a tuition statement, Form 1098-T (Figure 2), to students and to the IRS.

Tuition statements provide some of the necessary information for calculating education tax credits. Each

F1098-T provides data on the amount of tuition paid and scholarships or grants administered through

the school as well as indicators for if the student was enrolled at least half time and for if the student

was a graduate student. After removing duplicate ﬁlings and ﬁlings with an invalid taxpayer identiﬁcation

number (TIN), we are left with 22.944 million F1098-T ﬁlings for TY2022 representing 21.053 unique

students. 8.2% of these unique students (1,717 million) received a F1098-T from more than one

institution. Following the same procedure as Cronin & Gray-Hancuch (2024), we reconcile this F1098-T

data with enrollment data from the Department of Education during academic year 2020-2021 (the most

recent year available) and ﬁnd that 14.2% of students did not receive a F1098-T. 7

Figure 2. Form 1098-T Tuition Statement

As noted by Ackerman et al. (2014), Bulman & Hoxby (2025), and Cronin & Gray-Hancuch (2024) among

others, educational institutions may not issue a F1098-T for valid reasons and the student may still be

eligible for an education tax credit. A school may choose not to submit a F1098-T if the tuition paid is $0.

When the tuition paid is $0 because the student received scholarships or grants to oﬀset the tuition, the

school may choose not to submit a F1098-T, and the student may still be credit eligible because the

student may choose to count the scholarship or grant against other components of the cost of

attendance and claim a credit on the remaining tuition expenses. Cost of attendance for student ﬁnancial

aid purposes includes all costs to attend school. In addition to tuition and fees, cost of attendance

includes: books, course materials, supplies, and equipment; cost of housing and food (or living

expenses); transportation expenses; loan fees (excluding any loan fees for non-federal student loans);

miscellaneous expenses (including a reasonable amount for the documented cost of a personal

computer); allowance for childcare or other dependent care; costs related to a disability; and reasonable

As done by Cronin & Gray-Hancuch (2024) for TY2017, we start with Department of Education academic

enrollment for degree seeking students during academic year 2020-2021. We add a Department of Education

estimate of nondegree seeking students and subtract an estimate of non-resident aliens. This gives an estimate of

U.S. citizens and resident aliens at Title IV institutions equal to 24.2 million students. To this estimate we use the

authors adjustment of 11% to increase the academic year estimates to represent a calendar year. Finally, we

subtract our count of F1098-T students that are enrolled in multiple institutions to arrive at a calendar year 2021

enrollment of 25.156 million students. This enrollment estimate includes 3.6 million more students than appear in

F1098-T data for tax year 2021.

costs for eligible study abroad programs. 8 IRS Publication 970 includes several examples of how a

taxpayer may choose to count scholarships against the cost of attendance ﬁrst and receive a tax credit on

the remaining tuition dollars. 9 Cronin & Gray-Hancuch (2024) describe this complexity around calculating

education credits when the student receives scholarships as a reason why some students do not claim an

education credit.

Table 3 contains counts and amounts for F1098-T line items for unique students in tax years 2021 and

2022. We use the half-time enrollment and graduate student indicators to classify each F1098-T as either

likely AOTC (enrolled at least half-time and not a graduate student) or likely LLC (enrolled less than halftime or a graduate student). These classiﬁcations are not perfectly aligned with AOTC and LLC eligibility

rules and we will go into detail in section 4.2 about how well these rules line up with education credit

claims. For TY2022, we classify 69.5% of the 21 million unique students as having a likely AOTC F1098-T

and 34.5% as having a likely LLC F1098-T. Note that students with multiple F1098-Ts may have both a

likely AOTC F1098-T and a likely LLC F1098-T. For example, a student who completes their undergraduate

studies in May of one year and begins graduate school in August of the same year could have one likely

AOTC F1098-T from the undergraduate institution and a likely LLC F1098-T from the graduate school.

Tuition Statements do not include textbook expenses, drug felony information, year of postsecondary

education, or degree seeking status.

Step 2: Tax Returns with F8863 Education Credits

Taxpayers claim education tax credits using F8863 “Education Credits” (included in the appendix). The

taxpayer begins on page 2 of the form where they are asked to provide the SSN for each student and

answer a series of questions to determine AOTC and LLC eligibility. Based upon the answers to these

eligibility questions, the taxpayer then calculates either a tentative AOTC for each individual student or a

combined amount of eligible LLC tuition across all eligible LLC students on the return. Tentative AOTC

credit amounts are summed up and entered on page 1, where the income limitations are applied, and

the credit is split into a refundable and nonrefundable portion. Eligible LLC tuition amounts are also

entered on page 1, income limitations are applied, and the income limited credit is combined with the

nonrefundable portion of the AOTC to get a combined nonrefundable education credit. The taxpayer is

directed to a credit limit worksheet to calculate the portion of the nonrefundable education credits to

enter on line 19. 10 This nonrefundable education credit after the credit limit worksheet is entered on

Schedule 3, line 3. Finally, Schedule 3 nonrefundable credits are entered on the Form 1040 line 20,

combined with the nonrefundable portion of the child tax credit (line 19), and subtracted from tax

liability.

We create a calculator to compute each line of the F8863 as well as the ﬁnal portion of nonrefundable

education credits able to be claimed against tax liability for each tax return. Table 4 contains some of

these key lines for tax years 2021 and 2022. 11 In TY2022, 8.4 million tax returns claimed $12.7 billion in

(US Department of Education, 2024)

(Internal Revenue Service, Department of the Treasury, 2022)

The credit limit worksheet determines the amount of nonrefundable credit the taxpayer can claim against tax

liability. Allowable nonrefundable education credits are limited to tax liability before credits (F1040, line 18) less the

following tax credits: foreign tax credit, credit for child and dependent care expenses, retirement savings

contribution credit, and partner’s additional reporting year tax from Form 8978.

We limit our sample to taxpayers with valid taxpayer identiﬁcation numbers (“TINs”) who had a return posted by

12/31/2023 for TY2021 and by 04/30/2024 for TY2022.

Table 3. Form 1098-T Tuition Statement Counts and Amounts

TY 2021

Count

Amount

(K)

($M)

21,571

23,390

21,236 227,558

1,059

1,615

12,528 105,337

385

713

TY 2022

Count

Amount

(K)

($M)

21,053

22,944

20,727

228,762

1,023

1,740

12,416

108,133

344

628

3,902

18,362

3,904

3,762

18,020

3,860

Number of students that look AOTC eligible

Number of 1098ts

Line 1 - Qualified tuition

Line 2 - Prior year tuition adjustments

Line 5 - Scholarships or Grants

Line 6 - Prior year scholarship adjustments

Line 7 - Tuition includes amount for academic period

beginning next calendar year

Line 8 - Half time student indicator

Line 9 - Graduate student indicator

Line 10 - Insurance contract reimbursement / refund

14,932

15,719

14,485

699

10,120

247

167,461

1,001

87,821

412

14,634

15,408

14,163

647

10,010

212

168,581

1,008

90,075

362

2,875

14,932

2,799

14,634

Number of students that look LLC eligible

Number of 1098ts

Line 1 - Qualified tuition

Line 2 - Prior year tuition adjustments

Line 5 - Scholarships or Grants

Line 6 - Prior year scholarship adjustments

Line 7 - Tuition includes amount for academic period

beginning next calendar year

Line 8 - Half time student indicator

Line 9 - Graduate student indicator

Line 10 - Insurance contract reimbursement / refund

7,382

7,670

6,751

360

2,408

138

60,097

614

17,516

301

7,262

7,536

6,563

376

2,406

132

60,181

732

18,057

266

1,026

3,430

3,904

963

3,386

3,860

Total number of unique students

Number of 1098ts Received

Line 1 - Qualified tuition

Line 2 - Prior year tuition adjustments

Line 5 - Scholarships or Grants

Line 6 - Prior year scholarship adjustments

Line 7 - Tuition includes amount for academic period

beginning next calendar year

Line 8 - Half time student indicator

Line 9 - Graduate student indicator

Line 10 - Insurance contract reimbursement / refund

education credits -- $5.0 billion in refundable AOTC and $7.6 billion in nonrefundable education

credits. 12. While total education credits have decreased from TY2021 to TY2022, continuing the

downward trend shown in Figure 1, average credit amounts are similar between the two years.

