Economics Research Paper

This will be a literature review of 6 – 7 journal papers.
I would like to receive one cohesive paper/essay that summarizes all of them at the same time integrated together. It should not a summary of each paper, but an intertwined summary of each of the main topics from each of the articles in each paragraph. each of the papers should be cited at least once or twice.

THREE ESSAYS ON FINANCIAL PLANNING FOR COLLEGE
Prepared by:
Brigham Taft Dorman, AFC®
A DISSERTATION
IN
PERSONAL FINANCIAL PLANNING
Submitted to the Graduate Faculty
of Texas Tech University in
Partial Fulfillment of the Requirements
for the Degree of
DOCTOR OF PHILOSOPHY
Approved:
Michael Finke
Chairperson
Russell James III
Deena Katz
Swarn Chatterjee
Accepted by
Mark Sheridan, Ph.D.
Dean of Graduate School
August 2016
© 2016 Brigham Taft Dorman
Texas Tech University, Brigham T. Dorman, August 2016
ii
ACKNOWLEDGEMENTS
To my supportive wife and children, Shauna, Kayla, and Anna Dorman
I would like to thank my committee members for their continuous support and
contributions while writing this dissertation. I extend my appreciation for the guidance
and dedication of my committee chair, Dr. Michael Finke, in helping me finish this work
before starting my first academic position. He demonstrates a high level of devotion to
research, of which I’m honored to have learned from his skill. My appreciation extends to
the rest of my committee, Deena Katz, Dr. Russell James, and Dr. Swarn Chatterjee.
Special thanks to my fellow research colleagues Harold Evensky, Dr. Barry
Mulholland, and Dr. Rachel Bi for their patients while I completed this dissertation. Their
unwavering support has allowed me to finish before graduating. I would also like to thank
Dr. Vickie Hampton and Dr. Bill Gustafson for teaching me the skills needed, and
providing the opportunities for program development, and their assistance in making a
smooth transition from student to professor.
I would like to thank my fellow Ph.D. student colleagues for their support,
willingness to take charge in group projects, and their understanding of my time
commitment to my research. Thanks to the unforgettable friendships my family and I have
made during our time at Texas Tech University.
Finally, I thank my family for their kind, caring, and loving support, as well as their
various sacrifices that permitted me to pursue this Ph.D. Shauna took charge in helping
support the family financially, emotionally, and remained steadfast and caring during times
of struggle as well as when I needed to dedicate extended hours toward my school work.
Texas Tech University, Brigham T. Dorman, August 2016
iii
She was my ultimate “cheerleader” that rooted me on and stayed the course with me during
this journey. Kayla and Anna brought unmeasurable amounts of joy, support, and
wonderful prayers to our family. Our family has been strengthened, brought closer, and
more meaningful having taken this ride together.
It is an honor to attain the academic level of a Ph.D., especially from such a
dedicated and reputable financial planning program.
Texas Tech University, Brigham T. Dorman, August 2016
iv
TABLE OF CONTENTS
ACKNOWLEDGEMENTS…………………………………………………………………………………… ii
ABSTRACT………………………………………………………………………………………………………….v
LIST OF TABLES……………………………………………………………………………………………… vii
LIST OF FIGURES …………………………………………………………………………………………… viii
I. INTRODUCTION ……………………………………………………………………………………………1
II. THE EFFICIENCY OF 529 PLAN PUBLIC POLICY …………………………………………3
III. 529 TAX SHOCKS: FORGETTING THE UNEXPECTED ………………………………..36
IV. APPENDIX: MODEL OF ACCOUNT CHOICE FOR COLLEGE SAVING
HOUSEHOLDS………………………………………………………………………74
V. ASSET REPOSITIONING & COLLEGE FINANCIAL AID MAXIMIZATION ….83
VI. CONCLUSION…………………………………………………………………………………………….122
Texas Tech University, Brigham T. Dorman, August 2016
v
ABSTRACT
The study of education planning is imperative to households, financial
professionals, and public policy. This dissertation, consisting of three essays on financial
planning for college, investigates the how households may optimize their savings for a
college education. The first essay examines the relationship between participating in 529
education savings plans and college enrollment. The second essay evaluates optimal
account choice, given college completion probabilities. The Third essay investigates
financial aid maximization for households that are penalized for saving.
These three essays advance current knowledge of education planning by providing
insight to the efficiency of public policy and how households with prospective college
students can better plan for the cost of a postsecondary education. This dissertation
supports the idea that 529 education savings plans may be inefficient products for
household college savings. High income households are more likely to participate in 529
education savings plans, but due to low college completion rates, it is important that college
savers are properly informed of the potential penalties when making a sub-optimal account
choice. Financial professionals can provide this important information to consumers.
This study finds a lack of evidence to support the goal of public policy for 529
education savings plans to increase college enrollment. Households that have higher
income, more educated parents, and begin saving early for a child’s college education have
a higher probability of participating in a 529 plan. The use of a 529 plan explains very
little of college completion, and with college completion rates below 50%, households may
be better off avoiding them. This study finds that low income households are better off not
participating in 529 plans due to recent changes in tax law, while middle and high income
Texas Tech University, Brigham T. Dorman, August 2016
vi
households may benefit from 529 plan participation if the child’s probability of completing
college is greater than 50%. Households that do save are penalized on the FAFSA
application. This study finds that these households may maximize financial aid by
repositioning assets into home equity. Findings from this dissertation can help policy
makers, financial professionals, and households make more informed decisions with
respect to college affordability. Contrary to the goal of public policy, 529 education
savings plans, which are designed to reduce the cost of college, may not improve college
enrollment. Agreeing with prior research, low income households are better off not
participating in tax-advantaged accounts.
Texas Tech University, Brigham T. Dorman, August 2016
vii
LIST OF TABLES
1.1: Descriptive Statistics of Saving Timing……………………..……………… 15
1.2: Descriptive Statistics of Education Savers by Account Type…………..…… 18
1.3: Who Saves for a Child’s College Education and When, Multivariate
Logistic Analysis………………………………………………………………… 22
1.4: Multivariate Logistic Analysis on 529 Plan Participation……………..…… 26
1.5: Logistic Regression Analysis on College Enrollment…………………..….. 30
2.1: Degree Completion Coefficients…………………………..………………… 51
2.2: OLS Regression of 529 Participation on College Completion
Probabilities…………………………….………………………………………… 70
3.1: Descriptive Statistics of Households that Completed the FAFSA………..… 97
3.2: Foundational Assumptions of Hypothetical Household………………………….. 99
3.3: Change in Household EFC Score for Every $10,000…………………………….. 106
3.4: Logistic Regression on FAFSA Application Behavior…………………..…. 113
Texas Tech University, Brigham T. Dorman, August 2016
viii
LIST OF FIGURES
1.1: Total Assets in State Sponsored 529 Plans in 2013………………………… 9
1.2: Historical Capital Gains Tax Rates…………………………………………. 10
1.3: Net Tax Benefit for 529 Participation………………………………………. 12
2.1: Annual Amount Paid in Tax Penalties……………………………………… 38
2.2: Projected Annual Amount Paid in Tax Penalties…………………………… 38
2.3: Survival Function for SSA Population – Selected Calendar Years (1900,
1950, 2000, 2050, 2100)………………………………………………………… 40
2.4: Four and Six Year College Completion Rates……………………………… 40
2.5: Probability of College Completion by HS Grade, Total Sample…………… 57
2.6: Probability of College Completion by HS Grade, Non 529 Users…………. 58
2.7: Probability of College Completion by HS Grade, 529 Plan Users…………. 59
2.8: Probability of College Completion by Household Income…………………. 60
2.9: Probability of College Completion by Savings Time Preference…………… 61
2.10: Probability of College Completion by Parent Age/Life Cycle Stage (High
School Freshman Year)…………………………………………………………. 62
2.11: Probability of College Completion by Student Gender…………………… 63
2.12: Probability of College Completion by Student Ethnicity…………………. 63
2.13: Probability of College Completion by Parental Education
Attainment……………………………………………………………………….. 64
2.14: Scenario 1 – Low Income Household (0% Capital Gains Tax)…………… 66
2.15: Scenario 2 – Middle Income Household………………………………………. 67
2.16: Scenario 3 – High Income Household………………………………………… 68
2.17: High Income Growth Household……………………………………………… 69
3.1: Percentage Change in State Funding per Student, Inflation Adjusted, 2008-
2015……………………………………………………………………………… 85
3.2: Tuition vs. Income……………………………………………………………… 90
3.3: Household Home Equity Use for College Expenditures…………………… 93
3.4: Base Case of 30 year, $300k Mortgage at 3.5%……………………………………… 108
3.5: Marginal Change for Asset Re-positioning to Home Equity by Income
Level and Repositioning Amount……………………………………………….. 109
3.6: Marginal Change for Asset Re-positioning to Home Equity When
Increasing Mortgage Interest Rate………………………………………………. 110
3.7: Marginal Change for Asset Re-positioning to Home Equity When
Increasing Mortgage Amount………………………………………………………… 111
3.8: Marginal Change for Asset Re-positioning to Home Equity When
Decreasing Mortgage Amount…………………………………………………… 111
3.9: Why HOWS Fail to Complete the FASFA…………………………………………….. 114
Texas Tech University, Brigham T. Dorman, August 2016
1
CHAPTER I
INTRODUCTION
Households seek more efficient ways to save and pay for college. These essays
concentrate on three common areas where public policy, college saving households, and
financial planners can improve when addressing college funding: the efficiency of 529 plan
public policy, portfolio decision making for college, and asset re-positioning for federal
need-based aid maximization.
529 education savings plans were created to encourage American households to
save for college and reduce the increasing financial burden of affording a child’s postsecondary
education. Prior research demonstrates that 529 plan participation favors high
income households. 529 public policy speculates that 529 plan participation increases
college enrollment for middle-class households. This first essay studies the efficiency of
public policy around 529 education savings plans by first exploring 529 plan participants,
then investigating the impact of 529 plan participation on college enrollment. This essay
uses the High School Longitudinal Study of 2009 (HSLS), a dataset consisting of
approximately 23,000 households with High School freshman. Results suggest that public
policy on 529 plan participation fails to increase college enrollment for low and middle
income households.
Non-qualified withdrawals from 529 plans result in tax shocks. College completion
rates are less than 50% and households facing college completion uncertainties may be
allocating college savings to other financial goals or saving for college inefficiently
because of this unknown, thus failing to minimize unexpected tax shocks. To improve
college saving and affordability efforts, this second essay investigates how college saving
Texas Tech University, Brigham T. Dorman, August 2016
2
households may maximize their savings while minimizing the expected tax liability from
unexpected non-qualified withdrawals by evaluating account choice in connection to
college completion probabilities using the HSLS. Results suggest that it may be suboptimal
for households with prospective college students, of low college completion
probabilities, to participate in 529 plans.
College expenditures have risen while state postsecondary education funding has
declined, placing an increased financial burden on households with college goals. This
challenge increases the demand for household EFC minimization. The Higher Education
Act’s 1992 amendment reclassified home equity from a non-qualified to a qualified asset
that no longer taxes households for saving. EFC minimization, an act to maintain current
consumption during college years, may be achieved through strategic asset re-positioning.
This study investigates the benefit of re-positioning non-qualified assets to home equity in
order to minimize household EFC. Findings suggest that converting non-qualified assets
may be beneficial because it decreases household EFC, reducing the savings gap for
households with both retirement and college savings goals. Additionally, we find evidence
that homeowners with non-qualified savings have higher probabilities of demanding EFC
minimization.
Texas Tech University, Brigham T. Dorman, August 2016
3
CHAPTER II
THE EFFICIENCY OF 529 PLAN PUBLIC POLICY
529 education savings plans were created to encourage American households to
save for college and reduce the increasing financial burden of affording a child’s postsecondary
education. In 2012, $2.1 billion of tax revenues were forgone in 529 plans
(Cramer & Schreur, 2013). College Savings Plans Network (2015) reports that assets in
529 plans grew 165%, through contributions and capital appreciation, from nearly $98
billion to nearly $260 billion from 2005 to 2015. The Office of Management and Budget
(2015) reports that, over the next decade, 529 plans may cause close to $30 billion of lost
tax revenues1
. With the expiration of the deduction for higher education expenses in 2013
and the possible expiration of the American Opportunity Tax Credit (AOTC) in 2017
(Office of Management and Budget, 2015), demand for 529 plans may increase.
Langstraat (2012) and Poterba & Samwick (2003) argue that wealthier households
benefit more from using tax-advantaged accounts, such as Individual Retirement Accounts
(IRAs). Gokhale & Kotlikoff (2003) find that lower-income households experience much
lower net benefit from participating in tax-deferred 401(k) accounts.
Low to middle income households may see significantly less benefit from deferring
educational expenses in tax-sheltered savings accounts. Chien & DeVaney (2002) and
Manly & Wells (2009) show that higher income and more educated households are more
likely to save for college. 529 plan participants receive tax-free growth as long as the

1
“The Congressional Budget Act of 1974 (Public Law 93– 344) requires that a list of ‘tax expenditures’ be
included in the budget. Tax expenditures are defined in the law as ‘revenue losses attributable to provisions
of the Federal tax laws which allow a special exclusion, exemption, or deduction from gross income or which
provide a special credit, a preferential rate of tax, or a deferral of tax liability.’” Estimated revenue losses
are calculated by following tax law as of July 1, 2014, and in present value terms.
Texas Tech University, Brigham T. Dorman, August 2016
4
monies withdrawn are used for qualified education expenses. High income 529 plan
participants benefit more from contributing since, at the margin, they receive a greater tax
break for each dollar contributed (S. M. Dynarski, 2004b; Force, 2009). Mulleneaux
(1999) argues that tax policy on 529 plans is pro-high income. Ryan (2008) suggests
enforcing adjusted gross income (AGI) limits on 529 plans to mitigate the policies
regressive nature.
Prior studies have estimated the tax benefits of 529 plan participation, but disagree
on the magnitude. For example, Dynarski (2004) finds that higher income households pay
as much as 80% less taxes on 529 plan assets versus lower income households. However,
Force (2009) uses a similar method to quantify 529 plan participation tax benefits and finds
that the tax benefit for high income households is less than half (39%) as large2
.
The current tax regime may experience efficiency loss3
(Dynarski & Scott-Clayton,
2006) and sub-optimally serve lower and middle income education savers (Cramer &
Schreur, 2013). Households in the highest marginal tax bracket benefit even when making
non-qualified withdrawals from 529 plans (Dynarski, 2004b; Hoxby, 1998). Dynarski
(2004a) investigates the impact of financial aid policy on household income and agrees
with DeGennaro (2004) that lower and middle income households may be better off saving
in taxable accounts or not saving at all for a child’s education4
.
Few households use 529 plans despite evidence that saving for a child’s postsecondary
education is one of the most important household financial goals, second only

2 Both studies calculate the tax benefits based on including state tax deductions, however at the time of this
writing, six states allow for tax deductions, 17 offer no state tax deduction, while the remaining majority
allow for “potential” tax benefits (J.P. Morgan Asset Management, 2014)
3 Tax incentives encourage households to participate in specific asset locations rather than evaluating optimal
asset location, which may result in lower net equity over the life cycle.
4 Lower income households may experience greater lifetime total taxes and less lifetime total consumption.
Low and middle income households may reduce need-based financial aid eligibility by saving for college.
Texas Tech University, Brigham T. Dorman, August 2016
5
to retirement (Fidelity, 2015; Sallie Mae, 2014). Miller (1997) demonstrates that 92% of
those who save for a child’s education feel it is the best investment they can make for their
child, and 31% list education as their top financial goal. Only 3% of parents who have
saved over half of their child’s education costs strongly disagree that they have a parental
duty to save for their child’s college education. Eid (2010) explains that upholding the
family standard on education is a major reason why parents invest. Some parents value
education because it allows their children to attain a higher socioeconomic status, and high
socioeconomic status households interpret education as a way of making sure the family
name remains at that socioeconomic level. Impoverished households see education as a
way out (Mickelson, 1990), while few low income parents save for their child’s education
to earn compound interest5
(Eid, 2010; Ekos, 2008). According to the Survey of Consumer
Finances (SCF), less than 3% of American households use 529 plans. Households that
save for both retirement and college have 21% higher income and 58% more IRA assets,
on average, than households that only save for retirement (S. M. Dynarski, 2004b;
Government Accountability Office, 2012). Lower and middle income households may not
participate in 529 plans because of limited financial resources or because they focus on
reducing search cost more than tax efficiency (Alexander & Luna, 2005; V. L. Bogan,
2014; Sallie Mae, 2015).