Line 7 contains a checkbox for taxpayers to indicate when they are not eligible for the refundable portion

of the AOTC. Instead, these taxpayers may claim the full amount of any AOTC as a nonrefundable credit.

A taxpayer is ineligible for a refundable AOTC if any they meet all three of the following conditions: 1)

under age 18 at the end of the taxable year; or age 18 at the end of the taxable year and earned income

is less than one-half of support; or over age 18 and under age 24 at the end of the taxable year and

earned income is less than one-half of support; 2) at least one of the taxpayer’s parents is alive at the

end of the taxable year; and 3) the taxpayer is not ﬁling a joint return. Around 2% of returns in our data

with a tentative AOTC check this box.

Table 4. F8863 Calculations for selected line items

Line 1 - Tentative AOTC

Line 7 - Tentative AOTC after Income

Limit

Line 7 - Checkbox

Line 8 - Refundable AOTC

Line 9 - Nonrefundable AOTC

Line 12 - Tentative LLC Amount

Line 18 - LLC after Income Phaseout

Line 19 - Nonrefundable Education

credits after credit limit worksheet

Nonrefundable Education Credits

Claimed on F1040

Refundable AOTC + Nonrefundable

Education Credits Claimed on Return

Count

(K)

6,164

TY 2021

Dollars

($M)

13,630

Avg

($)

2,211

Count

(K)

5,964

TY 2022

Dollars

($M)

13,313

Avg ($)

2,232

6,163

138

6,039

6,163

3,086

3,085

13,285

5,216

8,069

3,170

3,081

2,156

864

1,309

1,027

999

5,963

121

5,801

5,910

2,782

2,780

12,944

5,045

7,782

2,887

2,792

2,171

870

1,317

1,038

1,004

7,946

8,245

1,038

7,345

7,691

1,047

7,943

8,239

1,037

7,287

7,628

1,047

9,036

13,455

1,489

8,462

12,674

1,498

Table 5 distributes the total education credit claims across AGI for both tax years. Average credit claims

increase with income, reﬂecting the fact that higher income taxpayers are more likely to have enough tax

liability to claim the full size of any nonrefundable credit. Appendix 3 contains a distribution for the

education credit components: refundable AOTC, nonrefundable AOTC, and LLC.

Step 3: Identify Persons on Individual Returns and link to F1098-Ts and F8863s

We join the F1098-Ts to persons who appear on a F1040 tax return. For tax year 2021, our data includes

160 million F1040 ﬁlings representing 301.7 million individual persons. We match a F1098-T to 24 million

persons on a return. This number is larger than the number of unique students with a F1098-T (21.6

million) and reﬂects the fact that many students appear on multiple tax returns. A student may appear

on their parents’ tax return as a dependent and on a separate return as a primary taxpayer. A student

See Appendix 2 for a comparison of our estimates to the SOI line count estimates for TY2021 (the latest year

available).

may be erroneously claimed on two or more tax returns as well. We are careful when calculating

underclaims to ensure we only assign an underclaim to an individual student once. We can further

separate persons on a return with a F1098-T into those who appear on a return with a F8863 for

Education Credits (9.1 million) and those who do not (14.9 million).

Table 5. Total Estimated Education Credit Claims Distributed by AGI

Number

Amount

Average

AGI

(k)

Distribution

($m)

Distribution

TY 2021

917

<= 0

0.9%

0.6%

965

1 to 25k

2,503

27.7%

2,417

18.0%

1,569

25k to 50k

2,461

27.2%

3,860

28.7%

1,705

50k to 75k

1,336

14.8%

2,279

16.9%

1,694

75k to 100k

894

9.9%

1,514

11.2%

1,883

100k to 200k

1,758

19.5%

3,309

24.6%

> 200k

0.0%

1,489

Total

9,036

100.0%

13,455

100.0%

<= 0

1 to 25k

25k to 50k

50k to 75k

75k to 100k

100k to 200k

> 200k

Total

2,275

2,279

1,277

828

1,715

8,462

TY 2022

1.1%

26.9%

2,220

26.9%

3,626

15.1%

2,185

9.8%

1,384

20.3%

3,177

0.0%

100.0%

12,674

0.7%

17.5%

28.6%

17.2%

10.9%

25.1%

0.0%

100.0%

932

976

1,591

1,711

1,672

1,853

1,498

Step 4: Apportion F8863 return level credits to students on a return

Education credit claims in the data available for this project are recorded at the return level. But because

credit eligibility depends on characteristics of an individual student, we want to apportion the return

level credits to the individual students on the return. We identify students in two ways: by the presence

of a F1098-T or from being listed as a student on F8863. For tax year 2021, our dataset includes 9.104

million returns with a F8863 for education credits representing 10.0 million students. 9.1 million of these

students have a F1098-T (90.4%) and 969 thousand do not (9.6%).

To apportion the return level credits to students we begin with the series of eligibility questions on page

2 of F8863 in Figure 3. The data used for this project contains answers for up to three students per tax

return. 13 An AOTC eligible student is one with less than 4 prior AOTC claims (Q23 = No), who is enrolled

at least half-time in a degree seeking program (Q24 = Yes), who is in the ﬁrst four years of postsecondary

Forty-four percent of returns in our data have a single student and 79% have 3 or fewer students. For the 21% of

returns with more than 3 students, we only observe the answers to these questions for three students.

education (Q25 = No), and who has no felony convictions for possession or distribution of a controlled

substance (Q26 = No). An LLC eligible student has any other combination of answers to the eligibility

questions.

Figure 3. F8863 Student Education Credit Eligibility Questions for TY2021 and TY2022

To apportion credits, we classify each student as either AOTC or LLC eligible according to the answers to

Q23 – Q26. Then we implement a series of iterative rules. First, if only one student on a return has

indicators for a type of a credit (AOTC or LLC), that student gets the full amount of the credit claimed on

the return up to the maximum allowable credit per student for AOTC. If the student checkboxes indicate

eligibility for one type of credit but the only credits claimed on the return are for the other education tax

credit and the return has no other students, we give the student the other type of credit in amounts

equal to what was claimed by the return. This rule apportions 89.4% of AOTC credit dollars and 96.6% of

LLC dollars in TY2022 (and 89.6% of AOTC dollars and 96.5% of LLC dollars in TY2021).

Second, we divide any remaining unmatched credit amounts claimed on a return across students with a

particular type of credit proportionally according to their share of overall tuition on F1098-Ts. For

example, consider a return with two AOTC eligible students where F1098-T tuition amounts are $4,000

for student 1 and $2,000 for student 2. We assign two-thirds ($4,000 / $6,000) of the return level AOTC

credits to student 1 and one-third to student 2. We only consider a maximum of $4,000 tuition for AOTC

eligible students as that is the maximum amount of tuition that can be used for the credit. At this point,

we also apportion any remaining unmatched AOTC credits to student with matched credits that are less

than the per student maximum AOTC amount. The idea for this rule is that these amounts should reﬂect

unobserved qualiﬁed textbook expenses for students where qualiﬁed tuition is smaller than the amount

used to calculate the AOTC credit on the return. This rule apportions an additional 8.6% of AOTC credits

and 2.2% of LLC credit dollars.