5 Ekos, 2008 investigates why low income Canadian households do not save for college. They find that low
financial illiteracy is positively related to the lack of college savings. Survey participants (n=901) are
questioned about compound interest and specifically if compound interest triggers saving. “What is
compound interest? What does that mean and how does it work? Has your understanding of compound
interest had any effect in motivating you to save? Do you think that in general, parents understand that as a
result of compound interest, they can have a much larger investment for their child’s post-secondary
education if they start early? And do you think that it motivates parents to start saving earlier or do you think
that it doesn’t make a difference?”
Texas Tech University, Brigham T. Dorman, August 2016
6
College enrollment may be influenced by post-secondary education savings
behavior. 60% to 70% of high school graduates enroll in post-secondary education
institutions (Bleemer & Zafar, 2015; Garneau, 2012), but only 23% are prepared for the
rigors of completing a four year degree (Greene & Winters, 2005; Porter & Polikoff, 2011).
Elliott (2009) investigates the gap between college aspiration and college expectation with
low income households and the role of Child Development Accounts (CDAs) in reducing
this gap. He finds that child education savings is a strong predictor of reducing college
aspiration and increasing college expectation. In other words, the mere existence of
education savings may increase the expectations and affordability of college attendance,
but being able to afford a post-secondary education because of 529 plan participation may
not translate directly into greater enrollment probabilities.
College savings and household income may contribute to higher college completion
rates. The U.S. Department of Education (2010) reports that 40% of students attending 4-
year institutions graduate within four years, and 60% graduate within six years. DeAngelo,
Pryor, & Franke (2011) finds that four-year college completion rates are below 40%.
Graduation rates are less than 10% when low to middle income households are not saving
for college, but rise to just over 30% when savings are involved (Elliot, Song, & Nam,
2013). Isaacs et al. (2008) used the PSID to investigate the relationship between college
graduation rates and household income. They find that college completion rates are
approximately 11% when parental income is in the bottom quintile and approximately 53%
when parental income is in the top quintile.
This paper investigates the efficacy of 529 plan public policy. First, we evaluate
when households begin saving for education through 529 plans. Second, we investigate
Texas Tech University, Brigham T. Dorman, August 2016
7
who saves specifically in sheltered education savings accounts. And finally, we examine
whether prospective college students of households who participate in tax sheltered
education savings vehicles are more likely to enroll in college.
Results show that higher-income households, who benefit the most from a taxsheltered
education savings plan, save more and begin saving earlier for a child’s
education. Factors associated with a recognition of the value of higher education, including
parental education, age, and race, are also strong predictors of early saving in 529 plans.
This suggests that parents who were most likely to save for a child’s educational expenses,
in the absence of a tax-sheltered option, are also most likely to take advantage of a 529
plan.
Participation in a 529 plan appears to not increase the likelihood of college
enrollment. This is consistent with Mulleneaux (1999) and Dynarski & Scott-Clayton
(2006). Parental expectations and behaviors consistent with preparation for college are
among the most important predictors of college enrollment.
Background
Concerns over low saving efforts leads to public policy promoting participation in
tax-advantaged accounts. College tuition continues to rise faster than inflation, despite the
fact that tuition of public four-year, in-state institutions, has increased by less than 3% for
the first time since 1974 (Baum & Ma, 2014). 529 plans are primary savings vehicles in
post-secondary education affordability. 529 plans began in Michigan as prepaid tuition
plans in 1986. A decade later, Congress permitted federal tax deferral on earrings in 529
plans through the Small Business Job Protection Act of 1996 (SBA’96) and Internal
Texas Tech University, Brigham T. Dorman, August 2016
8
Revenue Code 529 (IRC 529) was formed.6
Participation in 529 plans surged with the
enactment of the Taxpayer Relief Act of 1997 (TRA) by President Clinton on August 5,
1997. TRA focused on encouraging college enrollment and improving the affordability on
a post-secondary education for low and middle income households (Mulleneaux, 1999;
Staff of the Joint Committee on Taxation, 1997). Additionally it reduced federal tax rates,
made Roth IRA capital gain exemptions permanent, introduced the student loan interest
deduction, established Coverdell Education Savings Accounts (ESA), and exempted IRA
withdrawals for qualified education expenses from some federal taxes. However, the
expected impact on enrollment was opaque.7

From 1996 to 2000 capital gains in 529 plans were tax deferred, until the passing
of the Economic Growth and Tax Relief Reconciliation Act of 2001 (EGTRRA) by
President George W. Bush, granting temporary tax exemption status to 529 plan capital
gains. Through the Pension Protection Act of 2006 (PPA), Congress solidified the
EGTRRA provisions, making qualified post-secondary education expenses permanently
tax exempt for 529 plans. Unlike IRAs and other retirement savings accounts, 529 plans
are regulated at the state level. At least two attempts have been made to revert back to
taxing qualified withdrawals. To date, neither have succeeded.
Since EGTRRA, asset growth in 529 plans have been tax exempt. Figure 1.1 shows
that since EGTRRA 529 savings plans took over the majority of 529 plan assets with 529
prepaid tuition plans dropping from 66% to 48% of the market share.

6
IRC 529 classifies education savings programs that are exempt from taxation as Qualified Tuition Programs.
7 Lyke (1997) reports to congress that “It is also unclear whether the new tax benefits…will result in additional
enrollment or other investment in education. However, it is likely that most of the benefits will accrue to
families whose students would enroll anyway…”
Texas Tech University, Brigham T. Dorman, August 2016
9
This takeover was largely driven by new accounts and contributions to 529 plans
since market performance was bleak in 2001. The S&P 500 Index dropped
approximately 13%. Since then prepaid tuition plans have lost market share every year,
with the exception of 2008. This trend demonstrates the demand for tax-free growth on
education savings assets.
Theoretically, 529 plans may incentivize college saving behavior by providing tax
benefits. Tax-free growth makes 529 plans attractive when assets are used for qualified
post-secondary education expenses.8
However, the comprehension of such benefits has
been questioned since benefits change with tax laws, making 529 plans more complex for

8 Qualified post-secondary education expenses generally are classified as room & board, tuition, books, fees,
supplies, and equipment.
81% 66% 48% 31% 23% 19% 17% 15% 13% 15% 13% 12% 12% 11% 10%
$8 $12 $18
$35
$58
$79
$98
$122
$144
$116
$144
$167 $170
$193
$227
$-
$50
$100
$150
$200
$250
$300
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Billions (2013 Dollars)
Year
Figure 1.1: Total Assets in State Sponsored 529 Plans in 2013
(Source: Baum, Elliott, & Ma, 2014)
Savings Plans
Prepaid Plans
Texas Tech University, Brigham T. Dorman, August 2016
10
planning purposes. A majority of households do not have a plan for affording postsecondary
education expenses (Sallie Mae, 2013).
Force (2009) likens the tax benefits of 529 plan participation to Roth IRAs. The
two main tax advantages of 529 plan participation are tax deferred growth9 and tax free
withdrawals10. Contributions to 529 plans are not tax deductible at the federal level11, thus
the incentive for 529 plan participation is avoidance of capital gains, interest, and dividend
income taxes. As tax rates decline, the demand for 529 plans may also decline. Figure 1.2
demonstrates that, since 1997, capital gains tax rates have declined for low income
households and remained between 15%-20% for high income households. Thus, it may be
expected that 529 plan participation would decline with low income households and remain
fairly level or increase with higher income households, causing a greater participation gap
between household income levels.
State tax benefits often provide education savings, but benefits vary by state. 529
savings plan participants may be eligible for a state tax deduction or be exempt from state

9 Less the withdrawal penalty of 10% if assets are not used for educational purposes.
10 If qualified distributions.
11 Individual states may permit a deduction, but few do.
0%
5%
10%
15%
20%
25%
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
Figure 1.2: Historical Capital Gains Tax Rates
Highest Capital Gains Tax Rate Lowest Capital Gains Rate
Texas Tech University, Brigham T. Dorman, August 2016
11
income tax on qualified withdrawals. Such sheltering allows participants to avoid taxes at
the federal and state levels. Unlike IRAs, 529 plans have essentially no income limitation
for contribution eligibility, but they do have lifetime account balance limits that are set by
each state. Any household of any income level can contribute to a 529 plan. However, for
some households it may be sub-optimal to contribute above the annual gift tax exclusion
limit.
Flexibility options make 529 plans attractive. 529 plan participants retain control
over portfolio choice, can transfer assets among beneficiaries, and can change portfolio
allocation up to twice a year. This means that households could begin saving in a 529 plan
for multiple children once the first child is born. We recognize that 529 plans are attractive
estate planning tools in which contributions can reduce ones taxable estate and still allow
for the estate holder revoke the assets12
.
According to life cycle theory (Attanasio & Weber, 2010; Hanna, Fan, & Chang,
1995; Modigliani & Brumberg, 1954), young households with children are often in the
lower-earning stage of the life cycle and may optimally allocate fewer resources to fund an
educational expense that will occur in the future when earnings will likely be higher. Older
households will be more likely to reduce consumption in the present in order to increase
consumption during the years in which they choose to fund a child’s college education.
Benefit of 529 Plan Participation
The United States tax regime is progressive. Net contributions to 529 education
savings plans result in higher up front tax liabilities for higher income households and
lower up front tax liabilities for lower income households. However, lower income