Finally, we divide remaining unmatched credit amounts evenly over students listed on a F8863 who do

not yet have any credits assigned after the prior steps. This rule apportions an additional 2.0% for AOTC

and 0.5% for LLC credit dollars in TY2022. We are left with approximately 17 thousand students on

returns that have education credits who have no education credit assigned. These are students identiﬁed

by a F1098-T who we do not see on an F8863 in our data. 14 We assume the remaining unmatched return

credit dollars go with these students. Table 6 shows the result of this apportioning for total education

credit claims and for the individual components for TY2021 and TY2022. For TY2022, 8.5 million returns

Recall that the data available for this project includes F8863 information for up to three students per tax return.

have positive education credit claims covering 8.9 million individual students and for TY2021 9.036

million returns have education credit claims covering 9.5 million individual students.

Table 6. Return Level Education Tax Credits Apportioned to Students on the Return

Return Level Claims

Apportioned To Students

Returns

Amounts

Returns

Amounts

(thousands)

($ millions)

(thousands)

($ millions)

TY2021

Total Education Credits

9,036

13,455

9,545

13,436

Refundable AOTC, Line 8 F8863

6,039

5,216

6,371

5,215

Nonrefundable AOTC, Line 9 F8863

6,163

8,069

6,496

8,068

Nonrefundable LLC, Line 18 F8863

3,085

3,081

3,104

3,061

Nonrefundable Education Credits

Claimed

7,943

8,239

8,423

8,221

TY2022

Total Education Credits

8,462

12,674

8,946

12,656

Refundable AOTC, Line 8 F8863

5,801

5,045

6,127

5,044

Nonrefundable AOTC, Line 9 F8863

5,910

7,782

6,236

7,780

Nonrefundable LLC, Line 18 F8863

2,780

2,792

2,796

2,773

Nonrefundable Education Credits

Claimed

7,287

7,628

7,733

7,612

Step 5. Students on a Return with no F8863

The ﬁfth and ﬁnal step considers F1098-T students who appear on tax returns with no education credit

claim (14.8 million students in TY2022). For students who appear on multiple returns, we remove

instances where the student is a dependent ﬁler. We are left with 12.5 million students identiﬁed by a

F1098-T who appear on a F1040 with no education credit claims for TY2022 (12.7 million students for

TY2021).

The TY2022 total population of students on tax returns is 21.99 million (22.8 million in TY2021). This is

the total number of Step 4 students on F8863 returns (9.5 million) plus the Step 5 total number of

students on returns with no F8863 (12.5 million). This population is not a distinct set of students as some

students do appear on multiple returns with F8863 ﬁlings. We are careful not to assign a credit claim or a

credit underclaim to any student more than once even when they appear on multiple tax returns.

4. Identifying apparently-eligible underclaimants

4.1 Classifying Students

We want to identify students in our data who might have a credit underclaim. We do not estimate

underclaims for students who have credit claims, rather we assume that existing education credit claims

are for the correct education credit and for the correct amount. 15 We consider for potential underclaims

any student on a return who does not have an education credit and who does not have any education

credit claim on any other tax return. To do this, we separate student ﬁlers into three broad buckets.

Group 1 contains students who have an identiﬁed education credit claim (42%). Group 2 contains

students identiﬁed by a F1098-T who do not have an education credit but who appear on a return with

an education credit claim for another student (2%). Group 3 contains students identiﬁed by a F1098-T

with no credit claims who appear on a tax return that also has no other education credit claims (56%).

Potentially eligible underclaimant students are in Groups 2 and 3.

Group 1 represents claimant students (9.6 million in TY2021 and 9.0 million in TY2022). To get a set of

nonclaimant students we further subdivide Groups 2 and 3. From Group 2, we remove 17 thousand

students on returns with unmatched credits under the assumption that those return level unmatched

credits should go with these students. We remove an additional four thousand students in TY2021 (three

thousand in TY2022) who appear on a diﬀerent tax return with an education credit claim. This leaves 428

thousand students in TY2021 (463 thousand in TY2022) with no education credit claim who appear on a

tax return with education credit claims for at least one other student. These are the Group 2 students

who might have an underclaim. Likewise, from Group 3 we remove 1.8 million students in TY2021 (1.7

million in TY2022) who appear on a diﬀerent return with an education credit claim leaving 10.97 million

potential nonclaimants in TY2021 (10.8 million in TY2022). Underclaims for students in Group 2, who

appear on a return with an education credit claim for at least one other student, may occur for diﬀerent

reasons than for students in Group 3, who appear on a return with no education credit claims.

We look at tax return observable characteristics of student claimants and nonclaimants in Table 7a and

7b. Each tax year is presented as a separate panel. We separate student claimants by F1098-T presence,

and we also separate student nonclaimants by whether or not they appear on a tax return with any

education credit claims for other students. In TY2021, 3.8% of student nonclaimants appear on a return

with an education credit claim for at least one other student. Students without any F1098-T are older

(average age 31) than students with a F1098-T (average age 26). Student nonclaimants on returns with

another education credit are younger and more likely to be a dependent than both student claimants

and student nonclaimants on a return with no education credit claims.

Tables 7a and 7b also contains statistics on prior education credit claims. We look at F8863 ﬁlings for tax

years 2012-2022 and use the credit eligibility questions to identify if a student has an AOTC or LLC claim

for each tax year. Considering all students in our data for TY2021, 49% had a prior AOTC claim and 16%

had a prior LLC claim while 54% had any prior education credit claim. Student claimants are more likely

to have a prior education credit claim (72.7% for students with a F1098-T and 64.7% for students with no

F1098-T) than student nonclaimants (47.1% for students on returns with any education credit and 38.9%

for students on returns with no education credit claims). Students with a prior claim are more likely to

have had a prior AOTC claim than a prior LLC claim. We also look at the proportion of students with four

or more prior AOTC claims (11%). These students are ineligible for a current year AOTC but could still be

eligible to claim an LLC.

Prior work (IRS Research, Applied Analytics & Statistics, 2018) estimated $0.9 billion in AOTC underclaims for

TY2012. LLC underclaims totaled $0.5 billion and often arose from taxpayers who incorrectly claimed the AOTC

when they were instead eligible for the LLC.

Table 7a. Classification and Characteristics of Students on Tax Returns, Tax Year 2021

Student

Nonclaimants

Student Claimants

on Returns

with Claims

with No claim

Return Has Credit

Student Has F1098-T

Student Has Credit

Potential Underclaims

Number of Students

(thousands)

Total

8,631

969

428

10,972

20,999

Single

Head of Household

Married, filing joint

Married, filing separate

with spouse exemption

Widower

43.5%

14.4%

42.0%

0.0%

Filing Status

45.9%

2.0%

27.7%

18.6%

26.3%

79.2%

0.0%

37.3%

17.4%

42.5%

2.6%

39.5%

16.7%

42.3%

1.4%

0.0%

0.1%

Primary

Spouse

Dependent 1 - 4

58.4%

10.4%

31.2%

52.1%

9.5%

38.4%

55.0%

9.9%

35.1%

<18

>18 - <= 23

>= 24

1.1%

5.8%

44.5%

48.5%

4.4%

8.4%

43.5%

43.8%

3.2%

7.2%

43.0%

46.6%

Average Age

Prior Education Credit Claims, 2012 - 2021

66.1%

59.8%

41.4%

34.6%

22.2%

14.5%

14.3%

11.1%

48.8%

15.9%

72.7%

64.7%

47.1%

38.9%

54.1%

16.0%

17.3%

8.9%

6.9%

11.1%

Any Prior AOTC Claim

Any Prior LLC Claim

Any Prior Education

Credit Claim

Four or more Prior AOTC

Claims

0.1%

0.0%

0.1%

0.2%

Type of Person on Return

71.8%

21.9%

8.1%

17.5%

20.0%

60.6%

Student Age

3.4%

11.8%

3.8%

11.4%

26.6%

38.4%

66.2%

38.4%

Table 7b. Classification and Characteristics of Students on Tax Returns, Tax Year 2022

Student

Nonclaimants

Student Claimants

on Returns

with Claims

with No claim

Return Has Credit

Student Has F1098-T

Student Has Credit

Potential Underclaims

Number of Students

(thousands)

Total

7,956

1,076

463

10,817

20,312

Single

Head of Household

Married, filing joint

Married, filing separate

with spouse exemption

Widower

42.5%

15.1%

42.3%

0.0%

Filing Status

44.9%

5.5%

31.9%

19.5%

23.2%

74.7%

0.0%

34.0%

18.0%

45.3%

2.5%

37.3%

17.7%

43.6%

1.3%

0.0%

0.2%

0.1%

Primary

Spouse

Dependent 1 - 4

57.1%

10.0%

32.9%

48.5%

9.3%

42.3%

52.5%

9.6%

37.9%

<18

>18 - <= 23

>= 24

1.2%

6.0%

45.2%

47.7%

0.1%

0.2%

Type of Person on Return

72.5%

23.2%

7.3%

15.0%

20.2%

61.8%

Student Age

3.6%

11.3%

3.9%

11.5%

26.7%

41.6%

65.7%

35.6%

4.6%

8.7%

43.8%

42.8%

3.4%

7.5%

43.4%

45.8%

Average Age

Prior Education Credit Claims, 2012 - 2021

65.9%

56.7%

42.0%

24.7%

13.5%

14.8%

33.7%

12.6%

47.7%

17.4%

73.4%

61.4%

47.8%

38.5%

53.6%

16.2%

17.0%

8.7%

6.9%

11.1%

Any Prior AOTC Claim

Any Prior LLC Claim

Any Prior Education

Credit Claim

Four or more Prior AOTC

Claims

0.1%

4.2 Data Limitations and Basic Approach to Estimate Underclaims

Tax data does not provide complete information to determine AOTC eligibility for nonclaimants. Without

the F8863 eligibility questions, we do not know if the student is degree seeking, if the student is in the

ﬁrst four years of school, or if the student has a felony conviction for a controlled substance. There is no

information reporting on textbook expenditures.

Another important limitation of this analysis is that we do not observe students without a credit claim

who have no F1098-T. As previously discussed, institutions may not issue a F1098-T when the student

pays no tuition because the amount is fully covered by scholarships or grants. The student may still be

eligible for a credit if they choose to count the scholarship or grant funds against room and board

expenses ﬁrst. As shown on Table 7, 10% of TY2021 students with an education credit claim in our data

do not have a F1098-T (12% in TY2022). Cronin & Gray-Hancuch (2024) estimate that 14% of students did

not receive a F1098-T in TY2017.

We assign a type (AOTC or LLC) to each F1098-T using the graduate student and half-time enrollment

indicators. Students not enrolled in graduate school and enrolled at least half-time are assigned likely

AOTC while students who are enrolled in graduate school or enrolled less than half-time are assigned

likely LLC. This assignment is not perfect and likely overstates the number of students with AOTC tuition.

Table 8 shows how these assignments align with education credit claims for student claimants with an

available F1098-T. For students with only AOTC classiﬁed tuition, 85% have an AOTC claim. For students

with an AOTC claim, 92% have at least one F1098-T classiﬁed as AOTC. Likewise, for students with only

LLC classiﬁed tuition, 81% have an LLC claim and 68% of LLC claimants have at least one F1098-T

classiﬁed as LLC.

Table 8. F1098-T Tuition Type versus Education Credit Claims, TY2021 and TY2022

Students with Education Credit Claim and With F1098-T Tuition > $0, Counts in thousands

TY2021

TY 2022

Has AOTC

Has LLC

Has AOTC

Has LLC

Claim

Total

Claim

Has AOTC tuition

5,006

890

5,896

4,670

780

Has LLC tuition

430

1,840

2,270

398

1,682

Has Both tuition

267

357

242

Total

5,702

2,821

8,523

5,310

2,536

Total

5,450

2,080

316

7,846

Table 8 suggests that using the graduate student and half-time indicators to predict credit eligibility for

nonclaimants will do fairly well to predict aggregate underclaims, but this method is likely to overstate

AOTC claims and to understate LLC claims. We consider three sets of AOTC eligibility assumptions for

underclaims estimates: an upper-bound estimate using the most permissive AOTC eligibility

assumptions, a lower-bound estimate using the most restrictive AOTC eligibility assumptions, and a more

middle of the road approach. For all three estimates, we follow this basic algorithm: calculate possible

AOTC and LLC credits for each student nonclaimant according to F8863 lines 27-31 making sure that each

student only gets one credit; aggregate estimated student credits up to the tax return and run through a

F8863 calculator with return level income phaseouts and credit limitations; apportion used return level

credits back to each individual student. We remove any students with ﬁling status married ﬁling separate

as these returns are ineligible to claim education credits. We also exclude student nonclaimants who ﬁle

from U.S. territories as bona ﬁde residents of the U.S. territories are generally not eligible to claim an

education credit on U.S. tax return.

We use the tuition amounts from F1098-T to estimate qualiﬁed expenditures. We do not adjust these

amounts by any scholarships that may be reported as taxpayers may choose to include scholarship

income in taxable income and claim a credit for the remaining tuition that was not covered by a

scholarship. 16

4.3 Upper-bound Approach

For the most permissive AOTC eligibility approach, we assume that any student not in grad school who is

enrolled at least half-time claims the AOTC. All other students claim the LLC. We include rules like those

behind the F8863 “line 7 checkbox” to deny AOTC refundability to certain ineligible students. We do not

observe if a student provides 50% of their own support and consequently cannot implement those tests.

We deny refundability to 1) any student who is a primary ﬁler and under 18 years of age and 2) any

student who is a primary ﬁler, with ﬁling status other than married, ﬁling joint who is between the ages

of 18 and 23 (inclusive). The top panel of Table 9 contains upper-bound underclaims estimates for both

TY2021 and TY2022. We estimate total upper-bound underclaims of $6.3 billion in TY2022 for 5.4 million

tax returns ($7.0 billion over 5.4 million returns in TY2021). Approximately forty percent of these total

education credit underclaims arise from the refundable portion of the AOTC ($2.7 billion in both tax

years). This panel also shows that not all tentative nonrefundable education credits are able to be

claimed by the taxpayer. In TY2021, tentative nonrefundable education credits total $8.97 billion ($7.2

billion tentative AOTC + $1.8 billion tentative LLC) but only $4.3 billion of these credits are able to be

claimed against tax liability.

We test this upper-bound approach by estimating claims for students with both a credit claim and a

F1098-T to see how well the predictions align with the actual claims. For TY2022, this procedure

overestimates the number of actual AOTC claims by 8.5% and underestimates the number of actual LLC

claims by 18%.