12 Subject to withdrawal penalties.
Texas Tech University, Brigham T. Dorman, August 2016
12
households bear lower marginal tax rates with respect to capital gains, dividends, and
interest income. Theoretically, households participating in taxable accounts experience no
tax benefits as income increases.
529 plan participation reduces the tax burden of this progressive tax regime. Figure
1.3 shows that the net benefit of 529 plan participation for households in different tax
brackets (income levels). Lower income households experience fewer tax savings. A
positive relationship exists between tax benefits and income. With high income
households, 529 plan participation creates a net benefit greater than low income households
when comparing to taxable account participation.
The use of 529 plans allows us to explore the impact of public policy on college
enrollment. Given the sizeable tax expenditure and relatively greater benefit to higherearning
taxpayers whose children are more likely to attend college, then 529 plan
participation may not achieve its purpose of improving the odds of enrolling in college.
We find that high income households, households that begin saving early, and more
educated households are more likely to participate in 529 plans than lower and middle
6.00%
11.00%
16.00%
21.00%
26.00%
$35K $55k $75k $95k $115k $135k $155k $175k $195k $215k $235k
Figure 1.3: Net Tax Benefit for 529 Participation
Texas Tech University, Brigham T. Dorman, August 2016
13
income, less educated, and households that delay college saving efforts. We find no
evidence to support the goal of public policy that 529 education savings plans improve
college enrollment for middle income households.
Analysis
To evaluate the impact of 529 plan participation on college enrollment, we look at
the High School Longitudinal Study of 2009 (HSLS) dataset provided by the National
Center for Education Statistics (NCES). The HSLS, represented by ten states, follows over
23,000 high school freshman, their parents, school administrators, and counselors from 944
high schools. It contains five waves of data collection beginning in 2009 (base wave) and
focuses on understanding the path high school students take into college and the work
force. It provides household demographic, college enrollment, and 529 plan participation
data.
Table 1.1 shows when households begin saving for college. Early college saving
households, who begin saving for a child’s college education before first grade, make up
only 25% of the sample, but more than half make over $95,000 a year. Late college saving
households, those who begin after first grade, make up 45% of the sample, while the
remaining 30% have yet to begin saving for college. Households with lower income begin
saving later for college. 49% of households that make over $235,000 a year are early
college savers, while 52% of households that make less than or equal to $15,000 have not
started to save for college. A similar trend is found among household education levels.
70% of households with at least a bachelor’s degree are early college savers, while 74% of
households with less than a bachelor’s degree have yet to begin saving for college.
Texas Tech University, Brigham T. Dorman, August 2016
14
Consistent with life cycle theory, younger households delay saving for college.
Households younger than age 35 make up 44% of non-college saving households compared
to 56% of early college saving households who are 45 or older.
Texas Tech University, Brigham T. Dorman, August 2016
15
Table 1.1: Descriptive Statistics of Saving Timing (n=10,431)
Early Savers Late Savers Non-Savers Total Sample
Before
1st Grade
Between
1st & 6th
Between
7th & 9th
Have not
Started All Households
(n=2,560) (n=2,717) (n=1,930) (n=3,224) (n=10,431)
PARENT/HOUSEHOLD DEMOGRAPHICS
Household Income
≤ $15,000 77 (14%) 74 (13%) 115 (21%) 283 (52%) 549 (5%)
15,000-35,000 187 (13%) 269 (18%) 313 (21%) 701 (48%) 1,470 (14%)
35,001-55,000 258 (17%) 355 (23%) 307 (20%) 640 (41%) 1,560 (15%)
55,001-75,000 307 (19%) 424 (26%) 315 (19%) 592 (36%) 1,638 (16%)
75,001-95,000 308 (24%) 352 (27%) 263 (20%) 386 (29%) 1,309 (13%)
95,001-115,000 310 (29%) 329 (31%) 188 (18%) 246 (23%) 1,073 (10%)
115,001-135,000 236 (32%) 235 (32%) 143 (19%) 129 (17%) 743 (7%)
135,000-155,000 218 (36%) 193 (32%) 93 (15%) 100 (17%) 604 (6%)
155,000-195,000 176 (37%) 174 (36%) 70 (15%) 62 (13%) 482 (5%)
195,000-235,000 165 (46%) 104 (29%) 53 (15%) 35 (10%) 357 (3%)
> $235,000 318 (49%) 208 (32%) 70 (11%) 50 (8%) 646 (6%)
Highest Level of Education
≤ High School 653 (16%) 870 (21%) 878 (21%) 1,759 (42%) 4,160 (40%)
Associates 353 (21%) 458 (27%) 331 (20%) 540 (32%) 1,682 (16%)
Bachelor’s 909 (32%) 872 (30%) 465 (16%) 620 (22%) 2,866 (27%)
Graduate Degree 645 (38%) 517 (30%) 256 (15%) 299 (17%) 1,717 (16%)
Parent/Guardian Age
< 35 110 (13%) 158 (19%) 202 (24%) 372 (44%) 842 (8%) 35 - 40 327 (17%) 474 (24%) 408 (21%) 759 (39%) 1,968 (19%) 40 - 45 698 (24%) 791 (27%) 540 (19%) 873 (30%) 2,902 (28%) 45 - 50 839 (30%) 781 (28%) 477 (17%) 731 (26%) 2,828 (27%) 50 - 55 441 (32%) 382 (28%) 205 (15%) 348 (25%) 1,376 (13%) > 55 145 (28%) 131 (25%) 98 (19%) 141 (27%) 515 (5%)
Household Race
White 1,963 (28%) 1,903 (27%) 1,167 (17%) 1,974 (28%) 7,007 (67%)
Asian 169 (22%) 176 (23%) 180 (24%) 239 (31%) 764 (7%)
Black 162 (16%) 270 (26%) 247 (24%) 359 (35%) 1,038 (10%)
Hispanic 163 (14%) 258 (23%) 242 (21%) 481 (42%) 1,144 (11%)
Other race 103 (22%) 110 (23%) 94 (20%) 171 (36%) 478 (5%)
Relationship to the Student
Biological Parent 2,435 (25%) 2,531 (26%) 1,758 (18%) 2,993 (31%) 9,717 (93%)
Non-Biological 125 (18%) 186 (26%) 172 (24%) 231 (32%) 714 (7%)
Marital Status
Married 2,096 (26%) 2,224 (27%) 1,476 (18%) 2,362 (29%) 8,158 (78%)
Non-married 383 (22%) 380 (22%) 340 (20%) 625 (36%) 1,728 (17%)
Never married 81 (15%) 113 (21%) 114 (21%) 237 (43%) 545 (5%)
Number of Siblings 2,560 (25%) 2,717 (26%) 1,930 (19%) 3,224 (31%) 10,431 (100%)
Texas Tech University, Brigham T. Dorman, August 2016
16
Table 1.1: Descriptive Statistics of Saving Timing Continued (n=10,431)
Early Savers Late Savers Non-Savers Total Sample
Before 1st
Grade
Between 1st &
6th
Between 7th &
9th
Have not
Started All Households
(n=2,560) (n=2,717) (n=1,930) (n=3,224) (n=10,431)
Birth Country
United States 2,230 (26%) 2,297 (27%) 1,524 (18%) 2,534 (30%) 8,585 (82%)
Other 330 (18%) 420 (23%) 406 (22%) 690 (37%) 1,846 (18%)
PARENT EXPECTATIONS
Academic Expectation
Associates 117 (15%) 146 (18%) 164 (20%) 377 (47%) 804 (8%)
Bachelor’s 826 (23%) 1,000 (28%) 690 (19%) 1,086 (30%) 3,602 (35%)
Graduate Degree 1,463 (29%) 1,400 (27%) 926 (18%) 1,325 (26%) 5,114 (49%)
Don’t Know 154 (17%) 171 (19%) 150 (16%) 436 (48%) 911 (9%)
PARENT INVOLVEMENT
Homework Assistance
Never 480 (24%) 448 (22%) 343 (17%) 759 (37%) 2,030 (19%)
< 1 day/week 849 (26%) 893 (27%) 570 (17%) 972 (30%) 3,284 (31%) 1 or 2 days/week 780 (22%) 969 (28%) 701 (20%) 1,048 (30%) 3,498 (34%) 3 or 4 days/week 301 (27%) 290 (26%) 221 (20%) 294 (27%) 1,106 (11%) ≥ 5 days/week 150 (29%) 117 (23%) 95 (19%) 151 (29%) 513 (5%) Discuss College Academics No 1,149 (21%) 1,284 (24%) 967 (18%) 1,987 (37%) 5,387 (52%) Yes 1,411 (28%) 1,433 (28%) 963 (19%) 1,237 (25%) 5,044 (48%) STUDENT ABILITY Student Enrolled in Honors Courses No 1,247 (22%) 1,344 (23%) 1,089 (19%) 2,050 (36%) 5,730 (55%) Yes 1,313 (28%) 1,373 (29%) 841 (18%) 1,174 (25%) 4,701 (45%) Academics Outside of School No 2,006 (23%) 2,230 (26%) 1,577 (18%) 2,747 (32%) 8,560 (82%) Yes 554 (30%) 487 (26%) 353 (19%) 477 (25%) 1,871 (18%) Reference Category is italicized Source: The High School Longitudinal Study of 2009 (HSLS:09) provided by the National Center for Education Statistics (NCES). Table 1.2 shows the descriptive statistics of college saving households who use education saving vehicles13 versus those who do not. Starting early on college saving increases the use of education saving vehicles. Approximately 60% of early college saving households use tax advantaged education savings vehicles, while 75% of late college saving households (who delay until at least sixth grade) use other saving vehicles 14. Only 13 529 education savings plans, 529 prepaid tuition plans, Education Savings Accounts (ESAs). 14 May include IRAs, taxable accounts, 401ks, trusts, or other means. Texas Tech University, Brigham T. Dorman, August 2016 17 14% of households that use education savings vehicles delay college saving until as early as sixth grade. Higher socioeconomic status and more sophisticated households take advantage of education saving vehicles. Of households that use education savings vehicles, more than half have annual income greater than $95,000. The vast majority of low income households (less than or equal to $15,000 a year) use non-education saving vehicles15 to save for college. Just over half (52.2%) of middle income households (income between $55,000 and $155,000) do not participate in education savings vehicles, while two-thirds of high income households (greater than $155,000 a year) use education saving vehicles. About 60% of well-educated households (graduate degree) participate in education saving vehicles compared to 65% of poorly educated households (high school) using noneducation saving vehicles. Consistent with life cycle theory, households age 45 and older participate more in tax advantaged education accounts. Younger households dis-save, which may be related to having higher levels of debt. 15 Non-education savings vehicles consist of taxable accounts (i.e. savings or brokerage accounts) or IRAs. Texas Tech University, Brigham T. Dorman, August 2016 18 Table 1.2: Descriptive Statistics of Education Savers by Account Type Education Vehicle Savers Non-Education Vehicle Savers Total Sample 529, Coverdell, Prepaid Tuition Other Saving Vehicle All Households (n=3,366) (n=3,748) (n=7,114) HOUSEHOLD DEMOGRAPHICS Started Saving Before 1st Grade 1,486 (59%) 1,042 (41%) 2,528 (36%) Between 1st & 6th Grade 1,411 (53%) 1,271 (47%) 2,682 (38%) Between 6th & 9th Grade 469 (25%) 1,435 (75%) 1,904 (27%) Household Income ≤ $15,000 89 (34%) 170 (66%) 259 (4%) $15,000 - $35,000 232 (31%) 519 (69%) 751 (11%) $35,001 - $55,000 351 (39%) 554 (61%) 905 (13%) $55,001 - $75,000 415 (40%) 616 (60%) 1,031 (14%) $75,001 - $95,000 404 (44%) 508 (56%) 912 (13%) $95,001 - $115,000 428 (52%) 390 (48%) 818 (11%) $115,001 - $135,000 323 (53%) 285 (47%) 608 (9%) $135,000 - $155,000 278 (56%) 221 (44%) 499 (7%) $155,000 - $195,000 248 (59%) 171 (41%) 419 (6%) $195,000 - $235,000 214 (67%) 104 (33%) 318 (4%) > $235,000 384 (65%) 210 (35%) 594 (8%)
Highest Level of Education
≤ High School 830 (35%) 1,533 (65%) 2,363 (33%)
Associates Degree 469 (42%) 652 (58%) 1,121 (16%)
Bachelor’s Degree 1,227 (55%) 1,000 (45%) 2,227 (31%)
Graduate Degree 840 (60%) 563 (40%) 1,403 (20%)
Parent/Guardian Age
< 35 143 (31%) 319 (69%) 462 (6%) 35 - 40 467 (39%) 724 (61%) 1,191 (17%) 40 - 45 944 (47%) 1,062 (53%) 2,006 (28%) 45 - 50 1,097 (53%) 974 (47%) 2,071 (29%) 50 - 55 545 (54%) 472 (46%) 1,017 (14%) > 55 170 (46%) 197 (54%) 367 (5%)
Household Race
White 2,429 (49%) 2,547 (51%) 4,976 (70%)
Asian 250 (48%) 267 (52%) 517 (7%)
Black 307 (46%) 355 (54%) 662 (9%)
Hispanic 245 (37%) 410 (63%) 655 (9%)
Other race 135 (44%) 169 (56%) 304 (4%)
Relationship to the Student
Biological Parent 3,187 (48%) 3,454 (52%) 6,641 (93%)
Non-Biological Parent 179 (38%) 294 (62%) 473 (7%)
Texas Tech University, Brigham T. Dorman, August 2016
19
Table 1.2: Descriptive Statistics of Education Savers by Account Type Continued
Education Vehicle
Savers
Non-Education
Vehicle Savers Total Sample
529, Coverdell,
Prepaid Tuition
(n=3,366)
Other Saving
Vehicle
(n=3,748)
All
Households
(n=7,114)
Marital Status
Married 2,760 (48%) 2,971 (52%) 5,731 (81%)
Non-married 483 (45%) 602 (55%) 1,085 (15%)
Never married 123 (41%) 175 (59%) 298 (4%)
Number of Older Siblings
None 1,578 (51%) 1,533 (49%) 3,111 (44%)
One 1,116 (47%) 1,251 (53%) 2,367 (33%)
Two or more 672 (41%) 964 (59%) 1,636 (23%)
Birth Country
United States 2,848 (48%) 3,124 (52%) 5,972 (84%)
Other 518 (45%) 624 (55%) 1,142 (16%)
PARENT EXPECTATIONS
Academic Expectation
Associates Degree 142 (34%) 280 (66%) 422 (6%)
Bachelor’s Degree 1,175 (47%) 1,311 (53%) 2,486 (35%)
Master’s Degree 910 (50%) 904 (50%) 1,814 (25%)
Doctoral Degree 951 (49%) 973 (51%) 1,924 (27%)
Don’t Know 188 (40%) 280 (60%) 468 (7%)
PARENT INVOLVEMENT
Homework Assistance
Never 579 (46%) 670 (54%) 1,249 (18%)
< 1 day a week 1,109 (48%) 1,181 (52%) 2,290 (32%) 1 or 2 days a week 1,132 (47%) 1,288 (53%) 2,420 (34%) 3 or 4 days a week 378 (47%) 422 (53%) 800 (11%) ≥ 5 days a week 168 (47%) 187 (53%) 355 (5%) Discuss College Academics No 1,541 (46%) 1,816 (54%) 3,357 (47%) Yes 1,825 (49%) 1,932 (51%) 3,757 (53%) STUDENT ABILITY Student Enrolled in Honors Courses No 1,549 (43%) 2,077 (57%) 3,626 (51%) Yes 1,817 (52%) 1,671 (48%) 3,488 (49%) Academics Outside of School No 2,656 (46%) 3,078 (54%) 5,734 (81%) Yes 710 (51%) 670 (49%) 1,380 (19%) Texas Tech University, Brigham T. Dorman, August 2016 20 Results We conduct a series of regression analyses to investigate the relationship between 529 plan participation and college enrollment.16 First, we investigate who uses 529 plans and look at the timing of when households begin saving for college. Life cycle theory says that households will save in times that income is higher and dis-save when income is lower. We make the following hypotheses': 1. Households with higher income and greater education will have lower odds of delaying college saving behavior. 2. Households with higher income will have higher odds of participating in 529 plans compared to middle income households. 3. Public policy on 529 plans fails to fulfill its intended purpose of increasing college enrollment for middle income17 households. Table 1.3 shows that the odds of a household starting to save before the child enters first grade rises with income. The odds that high income households (between $155,000 and $195,000) will begin saving for a child’s college education before first grade are 79.5% more than middle income households (between $55,000 and $75,000). The odds of a high income household (between $155,000 and $195,000) saving for college between first and 16 In each regression we exclude a wealth variable because it was not available in the data. 17 Because household income status varies across household size, and data provided by HSLS provides preset income categories, we recognize that household income status may not reflect true representation for some households. In an effort to categorize household income status, we define low income as any household that makes less than $55,000 a year, middle income households as any household that makes between $55,000 and $155,000 a year, and high income households as any household that makes more than $155,000 a year. These categories are approximately equal to those reported by Pew Research Center (2015), "The income it takes to be middle income varies by household size, with smaller households requiring less to support the same lifestyle as larger households. For a three-person household, the middle-income range was about $42,000 to $126,000 annually in 2014. However, a one-person household needed only about $24,000 to $73,000 to be middle income. For a five-person household to be considered middle income, its 2014 income had to range from $54,000 to $162,000." Texas Tech University, Brigham T. Dorman, August 2016 21 sixth grade are 44.3% more than a middle income household (between $55,000 and $75,000), and the odds of not saving are 65.4% less than middle income households. The odds that a household with an associates, bachelor's, or graduate degree will begin saving for a child's college education before first grade are 17.5%, 44.3%, and 56.7%, respectively, greater than households with a high school diploma or less. The odds of these similar households not saving for college are 13.7%, 23.4%, and 24.2%, respectively, less than households with a high school diploma or less. Texas Tech University, Brigham T. Dorman, August 2016 22 Table 1.3: Who Saves for a Child's College Education and When, Multivariate Logistic Analysis (n=10,431) Before 1st Grade Between 1st and 6th Grade Between 7th and 9th Grade Non-Savers Variable Odds Ratio S.E. p-value Odds Ratio S.E. p-value Odds Ratio S.E. p-value Odds Ratio S.E. p-value PARENT/HOUSEHOLD DEMOGRAPHICS Household Income ($55,001 - $75,000) < $15,000 0.883 0.130 0.396 0.525 0.076 0.000 *** 0.955 0.124 0.723 1.590 0.172 0.000 *** $15,000 - $35,000 0.773 0.083 0.016 ** 0.732 0.068 0.001 *** 0.993 0.095 0.941 1.370 0.110 0.000 *** $35,001 - $55,000 0.924 0.088 0.409 0.895 0.075 0.189 0.974 0.089 0.771 1.149 0.088 0.069 * $75,001 - $95,000 1.205 0.113 0.047 ** 0.997 0.085 0.974 1.137 0.108 0.175 0.800 0.065 0.006 *** $95,001 - $115,000 1.483 0.143 0.000 *** 1.162 0.104 0.093 * 0.982 0.103 0.861 0.620 0.058 0.000 *** $115,001 - $135,000 1.622 0.171 0.000 *** 1.197 0.120 0.071 * 1.133 0.132 0.283 0.454 0.052 0.000 *** $135,000 - $155,000 1.852 0.208 0.000 *** 1.214 0.131 0.073 * 0.897 0.119 0.416 0.437 0.055 0.000 *** $155,000 - $195,000 1.795 0.215 0.000 *** 1.443 0.165 0.001 *** 0.845 0.125 0.253 0.346 0.052 0.000 *** $195,000 - $235,000 2.775 0.364 0.000 *** 1.055 0.141 0.686 0.861 0.143 0.367 0.248 0.047 0.000 *** > $235,000 2.867 0.310 0.000 *** 1.233 0.133 0.053 *** 0.631 0.093 0.002 *** 0.203 0.033 0.000 ***
Highest Level of Education (≤ High School)
Associates Degree 1.175 0.091 0.036 ** 1.196 0.083 0.009 *** 0.906 0.067 0.184 0.863 0.056 0.023 **
Bachelor’s Degree 1.443 0.099 0.000 *** 1.179 0.075 0.010 *** 0.759 0.056 0.000 *** 0.766 0.049 0.000 ***
Graduate Degree 1.567 0.126 0.000 *** 1.098 0.085 0.224 0.705 0.063 0.000 *** 0.758 0.063 0.001 ***
Parent/Guardian Age (< 35) 35 – 40 1.163 0.142 0.218 1.288 0.136 0.017 ** 0.860 0.087 0.134 0.877 0.079 0.146 40 – 45 1.419 0.167 0.003 *** 1.345 0.140 0.004 *** 0.807 0.081 0.033 ** 0.774 0.070 0.005 *** 45 – 50 1.779 0.213 0.000 *** 1.354 0.146 0.005 *** 0.744 0.078 0.005 *** 0.666 0.063 0.000 *** 50 – 55 2.144 0.274 0.000 *** 1.427 0.167 0.002 *** 0.627 0.075 0.000 *** 0.597 0.064 0.000 *** > 55 2.238 0.350 0.000 *** 1.375 0.199 0.027 ** 0.735 0.110 0.040 ** 0.556 0.076 0.000 ***
Household Race (White)
Asian 0.802 0.107 0.099 * 0.860 0.101 0.198 1.580 0.194 0.000 *** 1.041 0.123 0.731
Black 0.604 0.059 0.000 *** 1.226 0.101 0.014 ** 1.396 0.122 0.000 *** 0.947 0.078 0.506
Hispanic 0.678 0.069 0.000 *** 1.050 0.091 0.572 1.166 0.112 0.110 1.078 0.089 0.367
Other race 0.940 0.115 0.611 0.972 0.112 0.804 1.092 0.134 0.474 1.044 0.112 0.690
Relationship to the Student (Biological Parent)
Non-Biological Parent 0.622 0.069 0.000 *** 1.103 0.104 0.300 1.422 0.140 0.000 *** 0.937 0.087 0.485
Marital Status (Married)
Non-married 1.290 0.092 0.000 *** 0.958 0.067 0.533 1.060 0.078 0.433 0.843 0.053 0.007 ***
Never married 1.245 0.170 0.108 1.073 0.130 0.558 0.912 0.113 0.455 0.877 0.093 0.214
Texas Tech University, Brigham T. Dorman, August 2016
23
Table 1.3: Who Saves for a Child’s College Education and When, Multivariate Logistic Analysis Continued (n=10,431)
Before 1st Grade Between 1st and 6th Grade Between 7th and 9th Grade Non-Savers
Variable
Odds
Ratio S.E. p-value
Odds
Ratio S.E. p-value
Odds
Ratio S.E. p-value
Odds
Ratio S.E. p-value
Birth Country (United States)
Other 0.653 0.065 0.000 *** 0.905 0.076 0.233 1.177 0.108 0.076 * 1.330 0.108 0.000 ***
PARENT EXPECTATIONS
Academic Expectation (Associates Degree)
Bachelor’s Degree 1.140 0.094 0.112 1.362 0.103 0.000 *** 1.190 0.094 0.028 ** 0.681 0.045 0.000 ***
Graduate Degree 1.484 0.120 0.000 *** 1.294 0.097 0.001 *** 1.113 0.087 0.172 0.585 0.038 0.000 ***
PARENT INVOLVEMENT
Homework Assistance (Never)
< 1 day a week 0.867 0.061 0.043 ** 1.158 0.079 0.031 ** 1.144 0.088 0.082 * 0.918 0.060 0.189 1 or 2 days a week 0.808 0.057 0.003 *** 1.263 0.085 0.001 *** 1.297 0.097 0.001 *** 0.794 0.051 0.000 *** 3 or 4 days a week 1.083 0.102 0.394 1.198 0.107 0.044 ** 1.272 0.125 0.015 ** 0.638 0.058 0.000 *** ≥ 5 days a week 1.403 0.167 0.004 *** 1.059 0.129 0.637 1.088 0.143 0.519 0.682 0.079 0.001 *** Discuss College Academics (No) Yes 1.201 0.059 0.000 *** 1.149 0.053 0.003 *** 1.149 0.061 0.009 *** 0.671 0.032 0.000 *** STUDENT ABILITY Student Enrolled in Honors Courses (No) Yes 0.940 0.049 0.233 1.157 0.057 0.003 *** 1.025 0.058 0.660 0.888 0.044 0.017 ** Academics Outside of School (No) Yes 1.253 0.077 0.000 *** 0.942 0.057 0.326 1.003 0.067 0.963 0.841 0.053 0.006 *** Intercept 0.121 0.022 0.000 *** 0.086 0.015 0.000 *** 0.189 0.030 0.000 *** 2.365 0.321 0.000 *** Reference Category in parentheses ***Statistically Significant at the 1% level **Statistically Significant at the 5% level *Statistically Significant at the 10% level Source: The High School Longitudinal Study of 2009 (HSLS: 09) provided by the National Center for Education Statistics (NCES). Texas Tech University, Brigham T. Dorman, August 2016 24 To maximize the tax benefits of a 529 plan, participants would need to start as early as possible and be in higher marginal tax brackets. The second logistic regression investigates which types of households have greater odds of participating in 529 plans. We include variables that capture when households begin saving for college as well as other household demographic and child cognition progress variables. We employ a dichotomous dependent variable of whether or not the participant uses a 529 plan. Table 1.4 shows that the odds of participating in a 529 plan declines as college saving behavior is postponed. The odds that a household that begins saving for college before first grade will use a 529 plan are 243% greater than households that delay saving for college until between seventh and ninth grade. The odds that a household that begins saving for college between first grade and sixth grade will use a 529 plan are 199% greater than households that delay saving for college until between seventh and ninth grade. Table 1.4 also shows that as income rises, households have greater odds of participating in 529 plans. The odds that a household that makes between $95,000 and $115,000 will use a 529 plan are 37.9% greater than households that makes between $55,000 and $75,000. The odds that a household that makes between $155,000 and $195,000 will use a 529 plan are 56.9% greater than households that makes between $55,000 and $75,000. And the odds that a household that makes between $195,000 and $235,000 will use a 529 plan are 129.7% greater than households that make that makes between $55,000 and $75,000. Older and more educated households have greater odds of participating in 529 plans. The odds that a household that has a bachelor's or graduate degree will use a 529 plan are 42% and 51.4% greater than households that have a high school diploma or less, Texas Tech University, Brigham T. Dorman, August 2016 25 respectively. Parents of prospective college students who are between the age of 40 and 45 have 36.4% greater odds of using 529 plans to save for college compared to younger parents (under 35). While parents who are between the age of 45 and 50, and 50 and 55, have 60.8% and 66.68% greater odds of saving through 529 plans, respectively, participation begins to decline after age 55. As household size increases, 529 plan participation declines. For each additional household member, the odds of the household participating in a 529 plan is 16.6% less. When children participate in honors or advanced placement courses, the odds of parents using a 529 plan to save for college are greater. The odds that a household that has a child in honors courses will use a 529 plan are 17.7% greater than households with children who do not take honors courses. Texas Tech University, Brigham T. Dorman, August 2016 26 Table 1.4: Multivariate Logistic Analysis on 529 Plan Participation (n = 7,114) Uses 529 Plan Odds Ratio S.E. p-value Began saving for college (Between 7th & 9th grade) Before 1st grade 3.428 0.241 0.000 *** Between 1st & 6th grade 2.992 0.204 0.000 *** Household Income ($55,001 - $75,000) < $15,000 0.906 0.145 0.534 $15,000 - $35,000 0.773 0.087 0.022 ** $35,001 - $55,000 1.000 0.100 0.998 $75,001 - $95,000 1.100 0.107 0.328 $95,001 - $115,000 1.379 0.137 0.001 *** $115,001 - $135,000 1.384 0.152 0.003 *** $135,000 - $155,000 1.406 0.165 0.004 *** $155,000 - $195,000 1.569 0.198 0.000 *** $195,000 - $235,000 2.297 0.329 0.000 *** > $235,000 1.829 0.215 0.000 ***