4.4 Lower-bound Approach

For an extreme lower-bound, we implement the most restrictive set of AOTC eligibility rules – we

assume that all nonclaimants are ineligible for the AOTC and only eligible to claim the LLC. As shown in

Table 1, AOTC eligibility criteria are more restrictive than LLC eligibility criteria. Generally, any student in

our sample with a F1098-T should be LLC eligible. Under this lower-bound approach, we estimate $2.95

billion in underclaims across 4.1 million returns for TY2022 ($3.5 billion in underclaims across 5.0 million

returns in TY2021). As was also the case in the upper-bound estimates, taxpayers are not able to use the

full amount of tentative nonrefundable credits, $6.4 billion in TY2022.

See (Internal Revenue Service, Department of the Treasury, 2022) for detailed examples of how a taxpayer may

elect to do this.

Table 9. Underclaims Estimates for Filers TY 2021 and TY 2022

TY 2021

Students Returns Amount

Ret

Students

(K)

($M)

Avg

(K)

Upper-bound

Ref AOTC

3,442

3,281

2,727

831

3,364

Nonref AOTC

4,916

4,754

7,194

1,513

4,600

Nonref LLC

2,364

2,334

1,774

760

2,215

Nonref Credits

5,255

5,045

4,313

855

4,320

Education Credits

6,053

5,806

7,040

1,213

5,602

Ref AOTC

Nonref AOTC

Nonref LLC

Nonref Credits

Education Credits

Ref AOTC

Nonref AOTC

Nonref LLC

Nonref Credits

Education Credits

1,975

3,397

3,873

5,244

5,670

7,258

5,225

1,864

3,285

3,804

5,035

5,435

6,993

5,017

TY 2022

Returns Amount

(K)

($M)

Reg

Avg

3,195

4,429

2,186

4,143

5,362

2,664

6,602

1,703

3,633

6,297

834

1,490

779

877

1,174

1,594

5,384

3,101

3,994

5,589

Middle Approach

855

2,027

1,639

3,224

815

3,581

793

4,321

1,028

5,007

1,907

3,103

3,516

4,145

4,790

1,631

4,969

2,903

3,419

5,049

805

1,601

826

825

1,054

6,773

3,493

Lower-bound

969

5,794

696

4,312

696

4,312

6,520

4,137

6,375

2,951

978

713

4.5 Middle Approach

Finally, we consider a middle-of-the-road approach to AOTC eligibility. We start with the upper-bound

F1098-T credit assignments according to the graduate student and half-time enrollment indicators. We

include the same line 7 checkbox rules and we add two additional rules: we deny AOTC to any student

with at least four prior observed AOTC claims and we deny AOTC to any student older than age 25. The

AOTC is only available for the ﬁrst four years of postsecondary education. While AOTC eligible students

may be older than 25, we observe that 75% of AOTC claimants are age 25 or under.

Under this middle approach, underclaims for ﬁlers total $5.0 billion in TY2022 over 4.8 million returns

($5.6 billion over 5.4 million returns in TY2021). For TY2022 this middle approach estimate is 20%

smaller ($1.3 billion) than the upper-bound estimate. Table 10 shows an AGI distribution for total

education credit underclaims under this approach. Average credit claims increase with AGI. One

explanation for this is that higher income taxpayers are more likely to have enough tax liability to use the

full amount of any tentative nonrefundable education credit.

Table 10. Middle Approach Underclaims Estimates for Filers by AGI

AGI

Number (k) # Distribution Amount $M $ Distribution Average $

TY 2021

<= 0

0.6%

0.5%

944

1 to 25k

1,411

26.0%

848

15.2%

601

25k to 50k

1,776

32.7%

1,952

34.9%

1,099

50k to 75k

860

15.8%

1,102

19.7%

1,282

75k to 100k

522

9.6%

623

11.2%

1,194

100k to 200k

835

15.4%

1,034

18.5%

1,239

> 200k

0.0%

Total

5,435

100.0%

5,589

100.0%

1,028

TY 2022

0.6%

0.5%

963

<= 0

1,131

23.6%

724

14.3%

640

1 to 25k

1,484

31.0%

1,648

32.6%

1,110

25k to 50k

784

16.4%

969

19.2%

1,236

50k to 75k

498

10.4%

593

11.7%

1,192

75k to 100k

866

18.1%

1,090

21.6%

1,258

100k to 200k

0.0%

> 200k

Total

4,790

100.0%

5,049

100.0%

1,054

5. Education Credit Participation Rates

Underclaims arise from two sets of students: students on returns with an existing education credit claim

and students on a return with no education credit claim. Thus, we have returns that are classiﬁed as

both Claimants (because the tax return has an education credit claim for at least one student on the

return) and Nonclaimants (because the tax return has at least one student identiﬁed by a F1098-T with

no education credit claim). Table 11 breaks out total underclaims estimates across these two diﬀerent

sets of tax returns (Claimants with an Underclaim and Nonclaimants with an Underclaim).

We estimate participation rates for ﬁlers both in terms of number of claimants and in amount of credits

claimed as in Table 12. The credit participation rate for ﬁlers is equal to Claims / Total Eligible Population

where the Eligible Population equals Claims + Underclaims. The participation rate for number of

claimants ranges from 62.6% to 65.9% in TY2021, depending on AOTC eligibility assumptions, and the

credit dollar participation rate ranges from 65.7% to 79.5%. This reﬂects the fact that the average

underclaim amount ($1,028) is smaller than the average education credit for claimants ($1,489).

Numbers shown are for TY2021 “middle approach” but the same is true for TY2022 and across all three

sets of AOTC eligibility assumptions.

Table 11. Number and Amount of Education Tax Credit Underclaims for Filers

TY 2021

TY 2022

Underclaim Claimed

Number of

Underclaim Claimed

Number of

Amount

Tax Units

Amount

Tax Units

(millions) (millions)

(thousands)

(millions)

(thousands)

Panel A. Upper-bound

Filers, Total

5,805

7,040

640

5,362

6,297

610

Non-claimants

5,447

6,513

4,982

5,742

Claimants

358

527

640

381

555

610

Filers, Total

Non-claimants

Claimants

5,435

5,091

344

Panel B. Middle

5,589

640

5,155

434

640

Filers, Total

Non-claimants

Claimants

5,017

4,699

318

Panel C. Lower-bound

3,493

640

3,256

236

640

4,790

4,428

362

5,049

4,584

465

610

4,137

3,811

326

2,951

2,707

244

610

Table 12. Education Credit Participation Rates for Filers

Number of Claims (thousands) Amount of Claims ($ billions)

TY2021

TY2022

TY2021

TY2022

9,104

8,609

13.5

12.7

Claims

Underclaims

5,435

5,362

7.1

6.3

Upper-bound

5,017

4,790

5.6

5.0

Middle

4,702

4,137

3.5

3.0

Lower-bound

Total Eligible (Claims + Underclaims)

14,539

13,971

20.6

19.0

Upper-bound

14,121

13,398

19.1

17.7

Middle

13,806

12,746

17.0

15.7

Lower-bound

Participation Rate (Claims / Total Eligible)

62.6%

61.6%

65.7%

66.8%

upper

64.5%

64.3%

70.7%

71.5%

middle

65.9%

67.5%

79.5%

81.1%

lower

Figures 4 and 5 look at education credit underclaims by state for TY2022. Figure 4 shows underclaim

counts and Figure 5 shows underclaim dollars. Under both metrics, the largest concentration of

underclaims are in the most populous states: California, Texas, Florida and New York.

Figure 4. TY2022 Education Credit Underclaimants by State

Figure 5. TY2022 Education Credit Underclaims by State

6. Understanding potential barriers to claiming the credit

Information frictions due to complex rules are often cited as a potential barrier to beneﬁts up-take. 17

Education credit eligibility rules are complex. Tables 13a and 13b present participation rates by four tax

return preparation methods: paid preparer, self-prepared on paper, software, and VITA/TCE Prepared.

Underclaim counts on this table come from the “middle” approach of AOTC eligibility assumptions.

Regarding the tables below, a Claimant is a return with an F8863. A Claimant return may have a potential

underclaim if there is a student on the return, as identiﬁed by an F1098-T, who does not have any

education credit claim. Out of the 405 thousand Claimant returns with a potential underclaim in TY2021,

we estimate that 344 thousand have an education credit underclaim. Non-claimant returns have no

education credit claims. Out of the 9,892 thousand non-claimant returns, we estimate that 5,091

thousand have an education credit underclaim under the middle approach. Altogether in TY2021, we

identify 19.0 million tax returns with students and 14.2 million returns have either a claim or an

underclaim, which leads to the overall participation rate of 64%.

Self-Prepared paper ﬁled returns have the lowest credit-take up rates (43% in TY2021 and 35% in

TY2022), indicating that a lack of information may hinder credit take-up. We do not ﬁnd large diﬀerences

in take-up rates across the three preparation methods that include preparation assistance for either tax

year.

Table 13a. Tax return filers with students by education credit claim status and preparation method, TY

2021 (thousands)

Claimant

No potential underclaim

Potential underclaim

Underclaim > $0

No Underclaim

Non-claimant

Underclaim > $0

No Underclaim

Returns

9,104

8,699

405

344

9,892

5,091

4,801

Paid

Preparer

4,278

4,056

221

189

4,584

2,177

2,406

SelfPrepared

181

103

Software

4,686

4,514

173

145

5,059

2,768

2,291

VITA, TCE

Prepared

Returns with Students

Education credit eligible returns

18,996

14,195

8,861

6,455

258

180

9,745

7,454

132

107

Credit participation rate for eligible

returns

64%

66%

43%

63%

60%

See (Currie, 2006), (Bhargava & Manoli, 2015), and (Cronin & Gray-Hancuch, 2024) among others.

Table 13b. Tax return filers with students by education credit claim status and preparation method, TY

2022 (thousands)

Claimant

No potential underclaim

Potential underclaim

Underclaim > $0

No Underclaim

Non-claimant

Underclaim > $0

No Underclaim

Returns

8,609

8,174

435

362

9,680

4,428

5,253

Paid

Preparer

4,076

3,826

250

211

4,538

1,904

2,634

SelfPrepared

143

Software

4,418

4,250

168

138

4,921

2,396

2,525

VITA, TCE

Prepared

Returns with Students

Education credit eligible returns

18,289

13,036

8,614

5,980

188

127

9,339

6,814

148

115

Credit participation rate for eligible

returns

66%

68%

35%

65%

61%

Tables 13a and 13b show that nearly half of non-claimant returns have no underclaim and 15% of

claimant returns with a potential underclaim have no underclaim. On Tables 14a and 14b, we explore

some reasons that a return with a potential underclaim may not be eligible for an education credit claim.

In TY 2021, we identify 10.3 million returns with a potential education credit claim, and we estimate 4.9

million of those returns have no underclaim under the middle approach for AOTC eligibility.

Table 14a. Individual tax return filers with potential underclaims and possible reasons for no

underclaim, TY2021 (Thousands of Returns)

Claimants w/ potential

Underclaim

And actual underclaim

And no underclaim

Non-claimants

And actual underclaim

And no underclaim

Total Potential Underclaims

Total With No Underclaim

Returns

405

344

9,892

5,091

4,801

10,297

4,862

F1098-T

tuition =

Income

over

phaseout

Not enough

tax liability

Return LLC

already

maxed out

653

2,622

1,559

653

2,622

1,559

656

2,626

1,569

Table 14b. Individual tax return filers with potential underclaims and possible reasons for no

underclaim, TY2022 (Thousands of Returns)

Claimants w/ potential

Underclaim

And actual underclaim

And no underclaim

Non-claimants

And actual underclaim

And no underclaim

Total Potential Underclaims

Total With No Underclaim

Returns

435

362

9,680

4,428

5,253

10,115

5,326

F1098-T

tuition =

Income

over

phaseout

Not enough

tax liability

Return LLC

already

maxed out

680

2,908

1,720

680

2,908

1,720

684

2,912

1,742

Thirteen percent of these returns with no underclaim, or 656 thousand, have reported F1098-T equal to

$0 and that is why they do not qualify for a credit under our methodology. Potential AOTC recipients may

have textbook expenditures that we do not see on tax return data that could make them eligible. Over

half of these credit ineligible non-claimant returns have income amounts that are larger than the

allowable limits. For tax years 2021 and 2022, both the AOTC and LLC begin to phase-out at $180,000 for

married taxpayers ﬁling a joint return and phases out entirely at $200,000. This phase-out range is

reduced to $90,000 - $100,000 for other ﬁling statuses. About a third of ineligible non-claimants do not

have enough tax liability to claim a nonrefundable credit. A taxpayer may be ineligible for the ATOC but

eligible for the LLC but lacks enough tax liability to claim the non-refundable LLC. Finally, 21 thousand of

the claimant returns with a potential underclaim have a student that appears to be eligible for the LLC

but the return has already claimed an LLC for the maximum amount. This represents one-third of

apparently ineligible claimant returns with a potential underclaim.

Lastly, we look at the proportion of underclaimant students where scholarship amounts on the F1098-T

are larger than tuition amounts. These students may not claim a credit because they misunderstand how

scholarship income interacts with education credit calculations. As outlined in Pub 970 and thoroughly

discussed by (Cronin & Gray-Hancuch, 2024), students may choose to count any scholarships received

against the full cost of attendance and claim a credit for the remaining tuition amounts. We ﬁnd that

around a third of underclaimant students have reported scholarship amounts that are greater than

tuition amounts, and this proportion increases to 40-46% for students with an AOTC underclaim and

ranges from 14-32% for students with an LLC underclaim. For F1098-T recipients with reported

scholarships greater than or equal to reported tuition paid, (Cronin & Gray-Hancuch, 2024) estimate an

AOTC take-up rate that is roughly one-half of the overall AOTC take-up rate for apparently eligible

students age 18-21 years old.

7. Education Credit Underclaims for Nonﬁlers

Education credit underclaims can arise from both tax return ﬁlers and non-ﬁlers. Student non-ﬁlers can

be identiﬁed by a F1098-T that does not match to any ﬁled tax return. Additional complexities arise

around determining the appropriate tax unit and ﬁling status for student non-ﬁlers. This section

discusses the methodology and underclaims estimates for non-ﬁlers.

The estimate of education credit underclaims for non-ﬁlers starts with the list of recipients of Forms

1098T who were not on a tax return either as a primary, secondary or dependent for the tax year in

question. An approximation to the tax unit composition and ﬁling status for the return that the nonﬁling

student would have appeared on if they had ﬁled is made by linking these individuals to tax returns from

three adjacent tax years (2022, 2020 and 2019 for the tax year 2021 estimate and 2023, 2021 and 2020

for the tax year 2022 estimate). In tax year 2021 about 80% of individuals could be matched to a return

as a primary, secondary or dependent or a combination of these in one of the three years. This

percentage dropped to about 76% for the tax year 2022 estimate. If a Form 1098T matches to more

than one of the three adjacent tax years, then the match to the return from the prior tax year is used

ﬁrst, then the subsequent tax year; and then ﬁnally, the return from two tax years prior. 18 When the

student appears as both a primary and dependent on a tax return, the return where the individual

matches to a dependent is used. The nonﬁling Form 1098T recipients who do not appear on a

neighboring year tax return are assumed to be single and not able to be claimed as a dependent.