Household Highest Degree (≤ High School)
Associates Degree 1.115 0.089 0.174
Bachelor’s Degree 1.421 0.100 0.000 ***
Graduate Degree 1.514 0.125 0.000 ***
Parents Age (< 35) 35 - 40 1.249 0.159 0.080 * 40 - 45 1.364 0.169 0.012 ** 45 - 50 1.608 0.203 0.000 *** 50 - 55 1.668 0.227 0.000 *** > 55 1.494 0.247 0.015 **
Household Race (White)
Asian 0.901 0.120 0.434
Black 1.381 0.137 0.001 ***
Hispanic 0.881 0.092 0.228
Other Race 1.114 0.143 0.399
Relationship to Student (Biological parent)
Non Biological Parent 0.746 0.082 0.008 ***
Marital Status (Married)
Non Married 1.222 0.093 0.009 ***
Never Married 1.476 0.214 0.007 ***
Birth Country (United States)
Other 1.065 0.106 0.527
Family Size
Number of Older Siblings 0.867 0.021 0.000 ***
Texas Tech University, Brigham T. Dorman, August 2016
27
Table 1.4: Multivariate Logistic Analysis on 529 Plan Participation Continued (n = 7,114)
Uses 529 Plan
Odds Ratio S.E. p-value
Academic Expectation (Associates degree)
Bachelor’s Degree 1.145 0.100 0.119
Graduate Degree 1.074 0.092 0.408
Homework Assistance (Never)
< 1 day a week 0.961 0.073 0.599 1 or 2 days a week 1.020 0.078 0.799 3 or 4 days a week 1.039 0.104 0.700 ≥ 5 days a week 1.044 0.139 0.744 Discuss College Academics (No) Yes 1.003 0.053 0.958 Student Enrolled in Honors Courses (No) Yes 1.176 0.064 0.003 *** Academics Outside of School (No) Yes 1.097 0.073 0.164 Constant 0.158 0.026 0.000 *** Reference Group in Parentheses ***, **, * statistical significance at the 1%, 5%, and 10% levels, respectively Pseudo R-Squared = 0.0954 Source: The High School Longitudinal Study of 2009 (HSLS: 09) provided by the NCES. The purpose of 529 plans is to increase college enrollment for middle-class18 households by making it more affordable through tax-advantaged savings. We create dummy variables for 529 plan participation, income categories, and 529 plan participation categorized by household income. This will provide a clearer picture of the impact 529 plan participation and different income levels have on college enrollment behavior. We posit that public policy on 529 plans fails to fulfill its intended purpose of increasing college enrollment for middle income households. The results of our logistic regression of 18 Congress previously drafted financial regulation (EGTRRA of 2001) that permanently exempted 529 college savings plans distributions from income taxation. Congress acknowledges that 529 plans "empower middle-class families to accumulate savings to offset the rising costs of attending college" in the 114th Congress, 1st Session (H.R. 529). Texas Tech University, Brigham T. Dorman, August 2016 28 529 plan participation and household demographics on college enrollment are shown in Table 1.5. 529 plans are recognized as parental assets on the Free Application for Federal Student Aid (FAFSA) and reduce federal need-based aid. Table 1.5 shows that there is a lack of statistical evidence that 529 plan participation increases the odds of college enrollment. Rather, it shows that greater odds of college enrollment are related to household demographics. Household income is positively related to college enrollment. Compared to households with $55,000-$75,000 of income, prospective college students from households with $75,000-$95,000, $115,000-$135,000, and $155,000-$195,000 have 42.4%, 49%, and 62.9%, respectively, greater odds of enrolling in college. Among 529 participating households, few income levels are statistically significant, and only at the 5% and 10% levels. The odds that a child from a 529 participating household with income between $95,000 and $115,000 will enroll in a college degree program are 34.4% less than a 529 participating household with income between $55,000 and $75,000. Parental education attainment, parent age, family size, degree expectations, and honors course participation, are all statistically significant at the one percent level in explaining college enrollment. The odds that a prospective college student will enroll increases by 11.2% for each additional level of parental education attainment. The odds that a prospective college student from a household with expectations that the child will obtain a bachelor's or graduate degree will enroll in a college degree program are 77.9% and 78.7% respectively greater than households that expect an associate's degree. Texas Tech University, Brigham T. Dorman, August 2016 29 The odds that a high school honors course participating student will enroll in a college degree program are 22.3% greater than households with non-honors course participating children. Additional family members of a household require economic resources. For each additional household member, the odds of a prospective college student enrolling declines by 7%. Texas Tech University, Brigham T. Dorman, August 2016 30 Table 1.5 Logistic Regression Analysis on College Enrollment (N = 7,452) Dependent Variable = Enrollment Odds Ratio Std. Err. P-Value 529 Participation 1.227 0.163 0.122 Household Income (> $55,000 and ≤ $75,000)
≤ $15,000 0.529 0.095 0.000 ***
> $15,000 and ≤$35,000 0.793 0.098 0.060 *
> $35,000 and ≤$55,000 0.936 0.111 0.578
> $75,000 and ≤$95,000 1.424 0.179 0.005 ***
> $95,000 and ≤$115,000 1.533 0.214 0.002 ***
> $115,000 and ≤$135,00 1.490 0.234 0.011 **
> $135,000 and ≤$155,00 1.163 0.196 0.368
> $155,000 and ≤195,000 1.565 0.310 0.024 **
> $195,000 and ≤$235,00 1.629 0.395 0.044 **
>$235,000 1.549 0.286 0.018 **
529 Participation by Household Income (> $55,000 and ≤ $75,000)
≤ $15,000 1.287 0.374 0.385
> $15,000 and ≤$35,000 0.983 0.202 0.933
> $35,000 and ≤$55,000 0.884 0.169 0.518
> $75,000 and ≤$95,000 0.707 0.137 0.073 *
> $95,000 and ≤$115,000 0.656 0.132 0.037 **
> $115,000 and ≤$135,00 0.682 0.152 0.086 *
> $135,000 and ≤$155,00 0.920 0.217 0.724
> $155,000 and ≤195,000 0.696 0.183 0.168
> $195,000 and ≤$235,00 0.596 0.179 0.086 *
>$235,000 1.062 0.260 0.806
Saving Timing 0.960 0.033 0.234
Parents Education 1.112 0.025 0.000 ***
Parents Age 1.020 0.004 0.000 ***
Household Ethnicity (White)
Asian 0.681 0.088 0.003 ***
Black 1.152 0.108 0.131
Hispanic 1.136 0.115 0.207
Multi-Race 0.811 0.109 0.120
Other Race 1.021 0.291 0.941
Parental Relationship (Biological)
Non-Biological Parent 0.664 0.066 0.000 ***
Marital Status (Married)
Never Married 0.833 0.110 0.166
Non Married 0.848 0.063 0.026 **
Parent Nationality (U.S.)
Non-US Parent 0.926 0.089 0.419
Family Size 0.930 0.021 0.002 ***
Academic Expectation (Associates Degree)
Bachelor’s Degree 1.779 0.193 0.000 ***
Graduate Degree 1.787 0.195 0.000 ***
No Education Expectation 1.173 0.160 0.242
Discussed Academics 1.014 0.053 0.786
Enrolled in Honors Courses 1.223 0.067 0.000 ***
Academic Participation Outside of School 0.993 0.065 0.915
Constant 0.331 0.078 0.000 ***
Reference Group in Parentheses
***, **, * statistical significance at the 1%, 5%, and 10% levels, respectively
Source: The High School Longitudinal Study of 2009 (HSLS: 09) provided by the NCES.
Texas Tech University, Brigham T. Dorman, August 2016
31
Conclusion
The goal of public policy is to promote specific economic behavior. IRC
529 and TRA were implemented to increase college enrollment by making college more
affordable for middle income households through tax-advantaged savings. Because the
TRA primarily benefits higher-earning households, who have lower financial barriers to
funding higher education, the expected increase in college enrollment for middle income
households may not occur. We investigate whether 529 plan participation, at the middle
income level, increases the odds of a child enrolling in a college degree program.
Empirical results lack statistical evidence that 529 plan participating households
have higher odds of college enrollment over their counterparts. Higher-earning and more
educated households have greater odds of saving for college with 529 plans and lower odds
of delaying college saving or not saving at all. Tax sheltered education savings
opportunities attract those who will benefit the most financially from reducing taxable
income, and those who see the greatest benefit from investing in a child’s college
education. The tax expenditure on 529 plans effectively encourages higher socioeconomic
status parents to save more and earlier for a child’s education.
The same higher-earning households who place a greater value on funding higher
education may not need this tax expenditure to increase the likelihood that their children
will enroll in college. Results show a lack of evidence that 529 plan participation impacts
college enrollment, consistent with our hypothesis that 529 plan participation is less
important than a number of other factors at improving enrollment rates of higher education.
Our results also echo the claim of Dynarski & Scott-Clayton (2016) that educational tax
subsidies are inefficient. Consistent with Ionescu (2009), we find that child cognitive
Texas Tech University, Brigham T. Dorman, August 2016
32
ability and endowed human capital increase the probability of college enrollment. Parental
education attainment, parent age, child degree expectations, and honors course
participation have a stronger impact on post-secondary education enrollment behavior than
529 plan participation. Our results suggest that tax incentives may be a sub-optimal method
for encouraging college savings and improving college enrollment for middle income
households.
Texas Tech University, Brigham T. Dorman, August 2016
33
References
Alexander, R., & Luna, L. (2005). State-Sponsored College 529 Plans: An Analysis of
Factors that Influence Investors’ Choice. Journal of the American Taxation
Association, 27(January), 29–50.
Attanasio, O. P., & Weber, G. (2010). Consumption and Saving : Models of
Intertemporal Allocation and Their Implications for Public Policy. Journal of
Economic Literature, 48(September), 693–751.
Baum, S., & Ma, J. (2014). Trends in College Pricing. The College Board.
Bleemer, Z., & Zafar, B. (2015). Intended College Attendance : Evidence from an
Experiment on College Returns and Costs. Federal Reserve Bank of New York Staff
Report, (Report No. 739).
Bogan, V. L. (2014). Savings Incentives and Investment Management Fees: A Study of
the 529 College Savings Plan Market. Contemporary Economic Policy, 32(4), 826–
842.
Chien, Y.-W., & DeVaney, T. (2002). Children’s Education as the Most Important
Savings Goal. Journal of Family and Consumer Sciences, 94(1), 64–70.
J.P. Morgan Asset Management. (2014). College Planning Essentials: A Comprehensive
Guide to Saving and Investing.
Cramer, R., & Schreur, E. (2013). Personal Savings and Tax Reform: Principles and
Policy Proposals for Reforming the Tax Code. New American Foundation.
CSPN. (2015). College Savings Plans Network 529 Report: An Exclusive Mid-Year
Review of 529 Plan Activity.
DeAngelo, L., Pryor, J. H., & Franke, R. (2011). Completing College : Assessing
Graduation Rates at Four-Year Institutions. Higher Education Research Institute,
Graduate School of Education & Information Studies, University of California, Los
Angeles.
DeGennaro, R. (2004). Asset Allocation and Section 529 Plans. International Journal of
Business, 9(2).
Dynarski, S. M. (2004a). Tax Policy and Education Policy: Collision or Coordination? A
Case Study of the 529 and Coverdell Saving Incentives (No. w10357). National
Bureau of Economic Research.
Dynarski, S. M. (2004b). Who Benefits From the Education Saving Incentives? Income,
Educational Expectations, and the Value of the 529 and Coverdell. National Tax
Journal, 57(2), 359–383.
Dynarski, S. M., & Scott-Clayton, J. E. (2006). The Cost of Complexity in Federal
Student Aid: Lessons from Optimal Tax Theory and Behavioral Economics.
National Tax Journal, LVIV, 319–356.
Dynarski, S., & Scott-Clayton, J. (2016). Tax Benefits for College Attendance (No.
w22127) (Vol. 53).
Texas Tech University, Brigham T. Dorman, August 2016
34
Eid, B. (2010). Parental Saving Behavior: An Analysis of Parent’s Decision to Save and
the Methods Used to Accumulate Savings.
Ekos. (2008). Research into Low Income Parents ’ Views and Experiences Regarding
RESPs , CESG and the CLB.
Elliot, W., Song, H., & Nam, I. (2013). Relationships Between College Savings and
Enrollment, Graduation, and Student Loan Debt. St. Louis, Mo.: Center for Social
Development, Washington University in St. Louis.
Elliott, W. (2009). Children’s College Aspirations and Expectations: The Potential Role
of Children’s Development Accounts (CDAs). Children and Youth Services Review,
31(2), 274–283.
Fidelity. (2015). Fidelity Investments: 9th Annual College Savings Indicator.
Force, W. H. T. (2009). An Analysis of Section 529 College Savings and Prepaid Tuition
Plans.
Garneau, C. (2012). Family Structure, Social Capital, And Educational Outcomes In
Two-Parent Families.
Gokhale, J., & Kotlikoff, L. J. (2003). Who Gets Paid to Save? Tax Policy and the
Economy, 17(MIT Press), 111–139.
Government Accountability Office. (2012). A Small Percentage of Families Save in 529
Plans.
Greene, J. P., & Winters, M. A. (2005). Public High School Graduation and CollegeReadiness
Rates: 1991-2002. Education Working Paper No. 8. Center for Civic
Innovation.
Hanna, S. D., Fan, J. X., & Chang, Y. R. (1995). Optimal life cycle savings. Financial
Counseling and Planning, 6(801), 1–15.
Hoxby, C. M. (1998). Tax Incentives for Higher Education. Higher Education, 12(MIT
Press), 49 – 82.
Ionescu, F. (2009). The Federal Student Loan Program: Quantitative implications for
college enrollment and default rates. Review of Economic Dynamics, 12, 205–231.
Isaacs, J. B., Sawhill, I. V, & Haskins, R. (2008). Getting Ahead or Losing Ground:
Economic Mobility in America.
Langstraat, C. J. (2012). The Individual Retirement Account : Retirement Help for the
Masses , or Another Tax Break for the Wealthy? St. John’s Law Review, 60(3), 437–
451.
Lyke, R. F. (1997). Tax Benefits for Education in the Taxpayer Relief Act of 1997.
Congressional Research Service.
Mae, S. (2013). How America saves for College: Sallie Mae’s National Study of Parents
With Children Under Age 18.
Mae, S. (2014). How America Saves for College: Sallie Mae’s National Study of Parents
with Children Under Age 18.
Texas Tech University, Brigham T. Dorman, August 2016
35
Manly, C. A., & Wells, R. S. (2009). Financial Planning for College : What Parents Do
To Prepare.
Mickelson, R. a. (1990). The Attitude-Achievement Paradox Among Black Adolescents.
Sociology of Education, 63(1978), 44–61.
Miller, E. (1997). Parents’ Views on the Value of a College Education and How They
Will Pay for it. The Journal of Student Financial Aid, 27(1), 7–20.
Modigliani, F., & Brumberg, R. (1954). Utility Analysis and the Consumption Function:
An Interpretation of Cross-Section Data. In Franco Modigliani.
Mulleneaux, N. (1999). The Failure To Provide Adequate Higher Education Tax
Incentives For Lower-Income. Akron Tax Journal, 14, 27–42.
Office of Management and Budget. (2015). Analytical Perspectives: Budget of the U.S.
Government, FY 2016. US Government Printing Office, Washington, DC.
Pew Research Center. (2015). The American Middle Class Is Losing Ground: No longer
the majority and falling behind financially. Washington, D.C.: December.
Porter, A. C., & Polikoff, M. S. (2011). Measuring Academic Readiness for College.
Educational Policy, 26(3), 394–417.
Poterba, J. M., & Samwick, A. A. (2003). Taxation and Household Portfolio
Composition: U.S. Evidence from the 1980s and 1990s. Journal of Public
Economics, 87(1), 5–38.
Ryan, K. A. (2008). Access Assured : Restoring Progressivity in the Tax and Spending
Programs for Higher Education. Seton Hall Law Review, 38(1).
Sallie Mae. (2015). How America Saves for College: Sallie Mae’s National Study of
Parents with Children Under Age 18.
SJCT. (1997). General Explanation of Tax Legislation Enacted in 1997.
Texas Tech University, Brigham T. Dorman, August 2016
36
CHAPTER III
529 TAX SHOCKS: FORGETTING THE UNEXPECTED
Wealthier households benefit more from participating in tax advantaged accounts
than middle and lower income households. However, economic shocks may force
households to withdraw assets early in order to smooth current consumption or meet
unexpected liabilities. For a growing number of households, preparing for college
expenses with a 529 education savings plan brings an unpleasant surprise – owing
unanticipated income taxes and an additional 10% penalty. This unpleasant economic
event, which we refer to as a 529 tax shock, may be overlooked by college saving
households and financial planners.
The United States federal government promotes saving behavior through various
tax-advantaged accounts. For example, retirement savers may use Individual Retirement
Accounts (IRAs) or 401(k)s. 529 education savings plans, 529 prepaid tuition plans,
Education Savings Accounts (ESAs), and Individual Retirement Accounts (IRAs) provide
tax incentives to encourage education savings. These tax-advantaged accounts offer taxdeferred
asset growth and some permit tax-free withdrawals for qualified education
purposes. Early or non-qualified withdrawals trigger penalties. These penalties are meant
to discourage the use of tax sheltered assets for anything other than saving for education.
Although the tax benefits and penalties may encourage account participation, withdrawal
penalties may also discourage investors from participating in these accounts if there is a
possibility that the account may not be used for its intended purpose.
Tax shocks occur in tax sheltered accounts as a result of economic events that
trigger an unanticipated withdrawal. For example, an unexpected tax bill may be levied
Texas Tech University, Brigham T. Dorman, August 2016
37
from the use of a sheltered account as a result of job loss, income reduction, divorce, or
home purchase (Amromin & Smith, 2003). The shift from defined benefit to defined
contribution plans has increased the incidence of tax shocks from retirement account
withdrawals. Retirement accounts provide some protection against unanticipated
consumption shocks, but households who do not account for the tax expense from early
withdrawals may jeopardize long-run goals (Argento, Bryant, & Sabelhaus, 2015).
529 education savings plans are the dominant form of tax-sheltered education
saving in the United States, especially among high income households. These higherincome
households are more than twice as likely to use a 529 plan compared to middle and
lower income households (Sallie Mae, 2015). College Savings Planning Network reports
that there are over 12 million 529 accounts with the average balance of approximately
$21,000.
The bulk of the distributions from these plans will occur in the future – an effect
which will increase the awareness of tax shock consequences. In 2015, only 12% of 529
education savings plans experienced a distribution (College Savings Planning Network,
2015). Figure 2.1 demonstrates that, on average, tax penalties paid on 529 plan
withdrawals increased 20% annually between 2009 and 2013. Figure 2.2 shows that by
2020 nearly $830 million could be paid in unexpected tax liabilities through non-qualified
withdrawals from 529 plans. According to publically available data from CFP Board,
roughly 15% of CFP® practitioners offer education planning as part of their financial
planning services.
Texas Tech University, Brigham T. Dorman, August 2016
38
Figure 2.1: Annual Amount Paid in Tax Penalties
Source: IRS Statistics of Income (SOI)
Figure 2.2: Projected Annual Amount Paid in Tax Penalties
Source: IRS Statistics of Income (SOI), Authors Calculations
$0
$50,000,000
$100,000,000
$150,000,000
$200,000,000
$250,000,000
$300,000,000
2009 2010 2011 2012 2013
TOTAL TAX PENALTIES PAID
YEAR
$0
$100,000,000
$200,000,000
$300,000,000
$400,000,000
$500,000,000
$600,000,000
$700,000,000
$800,000,000
$900,000,000
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020
TOTAL TAX PENALTIES PAID
YEAR
Texas Tech University, Brigham T. Dorman, August 2016
39
Distribution rules for 529 plans are complex (Internal Revenue Service, 2014).
Households and financial planners face multiple challenges such as identifying qualified
and non-qualified education expenses, figuring tax liabilities (both normal and penalties),
reporting distributions, understanding how asset positioning impacts financial aid
eligibility, and claiming losses on 529 plan investments. Non-qualified distributions are
classified as withdrawals taken for expenses other than tuition, fees, books, supplies, and
equipment required for enrollment or attendance. Additionally, students must enroll at an
“eligible educational institution.19” Non-qualified withdrawals from 529 plans carry
multiple consequences. First, non-qualified withdrawals eliminate the tax benefit that
participants initially intended to receive. Second, because of this 529 tax shock, households
have an unexpected tax liability which may push households into higher marginal tax
brackets. Third, there is an opportunity cost to participating in a 529 rather than some other
tax advantaged investment vehicle such as an IRA or 401k.
College degree completion rates may be lower than college savers anticipate – and
are significantly lower than the risk of living to retirement age. Figure 2.3 shows that
individuals born around the year 2000 have about a 90% probability of living to age 59.5,
the age by which retirement account participants can begin withdrawals tax-penalty free.
Figure 2.4 shows that less than four in ten students graduate from college in four years.
Six year graduation rates are higher, approximately 60%, but may come with increased
education costs, fewer years to save for retirement, and potentially increased amounts of
college debt.