Once the primary, secondary and dependents and ﬁling status are identiﬁed for the estimated tax return,

students matched to the same primary taxpayer are combined into the same estimated tax return. Then

the primary and secondary (if married ﬁling joint) on the return is/are matched to all the information

returns with income. This income information is then compiled into a mock tax return and fed into a tax

calculator which allows the estimation of adjusted gross income, tax before credits, tax after credits and

then refundable and nonrefundable education credits.

These estimations require making the assumption not only that the composition of the tax return that

would have made the education credit claim is the same as the one identiﬁed from a neighboring year,

but also that the sum of the income found on information returns for the primary taxpayer (and

secondary, if present) on the return accurately represents the income on the tax return that would have

been ﬁled. Form 1099NEC nonemployee compensation is used for net self-employment income. In many

cases, of course, this estimation of income is incomplete or inaccurate since not all types of income is

reported to third parties and oﬀsets are usually not reported. All estimated tax returns are given the

standard deduction and no adjustments to income or nonrefundable credits are allotted except the

subtraction of one half of estimated self-employment income taxes and child tax credits calculated based

on the number of children meeting the age requirement.

Three diﬀerent estimates of education credit underclaims are made: One middle ground estimate and a

lower and upper bound estimate. The middle ground estimate uses the mock tax return to estimate ﬁling

status, adjusted gross income and tax after credits and assumes that all the primary taxpayers who are

under 24 years old are eligible to claim the AOTC as long as they pass the other tests (i.e. the Form 1098T

indicates that they are at least a half time student and not a graduate student and also it is evident from

ﬁlings going back to tax year 2012 that they have not already been claimed in four tax years for

refundable education credits).

For the tax year 2021 estimates, the tax return or returns for TY 2020 are used ﬁrst, then TY 2022 and then TY

2019. For the tax year 2022 estimates, the tax return or returns for TY 2021 are used ﬁrst, then TY 2023 and then

TY 2020.

A second, upper-bound estimate does not use the return line items calculated based on information

return income but instead assumes that the amount of nonrefundable credits that can be claimed is not

limited by tax owed (because tax liability is assumed to be $20,000 for all observations) and that

adjusted gross income is below the phaseout threshold. All taxpayers, including those under 24 years of

age, are assumed to be eligible to claim refundable education credits if they meet the other tests listed

above.

Finally, a third, lower-bound estimate is made which assumes that none of the taxpayers are eligible to

claim refundable education credits (which would be the case, for instance, if they were not pursuing a

degree) and calculates the nonrefundable credit amount based on an estimate of qualiﬁed educational

expenses from the Form 1098T and the income and tax calculated based on income from information

returns for the primary and secondary taxpayers from the neighboring year tax unit.

For all these estimates, the amount of qualiﬁed tuition on the Form 1098T is used as the estimate of

qualiﬁed educational expenses for the purpose of calculating the amount of education credit that can be

taken. In addition, all the returns with ﬁling status married ﬁling separate (based on the neighboring tax

year return) are considered not eligible to claim education credits.

Based on these assumptions, the estimates of the total underclaims of education credits associated with

the approximately 1.5 million nonﬁling students on Forms 1098T in tax year 2021 are $226 million, $1.0

billion and $2.1 billion for the lower-bound, middle ground and upper-bound estimates, respectively. In

the lower-bound estimate, there are about 0.5 million underclaims of education credits and in the upper

-bound estimate about 1.2 million. For the middle ground estimate, about 68.5% of the underclaims are

refundable credits and 31.5% are non-refundable. In addition, about 74.1% of the nonﬁling students

who are estimated to be eligible to claim education credits, appear to be eligible to claim AOTC based on

their Form 1098T and income.

In tax year 2022, the approximately 1.8 million nonﬁling Form 1098T recipients are associated with total

underclaims of education credits amounting to an estimated $258 million, $1.3 billion and $2.6 billion for

the lower-bound, middle ground and upper-bound estimates, respectively. The number of nonﬁling

students failing to make education credit claims is estimated to range from about 0.5 million to 1.5

million with a middle ground estimate of about 1.2 million. For the middle ground estimate, about

70.8% of the underclaims are refundable credits and about 29.2% of the underclaims are nonrefundable. In addition, about 78.1% of the nonﬁling students who are estimated to be eligible to claim

education credits are found to be eligible to claim AOTC.

Table 15a. Counts of Underclaims of Education Credits for Nonfilers, TY 2021 (thousands of returns)

Lower-Bound

Education Credits

AOTC

LLC

Refundable Credits

Non-Refundable Credits

Total Nonfiler F1098Ts

496

1,500

Middle Ground

996

737

258

737

460

Upper-Bound

1,213

807

406

807

1,123

Table 15b. Amounts of Underclaims of Education Credits for Nonfilers, TY 2021 ($ Millions)

Lower-Bound

Education Credits

AOTC

LLC

Refundable Credits

Non-Refundable Credits

226

Middle Ground

Upper-Bound

1,025

944

704

323

2,142

1,736

406

772

1,370

Table 16a. Counts of Underclaims of Education Credits for Nonfilers, TY 2022 (thousands of returns)

Lower-Bound

Education Credits

AOTC

LLC

Refundable Credits

Non-Refundable Credits

Total Nonfiler F1098Ts

545

1,838

Middle Ground

Upper-Bound

1,204

940

264

940

499

1,465

1,005

460

1,005

1,337

Table 16b. Amounts of Underclaims of Education Credits for Nonfilers, TY 2022 ($ Millions)

Lower-Bound

Education Credits

AOTC

LLC

Refundable Credits

Non-Refundable Credits

258

Middle Ground

1,267

1,177

897

370

Upper-Bound

2,600

2,136

464

962

1,638

8. Discussion and Concluding Remarks

Table 17 combines underclaim counts and dollar amount estimates for both ﬁlers and nonﬁlers for all

three approaches: upper-bound, middle, and lower-bound. Under the preferred middle approach, nonﬁlers represent 20.1% of the total estimated education credit underclaim dollar amount in TY2022

(15.5% in TY2021).

Adding total education credit claims to total education credit underclaims gives the total amount of

potential education credits. For TY2021 using the middle approach, this amount totals $20.1 billion

($13.5 billion in claims + $ 6.6 billion in underclaims). The credit participation rate across both ﬁlers and

nonﬁlers is then 67% ($13.5 / $20.1). For TY22 the total amount of potential claims under the middle

approach is $19.0 billion and the participation rate across ﬁlers and nonﬁlers is 66.7% ($12.7 / $18.9).

Table 17: Number and Amount of Education Tax Credit Underclaims for Filers and Nonfilers

TY 2021

TY 2022

Number of Underclaim

Claimed

Amount

Tax Units

Amount

Tax Units

Amount

(millions)

(thousands)

(millions)

(thousands)

Filers, Total

Non-claimants

Claimants

Non-filers

Total Underclaims

5,805

5,447

358

1,213

7,018

Filers, Total

Non-claimants

Claimants

Non-filers

Total Underclaims

5,435

5,091

344

996

6,431

Filers, Total

Non-claimants

Claimants

Non-filers

Total Underclaims

5,017

4,699

318

496

5,513

Panel A. Upper-bound

7,040

640

6,513

527

640

2,142

9,182

640

Panel B. Middle

5,589

640

5,155

434

640

1,025

6,614

640

Panel B. Lower-bound

3,492

640

3,256

236

640

226

3,718

640

Claimed

Amount

5,363

4,982

381

1,465

6,828

6,297

5,742

555

2,600

8,897

610

4,790

4,428

362

1,204

5,994

5,049

4,584

465

1,267

6,316

610

4,137

3,811

326

545

4,682

2,951

2,707

244

258

3,209

610

We ﬁnd that around one-third of apparently-eligible education tax credits are not claimed on tax returns

in tax years 2021 and 2022. Future work plans include proﬁling students underclaimants to identify

opportunities for outreach. In particular, we plan to look for underclaim clusters by return geography and

by educational institution. We continue to explore other rules for determining AOTC eligibility using

available tax administration data.

References

Ackerman, D., Cronin, J.-A., & Turner, N. (2014). Improving Form 1098-T: How a Revised Form Could

Increase Take-Up, Improve Compliance, and Lower Taxpayer Burden. IRS-TPC Research

Conference: Advancing Tax Administration.

Bhargava, S., & Manoli, D. (2015). Psychological Frictions and the Incomplete Take-Up of Social Beneﬁts:

Evidence from an IRS Field Experiment. American Economic Review, 105(11), 3489-3529.

Bulman, G. B., & Hoxby, C. M. (2015). The Returns to the Federal Tax Credits for Higher Education. Tax

Policy and Economy, 29(1), 13-88.

Cronin, J.-A., & Gray-Hancuch, J. (2024). Low-Income Barriers to Claiming Education Tax Credits. Oﬃce of

Tax Analysis Working Paper 125. Retrieved from

https://home.treasury.gov/system/ﬁles/131/WP-125.pdf

Currie, J. (2006). The Take-up of Social Beneﬁts. Public Policy and the Distribution of Income, 80-148.

Guyton, J., Manoli, D., Schafer, B., Sebastiani, M., & Turner, N. (2017). Credits for College: The Impacts of

Information on Take-Up of Tax-Based Student Aid and Enrollment. Proceedings of Annual

Conference on Taxation and Minutes of the Annual Meeting of the National Tax Association 110,

1-20.

Herlache, A., Turner, M. C., Hall, C., & Bell, E. (2021). Increasing Take-up of the American Opportunity Tax

Credit. The IRS Research Bulletin: Proceedings of the 2021 IRS / TPC Research Conference, 171179.

Internal Revenue Service, Department of the Treasury. (2022). Publication 970, Tax Beneﬁts for

Education.

IRS Research, Applied Analytics & Statistics. (2018, August). Compliance Analysis for Education Tax

Credits: Tax Years 2006-2012. Publication 5162.

Statistics of Income, Internal Revenue Service, Department of Treasury. (2012 - 2021). Publication 4801,

Individual Income Tax Returns Line Item Estimates. Retrieved from

https://www.irs.gov/statistics/soi-tax-stats-individual-income-tax-returns-line-item-estimatespublications-4801-and-5385

U.S. Department of Education, National Center for Education Statistics. (n.d.). Table 308.10. Total 12month enrollment in degree-granting postsecondary institutions. Integrated Postsecondary

Education Data System (IPEDS). Retrieved October 2024, from

https://nces.ed.gov/programs/digest/

US Department of Education. (2024). What does cost of attendance (COA) mean? Retrieved October 11,

2024, from https://studentaid.gov/help-center/answers/article/what-does-cost-of-attendancemean

Appendix

Appendix 1. F8863 Education Credits, TY2022

Appendix 2. F8863 Calculations for selected line items as compared to SOI Line Count Estimates,

TY2021

Population

SOI Line Count

Estimate

Count Dollars

Count

Dollars

(K)

($M)

(K)

($M)

Line 1 - Tentative AOTC

6,164 13,630

6,163

13,541

Line 7 - Tentative AOTC after Income Phaseout

6,163 13,285

6,159

13,176

Line 7 - Checkbox

138

Line 8 - Refundable AOTC

6,039

5,216

6,027

5,170

Line 9 - Nonrefundable AOTC

6,163

8,069

6,159

8,005

Line 12 - Tentative LLC Amount

3,086

3,170

3,191

3,230

Line 18 - LLC after Income Phaseout

3,085

3,081

3,190

3,144

Line 19 - Nonrefundable Education credits after credit limit

worksheet

7,946

8,245

8,123

8,280

Appendix 3. Middle Approach Underclaims Estimates for Filers by AGI Credit Component

TY 2021

TY2022

Count Count Amount Amount Number Count Amount

AGI

(k)

Refundable AOTC, F8863 Line 8

<= 0

1.7%

1.8%

1.3%

1 to 25k

424

22.7%

371

23.3%

415

21.8%

365

25k to 50k

524

28.1%

462

29.0%

518

27.2%

457

50k to 75k

312

16.7%

275

17.3%

330

17.3%

292

75k to 100k

202

10.8%

163

10.2%

213

11.2%

172

100k to 200k

371

19.9%

293

18.4%

406

21.3%

320

> 200k

0.0%

Total

1,864 100.0%

1,594

100.0%

1,907

100.0%

1,631

Nonrefundable AOTC, F8863 Line 9

<= 0

1.8%

108

2.0%

1.2%

1 to 25k

1,537 46.8%

2935

54.5%

1,298

41.8%

2451

25k to 50k

778

23.7%

1,194

22.2%

786

25.3%

1,213

50k to 75k

333

10.1%

457

8.5%

359

11.6%

497

75k to 100k

205

6.2%

250

4.6%

216

7.0%

264

100k to 200k

371

11.3%

440

8.2%

406

13.1%

480

> 200k

0.0%

Total

3,285 100.0%

5,384

100.0%

3,103

100.0%

4,969

Lifetime Learning Credit, F8863 Line 18

<= 0

2.2%

101

3.3%

1.5%

1 to 25k

1,313 34.5%

1,304

42.1%

1,148

32.7%

1,191

25k to 50k

1,044 27.4%

780

25.2%

977

27.8%

750

50k to 75k

545

14.3%

396

12.8%

533

15.2%

387

75k to 100k

329

8.6%

215

6.9%

316

9.0%

206

100k to 200k

487

12.8%

304

9.8%

490

13.9%

303

> 200k

0.0%

Total

3,804 100.0%

3,101

100.0%

3,516

100.0%

2,903

Nonrefundable Education Credits

<= 0

0.0%

711

17.8%

0.0%

593

1 to 25k

1,095 21.7%

477

11.9%

793

19.1%

359

25k to 50k

1,728 34.3%

1,489

37.3%

1,295

31.2%

1,190

50k to 75k

856

17.0%

827

20.7%

711

17.2%

677

75k to 100k

521

10.3%

460

11.5%

482

11.6%

422

100k to 200k

833

16.6%

741

18.5%

862

20.8%

770

> 200k

0.0%

Total

5,035 100.0%

3,994

100.0%

4,145

100.0%

3,419

Amoun

1.5%

22.4%

28.0%

17.9%

10.5%

19.6%

0.0%

100.0%

1.3%

49.3%

24.4%

10.0%

5.3%

9.7%

0.0%

100.0%

2.3%

41.0%

25.8%

13.3%

7.1%

10.4%

0.0%

100.0%

17.3%

10.5%

34.8%

19.8%

12.3%

22.5%

0.0%

100.0%

This is a copy of a public record, reproduced as it was published. It is not legal advice, and it may not be the version a court would rely on. Check the official source before you cite it.