19 According to IRS Publication 970, an eligible educational institution is generally “any college, university,
vocational school, or other postsecondary educational institution eligible to participate in a student aid
program administered by the U.S. Department of Education. It includes virtually all accredited public,
nonprofit, and proprietary (privately owned profit-making) postsecondary institutions.”
Texas Tech University, Brigham T. Dorman, August 2016
40
Figure 2.3: Survival Function for SSA Population – Selected Calendar Years (1900, 1950,
2000, 2050, 2100)
Source: U.S. Social Security Administration20
Figure 2.4: Four and Six Year College Completion Rates
Source: DeAngelo et al. (2011)

20 https://www.ssa.gov/OACT/NOTES/as120/LifeTables_Body.html
Texas Tech University, Brigham T. Dorman, August 2016
41
Another challenge faced by households who decide to save in a 529 plan is optimal
asset allocation. If the assets are used for educational purposes, the household faces an
after-tax portfolio allocation decision, but if the assets are not used for qualified educational
purposes (i.e. through a non-qualified withdrawal), households face a tax-deferred
portfolio allocation decision. A sub-optimal asset allocation may occur if savers fail to
consider tax consequences including the risk of failing to use invested assets for a qualified
educational expense.
The primary reason for saving for college is to provide economic resources for postsecondary
education expenses. We investigate how college completion probabilities
impact the college planning process. Results suggest that households that participate in
529 plans, with children of low college completion probabilities, experience greater
expected tax liabilities. We then ask two questions: first, how should college saving
households allocate their savings designated for college among different account options
when faced with uncertain college completion rates in order to minimize the probability of
a 529 tax shock, and, second, does the use of a 529 plan increase the probability of college
completion. We build a framework financial planners may follow to improve education
saving efforts that minimizes the probability of a 529 tax shock.
Literature Review
Life cycle theory (Modigliani & Brumberg, 1954) implies that taking early or nonqualified
withdrawals from tax-advantaged accounts is a function of life event shocks.
Early withdrawal behavior from retirement accounts appear to be driven by life events such
as a change in marital status, job loss or income reduction, purchase of a home, or increase
in family size (Amromin & Smith, 2003; Argento et al., 2015).
Texas Tech University, Brigham T. Dorman, August 2016
42
Taking an early or non-qualified withdrawal in either retirement accounts or a 529
plan triggers the 10% penalty as well as marginal income taxes. Empirical studies show
that wealthier households still benefit from participating in and taking early or nonqualified
withdrawals from tax-advantaged accounts. Bogan & Bogan (1982) and Mano &
Burr (1984) discuss and model the tradeoff between the use of taxable accounts and IRAs
for pre-retirement consumption prior to the age of 59.5. They find that with the 10% tax
penalty in place, wealthier households are still better off saving for pre-retirement spending
needs through an IRA because of the greater sheltering benefit over time. Similar results
for wealthier households are demonstrated with 529 plan participation, but 529 plan
participation limits lower income households from qualifying for financial aid (S. M.
Dynarski, 2004b). Hrung (2007) investigates household choice between tax-deductible
traditional IRAs and non-deductible IRAs (Roth) and finds that marginal tax rates are
positively related to the use of tax-deductible sheltering vehicle.
Specific financial goals such as saving for education or retirement involve the
selection of the most efficient account type. Some literature refers to the process of
selecting an optimal account type (with its own set of tax consequences) as asset location
or withdrawal location. Asset location refers to the decision to allocate different types of
assets (i.e. stocks and bonds) among taxable and tax-advantaged accounts. Tax liabilities
play an important role in asset location decisions. In general, tax-advantaged assets such
as municipal bonds and some equities should be held in tax-disadvantaged accounts (i.e.
taxable accounts), and tax disadvantaged assets such as bonds and REITs should be held
in tax-sheltered accounts (Dammon, Spatt, & Zhang, 2004; Poterba, Shoven, & Sialm,
2000; Shoven & Sialm, 2004; Zhou, 2009). Withdrawal location is the practice of using
Texas Tech University, Brigham T. Dorman, August 2016
43
multiple accounts for tax efficient withdrawal sequencing and greater portfolio longevity
(Horan, 2006; Jennings, Horan, & Reichenstein, 2010; Reichenstein, 2006).
Research on account choice and tax shocks from a college savings perspective is
limited. A student’s decision to not pursue or to discontinue their college education21 is a
life event that may trigger 529 tax shocks for households participating in a 529 plan (if the
529 plan assets are not transferred to another qualified beneficiary22). Assets held in Roth
IRAs and taxable accounts are not affected by this choice because Roth IRA assets can
then be re-designated for retirement and taxable account assets can be assigned to any
financial goal.
Tax benefits are maximized when 529 plans are used explicitly for qualified college
expenses rather than just reaching a specific age (i.e. 59.5 for retirement withdrawals).
Dynarski (2004a) shows that low income households make sub-optimal college savings
decisions when participating in 529 plans through naïve tax and financial aid planning.
Terry & Goolsby (2003) find that the benefits of using 529 plans for retirement
consumption are positively related to investment returns and marginal tax rates.
Few studies address account diversification for college savings goals. Horan
(2003) proposes that household withdrawal tax rates are critical to making an optimal
account choice. He compares 529 plans and ESAs to Roth IRAs and finds that households

21 Research shows that dropping out of college does not hinge upon solely academic failure (Duncheon,
2015).
22 According to IRS rules (Internal Revenue Service, 2014), transferring 529 plan assets between
beneficiaries is a tax exempt transaction. Qualified beneficiaries are the current beneficiaries “spouse, son,
daughter, stepchild, foster child, adopted child (or a descendant of any of them), brother, sister, stepbrother,
stepsister, father, mother (or an ancestor of the father or mother), stepfather, stepmother, niece, nephew, aunt,
uncle, son-in-law, daughter-in-law, father-in-law, mother-in-law, brother-in-law, sister-in-law, the spouse of
any individual previously listed, or a first cousin.” To the best of our knowledge, and at the time of this
writing, no study investigates the probability of a household transferring 529 plan assets between qualified
beneficiaries in the event of a 529 plan beneficiary choosing to discontinue their post-secondary education.
We leave this topic open for future research.
Texas Tech University, Brigham T. Dorman, August 2016
44
can make more tax efficient withdrawal decisions by diversifying across account types.
DeGennaro (2004) suggests that not only should some households avoid 529 plans
completely, but households should consider college savings accounts both inside and
outside a 529 plan for multiple reasons. For example, diversifying college savings across
account types provides flexibility for reallocation, if necessary. Additionally, it avoids
over-reliance on age based funds23, which have been questioned and shown to
underperform (D. M. Blanchett, 2011; Sandhya, 2011). Gokhale & Kotlikoff (2003),
although they do not look at 529 plans specifically, agree with DeGennaro (2004) that
avoiding tax-advantaged accounts may be optimal for some lower-income households.
Previous studies confirm that household and student characteristics affect college
completion rates (Magolda & Astin, 1993; Pascarella & Terenzini, 2005; Weidman,
Pascarella, & Terenzini, 1992). For example, household wealth and income are positively
related to college completion rates (Conley, 2001). Haveman & Wilson (2007) investigate
the effect of household income on educational attainment and find that college students
from households in the bottom quartile of the income distribution carry an 8% probability
of graduating, compared to a 30% probability for college students from households in the
top quartile24. Parental educational attainment is also positively related to college
completion rates. DeAngelo, Pryor, & Franke, (2011) study retention and degree
completion of over 200,000 students across 356 four-year non-profit institutions. They
find that college students from households with parents who have no college experience
(first generation college students) have a 14% lower probability of graduating.

23 Age based funds are synonymous to Target Date Funds. The term “Age Based” is commonly found in 529
plans, while the term “Target Date” is universal with retirement accounts.
24 The author does not identify whether these rates pertain to 4, 5, or 6 years of college attendance.
Texas Tech University, Brigham T. Dorman, August 2016
45
Academic preparation is positively related to college success, but many who enroll
are underprepared for the rigors of college life. Haveman & Wilson (2007) report that high
school graduation rates are 92% and 76% for high school students from households in the
top and bottom income quartiles, respectively. Additionally, they show that college
enrollment ranges between 21% and 64% of high school graduates. Greene & Winters
(2005) perform a detailed study on college preparation and completion rates. They find
that from 1991 to 2002, the number of students prepared for college increased only 9%,
from 25% to 34%.
Students are taking longer to complete their college education. The U.S.
Department of Education, through the NCES, reports that between 1996 and 2008, the
number of college students completing their undergraduate degree in six years has
increased from 55% to 60%25. Four-year college completion rates have been shown to be
below 40%, while six-year college completion rates may be as high as 61% (DeAngelo et
al., 2011). Attewell, Heil, & Reisel (2011) suggest that academic preparation is the
strongest predictor of college graduation, especially for students attending four-year
institutions, while financial aid is the most significant factor predicting graduation for
students attending two-year institutions.
This paper contributes to the existing literature by exploring the impact of college
completion probabilities by incorporating the risk of 529 tax shocks. First, we calculate 4,
5, and 6-year college completion rates among 9th and 12th graders following the
framework of DeAngelo et al. (2011). Second, we distinguish graduation rates among 529
plan participating and non-participating households. Next, we estimate a model to

25 https://nces.ed.gov/ipeds/trendgenerator/tganswer.aspx?sid=7&qid=20
Texas Tech University, Brigham T. Dorman, August 2016
46
determine account choice based on college completion probabilities. Finally, we perform
a least squares regression analysis to evaluate the relationship between 529 plan
participation and college completion rates. We limit account choice to 529s, Roth IRAs,
and taxable accounts and assume that 529 savers plan on using the funds for qualified
educational purposes only as well as spreading the education savings across all years of
attendance.
Tax Benefits of College Saving and Participation
Households save for college, participate in tax advantaged accounts, and incur
college expenditures primarily for the tax benefits and the vast majority of public policy
around education affordability focuses on college cost. In this section we briefly review
the menu of tax benefits that households receive by saving for and paying college
expenditures – tax deferment, income limitation rules, dependency deductions, tax credits,
and student loan interest deduction. For a more detailed review of the tax benefits for
college, see Dynarski & Scott-Clayton (2016). It is important to note that the complexity
of the tax laws make optimal tax planning difficult for the average households, thus many
households make sub-optimal decisions when planning college tax benefits (Turner, 2011).
Most households are familiar with the tax free growth on 529 savings. Thus, starting
early and maximizing education saving contributions can substantially reduce the cost of
college. Sallie Mae (2015) shows that a greater number of millennials are saving for
college, compared to generations before them. However, this cohort is participating less
in the financial markets26
.

26 Gallup. “Just over Half Americans Own Stocks, Matching Record Low.” Available at
http://www.gallup.com/poll/190883/half-americans-own-stocks-matching-record-low.aspx
Texas Tech University, Brigham T. Dorman, August 2016
47
Income is one of the most common measures when investigating college
enrollment, attendance, and even graduation. Higher income households, not only have
greater income tax benefits due higher marginal tax brackets, but benefit from the “no
income limit” provision for 529 plan participation. Unlike traditional IRAs, which phase
out contribution deductibility, and Roth IRAs, which phase out participation, through
income levels, 529 plans allow households of any income level to fully participate27
without phase outs. This means that high income households can benefit from participating
in 529 plans even when non-qualified withdrawal penalties apply (Dynarski, 2004b; Smith,
2010).
Households without prospective college students are ineligible for the extended
child dependency deduction. Traditionally, child dependency is claimed up to the age of
eighteen. However, households with children in postsecondary education institutions are
eligible for the dependency deduction up to the age of twenty-three. Since children from
higher income households have greater probabilities of enrolling in college, there may be
a positive relationship between income and the dependency deduction extension.
Tax credits, which reduce tax liabilities dollar-for-dollar and are subject to income
phase outs, primarily benefit households during college enrollment. College tax credits,
such as the American Opportunity Tax Credit (formerly known as the HOPE Credit), are
complex and cannot be overlapped, thus claiming them requires careful tax planning.
College tax benefits even exist post-graduation. College graduates can deduct their
student loan interest. Student loan interest is an above the line item, which means it is a
dollar-for-dollar deduction from a household’s total income. Unlike mortgage interest,

27 Full participation depends on contribution laws, including state maximums.
Texas Tech University, Brigham T. Dorman, August 2016
48
which requires the household to itemize their deductions, and is primarily found among
wealthier households, student loan interest can be deducted by all income levels.
Parent Perception on Child Educational Attainment
Research demonstrates that people overestimate personal skill and future outcomes
(Dunning, Heath, & Suls, 2004; Maxwell & Lopus, 1994). For example, Svenson (1981)
investigates perception of driving skills and shows that people believe they are above
average drivers. Klein & Weinstein (1997) find that households believe that negative
events are less likely to occur in their lifetime. Even investors think they are smarter than
they really are. However, this idea of unrealistic optimism (Weinstein, 1980) is found to
be skewed towards positive and more common events (Moore, 2007).
Parents demonstrate faith in their child’s ability to succeed academically by
participating in 529 plans, but not every child that enrolls in college will graduate.
Evidence shows that parents overestimate the cognitive ability of their children (Miller,
1986). Thus, if parents miscalculate their child’s cognitive ability, they may be less likely
to help the child prepare for future educational challenges, and underestimate the
probability of experiencing a 529 tax shock in the event that their child chooses not to
complete their postsecondary education, while failing to utilizing 529 assets for qualified
college expenses.
Data
We use data from the High School Longitudinal Study of 2009 (HSLS) dataset
provided by the National Center for Education Statistics (NCES), a nationally
representative longitudinal study representing over 23,000 high school freshman, their
parents, school administrators, and counselors from 944 high schools. It contains five
Texas Tech University, Brigham T. Dorman, August 2016
49
waves of data collection beginning in 2009 (base year) and focuses on understanding the
path high school students take into college and the work force. The HSLS fits well with
this study because it includes household demographics, college enrollment, student high
school academic performance, and 529 plan participation – including the time horizon of
plan participation.
This paper focuses on college savings and account choice. We discard from the
sample households that failed to provide information regarding their college savings plans.
The total sample consists of 6,119 households that indicated they are saving for a child’s
college education. We divide our sample into two groups: 529 participating households
(n=2,938) and non-529 participating households (n=3,181). This means 48% of the sample
uses 529s while the remaining 52% use some other method to fund college expenses. The
rationale behind dividing the sample as such is to help understand the impact 529 plan
participation has on college completion, and whether or not 529 plan participation provides
any significant advantages over saving via other methods with respect to affording college
expenses.
Measuring College Completion Probability
In this section we measure college completion probabilities of high school students
based on student characteristics. We compare those of both ninth (freshman) and twelfth
(senior) grade students who have indicated whether they are saving for college.
We first estimate the college completion probability (Pc) for each student in our
total sample by following the framework of DeAngelo et al. (2011). To calculate the
probability of completing a 4, 5, or 6-year degree we employ the following equations:
Texas Tech University, Brigham T. Dorman, August 2016
50
() =
(
)
[1+(
)]
, and [1]
=∝ + 11 + 22 + ⋯ + , where [2]
Xn = independent variable of student heterogeneity
βn = regression coefficient
α = constant
Xi = dependent variable of degree completion for a prospective college student (i).28
The college completion rates for high school freshman and seniors are a function
of High School grades (GPA), grade level math examination score (MATH), gender
(Gender), and race (RACE). Beta coefficients are obtained from the data merging of the
Cooperative Institutional Research Program’s (CIRP) Freshman Survey and the National
Student Clearinghouse (NSC), produced by Higher Education Research Institute (HERI).
They examine retention and college degree completion of over 210,000 college students
from more than 350 four-year non-profit institutions. Table 2.1 shows, with whites and
males as reference groups for race and gender, the beta coefficient values used in
calculating the students degree completion value (Xi), which is then placed into Equation
1 to compute the college completion probability for Xi.

28 Due to the HSLS data not having college completion data at the time of this writing, we employ the
coefficient values from DeAngelo et al. (2011).
Texas Tech University, Brigham T. Dorman, August 2016
51
Table 2.1: Degree Completion Coefficients
Input Variable β Coefficient
4-Year 5-Year 6-Year
Student GPA 0.309 0.323 0.322
Cognitive Grade Level Test Score 0.03 0.025 0.023
Gender: Female 0.558 0.322 0.232
Race: American Indian -1.003 -0.844 -0.82
Race: Asian -0.06 0.181 0.328
Race: African American -0.25 -0.278 -0.223
Race: Hispanic -0.337 -0.264 -0.149
Race: Other -0.158 -0.06 0.022
Race: Multicultural -0.321 -0.326 -0.309
Source: DeAngelo et al. (2011)
Lastly, we determine the anticipated probability of college graduation, at the senior
year of high school, for Xi by averaging Xi’s 4, 5, and 6-year college completion
probabilities.
Model of Account Choice for College Saving Households
The model in the Appendix illustrates optimal account choice. Account choice may
be driven by a function of the probability of college completion, college non-completion,
and the after-tax wealth accumulation or expected tax liability, resulting in future expected
wealth or the future expected tax liability. The probability of college completion (Pc) and
non-completion (1-Pc) is obtained by following the equations outlined in the data section.
The account choice with the highest expected wealth or smallest expected tax liability
(smallest tax shock) provides households with the greatest expected assets available for
consumption for college. We investigate the expected after tax wealth accumulation and
expected tax liability across twelve different account choice options. These options are:
Texas Tech University, Brigham T. Dorman, August 2016
52
A. 100% 529 plan
B. 100% Roth IRA
C. 100% Taxable account
D. 50% 529 plan/50% Roth IRA
E. 50% 529/50% Taxable account
F. 50% Roth IRA/50% Taxable account
G. 75% 529 plan/25% Roth IRA
H. 75% 529/25% Taxable account
I. 75% Roth IRA/25% Taxable account
J. 25% 529 plan/75% Roth IRA
K. 25% 529/75% Taxable account
L. 25% Roth IRA/75% Taxable account
To calculate the after tax expected wealth for the use of each account type we use
the following formulas:
() = + (1 − ) [3]
= − [4]
Roth IRA:
ℎ = [ [
(1+)

] (1 + ) + (1 + )

] − [( − )()] [5]
Taxable Account:
= [ [
(1+)

] (1 + ) + (1 + )

] − [( − )()] [6]
Texas Tech University, Brigham T. Dorman, August 2016
53
529 Education Savings Plan:
529 = [ [
(1+)

] (1 + ) + (1 + )

] − [(( − )()) + (( − )())] [7]
were
= [
(1 + )

] (1 + )
= (1 + )

= +
= + () +
= ℎ ℎ
=
1 − =
=
=
=
=
= ℎ
Texas Tech University, Brigham T. Dorman, August 2016
54
To calculate the expected tax liability (T) for each of the account choice options,
we employ the following equation:
(
) = + (1 − ), where [8]
= ℎ , and
= () + [( − )()] + [(( − )()) + (( − )())]. [9]
Given the complexity and sensitivity of this study, we assume that if a household
uses a 529 plan, the 529 plan is spread across all years of the child’s post-secondary
education. Thus, non-qualified funds, either used premature to graduation or left over
after graduation are subject to taxes and penalties. This assumption is to simplify the fact
that parents could dissolve the 529 funds in the earlier college years, and to the best of
our knowledge, there is a lack of data to support when parents either fully use up 529
funds or transfer them to another qualified beneficiary.
Non-Qualified Withdrawal Penalties – Post Great Recession
According to the IRS Statistics of Income (SOI), on average, penalties from nonqualified
withdrawals have increased by 20% annually. This trend may be a result of
market downturn during the Great Recession, of which high time preference households
may have withdrawn funds from 529 plans to smooth current consumption. This section
considers the trend of recent non-qualified withdrawals using data from 2009 to 2013. To
the best of our knowledge, this is the only data available to evaluate the trend of nonqualified
education expense behavior. Evidence suggests that if current trends continue,
American households could pay approximately $830 million in tax penalties for nonqualified
withdrawals from 529 plans by 2020.
Texas Tech University, Brigham T. Dorman, August 2016
55
Regression Analysis on 529 Plan Participation and College Completion
We finalize our analysis by evaluating the relationship between 529 plan
participation and college completion probabilities by employing the following ordinary
least squares regression:
Anticipated Probability of College Graduation (Pc) = α + β 1
529+ β 2GPA+ β 3 Female + β4 Test + β 5 Asian+ β6 American
Indian + β7 Black + β8Hispanic+ β 9 Multi Race + β10 Other
Race + ɛ
Where Pc is a continuous variable representing the probability that the prospective
college student will graduate with a post-secondary education degree between four and six
years; 529 is a dummy variable representing the 529 plan participation status of the
respondent; GPA is a continuous variable demonstrating the overall cognitive skill of the
prospective college student; Female is a gender dummy variable; Test is a continuous
variable corresponding to the respondent’s scores on a mathematics test. We posit that 529
plan participation will be positively related to college completion probabilities. Yet,
Dorman (Chapter II) demonstrates that there is a lack of evidence supporting public policy
that 529 plan participation improves college enrollment. Likewise, we expect 529 plan
participation to have little explanatory power in predicting college graduation rates.
Results
We model account choice for college saving households. First, we calculate college
completion rates among High School freshman and seniors, following the framework of
DeAngelo et al. (2011). Second, we distinguish graduation rates among 529 plan
participating households and non-participating households. Next, we demonstrate a model
Texas Tech University, Brigham T. Dorman, August 2016
56
to determine account choice based on college completion probabilities. Finally, we
perform a least squares regression analysis to evaluate the relationship of 529 plan
participation with college completion rates.
Figure 2.5 shows that mean graduation rates are below 50%. High School freshman
have a 43.82% probability of completing college. Rates increase by only 0.4% by the time
they graduate from High School. Figures 2.6 and 2.7 show that only a 5% college
completion probability gap exists between 529 participating and non-529 participating
households. High School freshman from non-529 participating households have a 41.4%
probability of completing college. By High School graduation, their college completion
probability rises to 41.68%. This results in only a 0.28% increase in college completion
probability. High School freshman from 529 participating households have a 46.6%
probability of completing college. By High School graduation, their college completion
probability rises to 46.97%. This results in about a 0.37% increase in college completion
probability.
Texas Tech University, Brigham T. Dorman, August 2016
57
Figure 2.5: Probability of College Completion by HS Grade, Total Sample (N = 6,119)
College completion estimates are calculated from heterogeneous characteristics regarding
cognitive ability and personal demographics. High school freshman have a mean
probability of 31.81%, 47.72%, and 51.92% for completing college within four, five, and
six years, respectively. High school seniors have a mean probability of 32.31%, 48.10%,
and 52.25% for completing college within four, five, and six years, respectively. For the
total sample, the mean probability of college completion is 44.02%.
6-Yr Senior
6-Yr Freshman
5-Yr Senior
5-Yr Freshman
4-Yr Senior
4-Yr Freshman
0
200
400
600
800
1000
1200
1400
Students
Probability of College Completion
Texas Tech University, Brigham T. Dorman, August 2016
58
Figure 2.6: Probability of College Completion by HS Grade, Non 529 Users (N = 3,181)
High school freshman have a mean probability of 29.48%, 45.21%, and 49.53% for
completing college within four, five, and six years, respectively. High school seniors have
a mean probability of 29.86%, 45.46%, and 49.72% for completing college within four,
five, and six years, respectively. For non 529 participating households, the mean
probability is 41.54%.
Source: Authors Calculation using HSLS:09; DeAngelo, Pryor, & Franke, (2011)
0
100
200
300
400
500
600
700
800
4-Years
HS Freshman
HS Senior
0
100
200
300
400
500
600
5-Years
HS Freshman
HS Senior
0
100
200
300
400
500
600
6-Years
HS Freshman
HS Senior
Texas Tech University, Brigham T. Dorman, August 2016
59
Figure 2.7: Probability of College Completion by HS Grade, 529 Plan Users (N = 2,938)
High school freshman have a mean probability of 34.33%, 50.96%, and 54.52% for
completing college within four, five, and six years, respectively. High school seniors have
a mean probability of 34.96%, 50.96%, and 54.99% for completing college within four,
five, and six years, respectively. For 529 participating households, the mean probability is
46.79%.
Source: Authors Calculation using HSLS:09; Deangelo, Pryor, & Franke, (2011)
0
100
200
300
400
500
600
4-Years
HS Freshman
HS Senior
0
100
200
300
400
500
600
5-Years
HS Freshman
HS Senior
0
100
200
300
400
500
600
6-Years
HS Freshman
HS Senior
Texas Tech University, Brigham T. Dorman, August 2016
60
To maximize the tax benefits of a 529 plan, participants would need to start as early
as possible and be in higher marginal income tax brackets. Figure 2.8 shows that for both
529 and non-529 participating households, the probability of college completion is
positively related to parental household income. Figure 2.9 shows that, among 529
participating households, starting to save before the child enters first grade, college
completion probabilities increase by nearly 6%, compared to 529 participating households
that postpone saving efforts until early High School years.
Figure 2.8: Probability of College Completion by Household Income
College completion probabilities are evaluated across 4, 5, and 6-year college completion
rates. At the mean, prospective college students from 529 participating households have a
1.91% greater probability of completing college compared to non-529 participating
households. The gap of college completion probability is the smallest among high income
households.
0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
70.00%
80.00%
90.00%
100.00%
Probability of Completing College
Household Income
529 Participants Non 529 Participants
Low Income Middle Income High Income
Texas Tech University, Brigham T. Dorman, August 2016
61
Figure 2.9: Probability of College Completion by Savings Time Preference
Figure 2.10 shows that middle aged 529 plan participating households have a
greater college completion probability. For early life cycle households, college completion
probabilities are no different between 529 plan participants and non-529 plan participants,
and college completion rates are the lowest.
48.37% 46.24%
42.38%
0%
20%
40%
60%
80%
100%
Before 1st Grade Between 1st & 6th Grade Between 7th & 9th Grade
Probability of College Completion
When Household Began Saving in 529 Plan
Texas Tech University, Brigham T. Dorman, August 2016
62
Figure 2.10: Probability of College Completion by Parent Age/Life Cycle Stage (High
School Freshman Year)
Figures 2.11, 2.12, and 2.13 show that among prospective college students from
households participating in 529 plans, female, non-white, and whose parents have higher
post-secondary education have higher college completion probabilities. Female students
have a 10% greater probability of graduating from a four-year degree granting institution
compared to a males. Non-white students from 529 participating households have a 5%
greater probability of completing college, compared to prospective college students from
white households. And 529 participating students who have parents with a bachelor’s
degree have a 50% probability of graduating from college.
0%
10%
20%
30%
40%
50%
60%
<35 35-40 40-45 45-50 50-55 >55
Probability of College Completion
Parent Age
529 Participants
Non 529 Participants
Texas Tech University, Brigham T. Dorman, August 2016
63
Figure 2.11: Probability of College Completion by Student Gender
Figure 2.12: Probability of College Completion by Student Ethnicity
41.42%
36.62%
51.85%
46.26%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
529 Participants (n=2,938) Non-529 Participants (n=3,181)
Probability of Completing College
Male
Female
45% 42%
47% 48% 48% 47%
40% 37%
42% 43% 43% 42%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Probability of College Completion
Student Ethnicity
529 Participants
Non 529 Participants
Texas Tech University, Brigham T. Dorman, August 2016
64
Figure 2.13: Probability of College Completion by Parental Education Attainment
Households may choose an account composition that produces the lowest expected
tax liability. Figures 2.14, 2.15, 2.16, and 2.17 demonstrate the results of our model on
four hypothetical scenarios and show that prospective college students with a less than 55%
minimum probability of completing college experience higher expected tax liabilities.
Figure 2.14 shows that low income households maximize college savings in taxable
accounts, while Roth IRAs provide the second lowest expected tax liability. 529 plan
participation results in the highest expected tax liability. Even allocating assets between a
Roth IRA and a taxable account produces a lower expected tax liability, primarily for
students with less than a 70% probability. Only 0.5% (n=30)29 of our entire sample are
students from low income households with a college completion probability of 70% or

29 To qualify as a low income household, income must be less than $36,900, demonstrating a 0% capital gains
tax. The seven respondents are based off a maximum income level $35,000 due to data limitations. Then
next income level choice for respondents is between $35,000 and $55,000.
38% 41%
50%
53%
36% 39%
46%
51%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
≤High School Associates
Degree
Bachelor’s
Degree
Graduate
Degree
Probability of College Completion
Highest Level of Parental Education Completion
529 Participants
Non 529 Participants
Texas Tech University, Brigham T. Dorman, August 2016
65
greater. No prospective college student in our sample demonstrates a 100% probability of
college completion.
Figures 2.15 and 2.16 show that high income households experience the greatest
tax benefits from participating in tax-advantaged accounts. Placing 100% of postsecondary
savings in a Roth IRA results in the lowest expected tax liability until
prospective college students reach a 50% probability of completing a college degree. If a
prospective college student has a 50% probability, diversifying between a Roth and taxable
account produces the lowest expected tax liability. Households with prospective college
students who demonstrate between 55% to 60% probability may experience the lowest
expected tax liability by participating 100% in taxable accounts. If the prospective college
student demonstrates a probability of graduating greater than 60%, the household may
maximize college savings by placing college savings primarily in 529 plans. Almost one
in five (n=1,357) of our entire sample are students from households with income greater
than or equal $35,000 and have a probability of completing a college degree greater than
or equal 60%. One in every four households have a child with a probability of graduating
from between 35% and 50%. In this case, the model results in 100% Roth IRA as the
optimal account choice for the lowest expected tax liability. However, households wishing
to participate in 529 plans, with prospective college students who have a probability of
completing a college degree between 35% and 50%, may experience low expected tax
liabilities by diversifying between 529 plans and Roth IRAs, placing the majority of the
college savings in the Roth IRA. Figure seventeen shows that households who experience
increasing income face similar account choices as middle and high income households.
Texas Tech University, Brigham T. Dorman, August 2016
66
Figure 2.14: Scenario 1 – Low Income Household (0% Capital Gains Tax)
Expected Total Tax Liability by Account Choice and Probability of College Completion
Probability
of College
Completion
100%
529
100%
Roth
IRA
100%
Taxable
Account
25/75
529/Roth
50/50
529/Roth
75/25
529/Roth
25/75
529/Taxable
50/50
529/Taxable
75/25
529/Taxable
25/75
Roth/Taxable
50/50
Roth/Taxable
75/25
Roth/Taxable
10%
15%
20%
25%
30%
35%
40%
45%
50%
55%
60%
65%
70%
75%
80%
85%
90%
95%
100%
Primary Assumptions: Household marginal tax rate = 15%, Initial contribution = $5,000, Years Until College = 18, Capital Gains Tax Rate = 0%, Additional Monthly
Contributions = $150
Texas Tech University, Brigham T. Dorman, August 2016
67
Figure 2.15: Scenario 2 – Middle Income Household
Expected Total Tax Liability by Account Choice and Probability of College Completion
Probability
of College
Completion
100%
529
100%
Roth
IRA
100%
Taxable
Account
25/75
529/Roth
50/50
529/Roth
75/25
529/Roth
25/75
529/Taxable
50/50
529/Taxable
75/25
529/Taxable
25/75
Roth/Taxable
50/50
Roth/Taxable
75/25
Roth/Taxable
10%
15%
20%
25%
30%
35%
40%
45%
50%
55%
60%
65%
70%
75%
80%
85%
90%
95%
100%
Primary Assumptions: Household marginal tax rate = 28%, Initial contribution = $7,500, Years Until College = 18, Capital Gains Tax Rate = 15%, Additional
Monthly Contributions = $300
Texas Tech University, Brigham T. Dorman, August 2016
68
Figure 2.16: Scenario 3 – High Income Household
Expected Total Tax Liability by Account Choice and Probability of College Completion
Probability
of College
Completion
100%
529
100%
Roth
IRA
100%
Taxable
Account
25/75
529/Roth
50/50
529/Roth
75/25
529/Roth
25/75
529/Taxable
50/50
529/Taxable
75/25
529/Taxable
25/75
Roth/Taxable
50/50
Roth/Taxable
75/25
Roth/Taxable
10%
15%
20%
25%
30%
35%
40%
45%
50%
55%
60%
65%
70%
75%
80%
85%
90%
95%
100%
Primary Assumptions: Household marginal tax rate = 39.6%, Initial contribution = $10,000, Years Until College = 18, Capital Gains Tax Rate = 20%, Additional
Monthly Contributions = $500
Texas Tech University, Brigham T. Dorman, August 2016
69
Figure 2.17: High Income Growth Household
Expected Total Tax Liability by Account Choice and Probability of College Completion
Probability
of College
Completion
100%
529
100%
Roth
IRA
100%
Taxable
Account
25/75
529/Roth
50/50
529/Roth
75/25
529/Roth
25/75
529/Taxable
50/50
529/Taxable
75/25
529/Taxable
25/75
Roth/Taxable
50/50
Roth/Taxable
75/25
Roth/Taxable
10%
15%
20%
25%
30%
35%
40%
45%
50%
55%
60%
65%
70%
75%
80%
85%
90%
95%
100%
Primary Assumptions: Initial Household marginal tax rate = 25%, Ending Household Marginal Tax Rate = 39.6%, Initial contribution = $5,000, Years Until College = 18, Years of
Additional Contributions = 13, Capital Gains Tax Rate = 20%, Additional Monthly Contributions = $300
Texas Tech University, Brigham T. Dorman, August 2016
70
Tax shock minimization is a function of 529 plan participation and high
probabilities of college completion. To investigate the relationship between 529 plan
participation and college completion probabilities, we conduct an ordinary least squares
regression. Table 2.2 shows a remarkably consistent relationship (R-squared of 0.8057).
Results show that 529 plan participation is positively related and statistically significant at
the 1% level, but explains very little of college completion probabilities. Participating in
a 529 plan only adds 0.0005 (R-squared = 0.8052 without 529 participation) to the fitted
data.
Table 2.2: OLS Regression of 529 Participation on College Completion
Probabilities (n=6,119)
Coefficient S.E. p-value
529 Plan Participant (Non-Participant) 0.008 0.002 0.000 ***
Student GPA 0.069 0.002 0.000 ***
Female (Male) 0.088 0.002 0.000 ***
Cognitive Grade Level Test Score 0.006 0.000 0.000 ***
Student Ethnicity (White)
Asian 0.056 0.004 0.000 ***
American Indian -0.163 0.020 0.000 ***
Black -0.072 0.004 0.000 ***
Hispanic -0.066 0.004 0.000 ***
Multi-Race -0.068 0.004 0.000 ***
Other Race -0.009 0.018 0.581
Constant -0.230 0.030 0.000 ***
Dependent Variable: Probability of College Completion
Reference Category in Parentheses
***, **, * Statistical Significance at the 1%, 5%, and 10% Level, respectively
R-Squared = 0.8057
Texas Tech University, Brigham T. Dorman, August 2016
71
Conclusion
Households participate in tax advantaged accounts primarily for the tax benefits
and the vast majority of public policy around education affordability focuses on college
cost. High income households may experience the greatest benefit of 529 participation
because it is tax-sheltered without any income limitations30, making it a popular “next
sheltering option.” They can still benefit even when taking non-qualified withdrawals.
Many students do not attend or complete college, while parents demonstrate
unrealistic optimism towards cognitive ability. This results in an inability to spend savings
that have been placed in a 529 plan, which results in a tax shock that reduces the ex-ante
efficiency of saving in a 529 plan. We investigate how the risk that a child will fail to
attend college impacts optimal account choice for college saving households.
When college completion probabilities are between 35% and 50%, results suggest
that low income households experience lower expected tax liabilities when they avoid 529
plans and use taxable accounts. Households subject to capital gains taxes experience lower
expected tax liabilities when they diversify between 529 plans and Roth IRAs, with the
majority of the college saving assets being placed in the Roth IRA. However, low expected
tax liabilities may also be experienced by diversifying between a Roth IRA and a taxable
account.
Even higher-income households who benefit more from tax-advantaged savings
than low and middle income households may need to evaluate their decision to participate
in tax-advantaged education saving accounts. Results show that few prospective college
students from high income households have a high enough probability of graduating with

3030 A traditional IRA has an income limitation by which contribution deductibility phases out and Roth IRAs
have an income limitation that phases out participation.
Texas Tech University, Brigham T. Dorman, August 2016
72
a post-secondary degree to warrant saving in a 529 plan. Consistent with DeAngelo et al.
(2011) college completion rates are lower than many parents might expect (less than 50%).
We find that 529 plan participation does very little to enhance the probability of completing
a college degree, suggesting that 529 plan participating households may experience a large
unexpected tax shock if the prospective college student chooses not to attend or
discontinues college attendance. However, it is important to note that if a household fully
utilizes 529 funds within the college years, the benefit of 529 participation still applies.
For example, if a household uses up one hundred percent of their 529 savings within the
first two years of their child’s post-secondary years, it is less likely they will take nonqualified
withdrawals.
Cognitive preparation for college academics explain 80% of college completion
probabilities. Our results suggest that tax shocks are highly possible when college
completion rates are low and that 529 plans may be sub-optimal accounts to invest in for
college, at least until college completion probabilities increase to over 50%.
Implication on Financial Planning
Findings from this study suggest that financial planners consider avoiding
investment in 529 plan when prospective college students have low probabilities of
graduation. Evaluating the probability that the child may or may not complete college
during the data gathering and analysis stage of the financial planning process is for
recommending an appropriate account for college savings. Even though this may be an
unconventional approach, it is important to note that financial planners may need to
consider the consequences of recommending 529 participation in the event that the
investment is not used to fund a qualified educational expense.
Texas Tech University, Brigham T. Dorman, August 2016
73
Future Research
To the best of our knowledge, this is the first study to address college completion
probabilities with account choice, and to address the unexpected event of tax shocks.
Similar to Doyle (1984), we can expand our research on account choice for college
planning by evaluating the breakeven point at which a college saving household may be
able to pay both the regular income tax and tax penalty, even when the child does not
complete or completely ignores college, and have an equal post tax amount in either a 529
or non-529 accumulated account. Essentially, further analysis could allow households to
estimate how early they would need to start saving for college to break even. Additionally,
future research should investigate withdrawal sequencing (D. Blanchett & Kaplan, 2013;
Horan, 2006) from college saving accounts. Another area for further research is to
determine the optimal order of contributions to sheltered accounts since higher income
households can still benefit in after-tax and after-penalty distributions from 529 plans.
Finally, investigating the probability that a household would shift or transfer 529 plan
assets from a child who fails to complete college to another qualified beneficiary would
benefit the existing research in financial planning for college.
Texas Tech University, Brigham T. Dorman, August 2016
74
CHAPTER IV
Appendix: Model of Account Choice for College Saving Households
To demonstrate the complexity of account choice for college saving households,
we use four simplified examples. Horan (2006), Jennings, Horan, & Reichenstein (2010),
and Reichenstein (2006) demonstrate that retirees may benefit from diversifying their
retirement savings among multiple tax-advantaged savings vehicles and withdrawal
sequencing may increase portfolio longevity. We address account choice from a college
savings perspective.
In the first scenario, we assume a young household, at time 0 (t=0 & birth of a
child), has income within the respectable taxable income level of 15% marginal tax. For
2014, that range is between $9,075 and $36,900. We classify this household as a “low
income household.” For simplicity, we assume this household remains in the 15%
marginal tax rate through their child’s college years. It is worthy to note that low income
households, under current tax law, carry no liability for capital gains tax, thus we assume
this same regulation remains for the given scenario. This low income household decides
to save for their newborn child’s college education, with an initial contribution of Ci and
equal additional monthly contributions (Ca) until the child enters college, assuming college
entry at age 19.
In our second example, we assume that at time 0 (t=0 & birth of a child), a
household decides to open an account designated for their newborn’s future college
education with an initial contribution of Ci. We assume that the household is a mid-life
cycle household (i.e. age 40) and remains in the same marginal tax bracket ™ during the
child’s pre-college years and that the child begins college at age 19. Each month the
Texas Tech University, Brigham T. Dorman, August 2016
75
household makes equal additional contributions (Ca) until the child enters college. Once
the child enters college, at t=19, the household discontinues all contributions and converts
the accumulated wealth to cash for preservation and college consumption purposes. For
simplicity purposes, we assume investment returns are net of inflation and fees.
The third household consists of a household that we classify as a “high income
household.” We assume this household starts a college savings account while in the highest
marginal tax rate, and remains in this same marginal tax rate through the child’s college
years. For 2014, the highest marginal income tax rate is 39.6% with taxable income being
greater than $406,750. It is worthy to note that once a household enters this marginal tax
rate, the household capital gains tax rates increases from 15% to 20%. For simplicity
purposes, we ignore the new 3.8% Medicare tax imposed on the lesser of a household’s net
investment income or modified adjusted gross income.
Our fourth and final scenario involves a household with what we will call “income
growth,” meaning the household starts out at a lower marginal tax bracket (i.e. 25%), but
by the time the child enters college, the household income has risen to a much higher
income tax bracket (i.e. 39.6%). We will also assume that the household, due to income
limitations earlier on, cannot and does not start making equal additional monthly
contributions until five years after the initial contribution has been made. The purpose of
delaying the additional contributions is to capture the effect of an increased marginal tax
rate on the basis apart from the initial contribution.
For a given scenario we calculate a households after tax expected wealth (or
expected total tax liability) as
() = + (1 − ), where
Texas Tech University, Brigham T. Dorman, August 2016
76
the after tax expected wealth of the household college savings portfolio (E(Wp)) is a
function of the after tax future value of the college savings portfolio when the child
completes college ( = − ), the after tax future value of the college savings
portfolio when the child does not completes college ( = − ), the probability of
college completion (Pc) and 1-Pc. Total taxes due at withdrawal (tw) is the sum of marginal
tax rates ™, capital gains tax rates (tcg), and any tax penalty (tp), depending on the account
choice of the household.
Taxable accounts provide households with the most flexibility, since assets can be
taken from the account at any time without penalty, and at the time of withdrawal,
households are only required to pay the capital gains rate (tcg). But, there is no tax break
for using the funds for qualified education purposes. Thus the after tax accumulated wealth
of a taxable account is
= [ [
(1 + )

] (1 + ) +
(1 + )

] − [( − )()]
and we ignore annual dividend income for simplicity purposes, but understand the
importance of it. The future value of the college savings portfolio (FVp) is simply the sum
of the future value of the initial contribution (
),

(1 + )

,
and the future value of the additional contributions (),
[
(1+)

] (1 + ), where
we recognize that the basis of the portfolio as
+ (),
with the assumption that additional contributions are equal payments (i.e. $250 monthly).
Texas Tech University, Brigham T. Dorman, August 2016
77
Roth IRA accounts may be the second easiest account type to understand, since
withdrawals are only taxed at the household’s marginal tax rate. Note, however, that the
marginal tax rate is only applied to the growth portion of the account because taxes were
paid at the time of the contribution and the assets are used for qualified education purposes.
Essentially, the marginal tax rate is simply a higher “capital gains” tax rate when assets are
used for post-secondary education. Assets used for qualified education purposes avoid
paying the 10% tax penalty. Thus the after tax accumulated wealth from using a Roth IRA,
assuming the child attends and completes college is precisely similar to that of the taxable
account with the only difference being the tax rate upon withdrawal:
ℎ = [ [
(1+)

] (1 + ) + (1 + )

] − [( − )()].
In the case of a Roth IRA dividend income is of no concern because it is tax deferred, or
tax free if withdrawn after existing for five years, and the account owner has reached 59.5.
The 529 plan may be the most complicated to understand due to the fact that if the
assets are not used for qualified college expenses, they are taxed at both the marginal tax
rate with an additional 10% tax penalty. Thus, the after tax accumulated wealth for using
a 529, assuming the child goes to and completes college is:
529 = [ [
(1 + )

] (1 + ) + (1 + )

],
But, assuming 529 plan savings have been accumulated, if the child decides not to enter
college or to discontinue their college education, the after tax accumulated wealth for using
a 529 plan is:
529 = [ [
(1+)

] (1 + ) + (1 + )

] − [(( − )()) + (( − )())].
Texas Tech University, Brigham T. Dorman, August 2016
78
What is the optimal account choice using these models? We assume that each
household, from the four scenarios above, can choose to invest in a 529 education savings
plan, a Roth IRA, a taxable account, or to diversify their college savings across these three
choices. We assume household marginal tax rates ™ and tax liabilities for contributions
are paid from a separate account, but taxes at the time of withdrawals (tw) carry different
rules across account type. Figures fourteen through seventeen show the results of
determining the optimal account choice as a function of expected tax liabilities and
potential tax shocks. For simplicity purposes we assume all scenarios experience a 5% real
rate of return.
Transferring 529 Assets
Up to this point we ignore the important feature that households participating in
529 plans can transfer the account to an alternative beneficiary if the primary beneficiary
decides not to use the assets. The impact of this option remains open to future research.
The choice of transferring assets effects both the departing beneficiary as well as the new
beneficiary with respect to a decision to attend college. With the assets no longer at the
disposal of the primary beneficiary, they may lose their interest at returning or going to
college. With the assets now at the disposal of the new beneficiary, it may come as a shock
they have essentially received a financial windfall for college, but may not have been
preparing for or have any plans to attend. This feature is a common estate planning
strategy31 in order to remove assets from one’s taxable estate.

31 There are two key measures when discussing 529 plans for estate planning purposes. First, the impact of
529 plans on grandparent estate plans, and second, the structure of the 529 plan has important financial aid
eligibility implications. For grandparents to maximize their financial support towards a grandchild’s
postsecondary education, and reduce their estate tax liability, it may be optimal to contribute the five-year
equivalent lump-sum. By doing this, and as early as possible, they can maximize their benefit of tax-free
compound interest. However, it is important to note that if the grandparent passes away during the five years,
529 plan assets are prorated and returned to the taxable estate. Additionally, it may be optimal to have 529
Texas Tech University, Brigham T. Dorman, August 2016
79
Finally, households that fail to recognize college completion probabilities when
evaluating account choice, and elects to invest in a 529 plan by default run the risk of an
unexpected 529 tax shock. Given the low college completion probabilities of prospective
college students, a household with a child that believes itself to be more likely to complete
college will realize a higher tax benefit through 529 plan participation, and is more likely
to prefer this account type for their college savings.

assets from grandparents be used for the final two years of college because of the new prior-prior year income
reporting rule. Households seeking to maximize federal need-based aid need to pay close attention to
ownership structure. For example, if a 529 plan is placed in the grandparent’s name – to reduce their estate
tax liability – and are withdrawn to pay for a child’s college expenses, the 529 withdrawals count against the
child’s FAFSA application as income to the child.
Texas Tech University, Brigham T. Dorman, August 2016
80
References
Amromin, G., & Smith, P. (2003). What Explains Early Withdrawals from Retirement
Accounts? Evidence from a Panel of Taxpayers. National Tax Journal, 56(3), 595–
612.
Argento, R., Bryant, V. L., & Sabelhaus, J. (2015). Early Withdrawals from Retirement
Accounts During the Great Recession. Contemporary Economic Policy, 33(1), 1–16.
Attewell, P., Heil, S., & Reisel, L. (2011). Competing Explanations of Undergraduate
Noncompletion. American Educational Research Journal, 48(3), 536–559.
Blanchett, D., & Kaplan, P. (2013). Alpha , Beta , and Now … Gamma. The Journal of
Retirement, 1(2), 29–45.
Blanchett, D. M. (2011). Target-Date Funds : A Temporary Solution to a Permanent
Problem. Journal of Financial Planning, 24(September), 60–67.
Bogan, E. C., & Bogan, T. R. (1982). Individual Retirement Accounts and Preretirement
Savings Goals. Financial Analysts Journal, 38(November-December), 45–47.
Conley, D. (2001). Capital for College: Parental Assets and Postsecondary Schooling.
Sociology of Education, 74(1), 59–72.
CSPN. (2015). College Savings Plans Network 529 Report: An Exclusive Mid-Year
Review of 529 Plan Activity.
Dammon, R. M., Spatt, C. S., & Zhang, H. H. (2004). Optimal Asset Location and
Allocation with Taxable and Tax-Deferred Investing. Journal of Finance, 59(3),
999–1037.
DeAngelo, L., Pryor, J. H., & Franke, R. (2011). Completing College : Assessing
Graduation Rates at Four-Year Institutions. Higher Education Research Institute,
Graduate School of Education & Information Studies, University of California, Los
Angeles.
DeGennaro, R. (2004). Asset Allocation and Section 529 Plans. International Journal of
Business, 9(2).
Doyle, R. J. (1984). IRAs and The Capital Gains Tax Effect. Financial Analysts Journal,
40(3), 60–66.
Duncheon, J. C. (2015). The Problem of College Readiness. The Problem of College
Readiness (Vol. 1).
Dunning, D., Heath, C., & Suls, J. M. (2004). Flawed Self-Assessment Implications for
Health, Education, and the Workplace. Psychological Science in the Public Interest,
Supplement, 5(3), 69–106.
Dynarski, S. M. (2004a). Tax Policy and Education Policy: Collision or Coordination? A
Case Study of the 529 and Coverdell Saving Incentives (No. w10357). National
Bureau of Economic Research.
Texas Tech University, Brigham T. Dorman, August 2016
81
Dynarski, S. M. (2004b). Who Benefits From the Education Saving Incentives? Income,
Educational Expectations, and the Value of the 529 and Coverdell. National Tax
Journal, 57(2), 359–383.
Dynarski, S., & Scott-Clayton, J. (2016). Tax Benefits for College Attendance (No.
w22127) (Vol. 53).
Gokhale, J., & Kotlikoff, L. J. (2003). Who Gets Paid to Save? Tax Policy and the
Economy, 17(MIT Press), 111–139.
Greene, J. P., & Winters, M. A. (2005). Public High School Graduation and CollegeReadiness
Rates: 1991-2002. Education Working Paper No. 8. Center for Civic
Innovation.
Haveman, R., & Wilson, K. (2007). Economic Inequality in College Access,
Matriculation, and Graduation. Economic Inequality and Higher Education Access,
Persistence, and Success, 20–59.
Horan, S. M. (2003). Choosing Between Tax-Advantaged Savings Accounts : A
Reconciliation of Standardized Pre- tax and After-tax Frameworks. Financial
Services Review, 12(4), 339–357.
Horan, S. M. (2006). Withdrawal Location with Progressive Tax Rates. Financial
Analysts Journal, 62(6), 77–87.
Hrung, W. B. (2007). Determinants of the Choice between Roth and Deductible IRAs.
The Journal of the American Taxation Association, 29(1), 27–42.
Jennings, W. W., Horan, S. M., & Reichenstein, W. (2010). Private Wealth Management:
A Review. Research Foundation Literature Reviews, 5(1), 1–68.
Klein, W. M. P., & Weinstein, N. D. (1997). Social Comparison and Unrealistic
Optimism about Personal Risk. In Health, coping, and well-being: Perspectives
from social comparison theory (pp. 25–61).
Magolda, M. B. B., & Astin, A. W. (1993). What Matters in College: Four Critical Years
Revisited. Educational Researcher, 22, 32.
Mano, R. M., & Burr, T. (1984). IRAs versus Nonsheltered Alternatives for Retirement
Savings Goals. Financial Analysts Journal, 40(May-June), 67–75.
Maxwell, N. L., & Lopus, J. S. (1994). The Lake Wobegon Effect in Student SelfReported
Data. The American Economic Review, 84(2), 201–205.
Miller, S. A. (1986). Parents’ Beliefs About Their Children’s Cognitive Abilities.
Developmental Psychology, 22(2), 276–284.
Modigliani, F., & Brumberg, R. (1954). Utility Analysis and the Consumption Function:
An Interpretation of Cross-Section Data. In Franco Modigliani.
Texas Tech University, Brigham T. Dorman, August 2016
82
Moore, D. A. (2007). Not So Above Average After All: When People Believe They Are
Worse Than Average and Its Implications For Theories of Bias in Social
Comparison. Organizational Behavior and Human Decision Processes, 102(1), 42–
58.
Pascarella, E. T., & Terenzini, P. T. (2005). How College Affects Students : A Third
Decade of Research. The JosseyBass higher and adult education series (Vol. 2).
Poterba, J. M., Shoven, J. B., & Sialm, C. (2000). Asset Location for Retirement Savers
(No. w7991). National Bureau of Economic Research.
Reichenstein, W. (2006). Withdrawal Strategies to Make Your Nest Egg Last Longer.
AAII Journal, 28(10), 5–11.
Sallie Mae. (2015). How America Saves for College: Sallie Mae’s National Study of
Parents with Children Under Age 18.
Sandhya, V. (2011). Agency Problems in Target-Date Funds. AFA 2012 Chicago
Meetings Paper.
Shoven, J. B., & Sialm, C. (2004). Asset location in tax-deferred and conventional
savings accounts. Journal of Public Economics, 88, 23–38.
Smith, H. (2010). Who Benefits Most from Section 529 Plans and Coverdell ESA
Accounts ? In The 57th Annual Conference of the American (Vol. 57, p. 28).
Svenson, O. (1981). Are We All Less Risky and More Skillful Than Our Fellow Drivers?
Acta Psychologica, 47(2), 143–148.
Terry, A., & Goolsby, W. C. (2003). Section 529 Plans as Retirement Accounts.
Financial Services Review, 12(4), 309.
Turner, N. (2011). Why Don’t Taxpayers Maximize their Tax-Based Student Aid?
Salience and Inertia in Program Selection. The B.E. Journal of Economic Analysis &
Policy, 11(1).
U.S. Department of the Treasury. IRS Publication 970: Tax Benefits for Education, Cat.
No. 25221V (2014).
Weidman, J. C., Pascarella, E. T., & Terenzini, P. T. (1992). How College Affects
Students: Findings and Insights From Twenty Years of Research. Contemporary
Sociology (Vol. 21).
Weinstein, N. D. (1980). Unrealistic Optimisim About Future Life Events. Journal of
Personality and Social Psychology, 39(5), 806–820.
Zhou, J. (2009). The Asset Location Puzzle: Taxes Matter. Journal of Economic
Dynamics and Control, 33(4), 955–969.
Texas Tech University, Brigham T. Dorman, August 2016
83
CHAPTER V
ASSET REPOSITIONING AND COLLEGE FINANCIAL AID MAXIMIZATION
Pro-saving households are penalized the most when it comes to college financial
aid eligibility. However, asset repositioning may reduce this burden by increasing the
eligibility for federal need-based aid. Asset re-positioning refers to the process of relocating
assets when economic shocks or short term goals approach, such as covering
postsecondary education expenses, a temporary reduction in income, or even the retirement
transition.
The Higher Education Act of 1965 introduced financial aid to prospective college
students, opening the doors of a postsecondary education to many more households. Its
1992 amendment removed home equity from its formula, increasing financial aid awards
for future college attendees of wealthier households.
During the last decade, the challenge of college affordability has increased, putting
pressure on household ability to maintain current consumption. This challenge is supposed
whether households cover postsecondary expenses with parent or student income, taxable
savings, tax-advantaged savings, non-traditional assets such as home equity, scholarships,
or student loans.
At the height, the 2014 – 2015 academic year, households spent 16% more on
college expenses, compared to the 2013 – 2014 academic year. This is the second largest
increase in average postsecondary education spending since the onset of the financial crisis
(2007 – 2008), with the 2009 – 2010 academic year reporting the largest increase in average
college spending, 24%. The surge in average college spending was triggered by higher
Texas Tech University, Brigham T. Dorman, August 2016
84
income households spending approximately $12,000 more than low and middle income
households (Sallie Mae, 2015).
The increase in average college spending is also manifest in the decline of state
higher education funding. Figure 3.1 shows that 94% of the states reduced their higher
education budget. Average state funding per student was cut by about 20% between 2008
and 2015. For the majority of the states, these figures indicate that state funding on higher
education is lower than that of pre-financial crisis, and that these cuts may be a direct result
of the decline in state tax revenues during the recession.
Texas Tech University, Brigham T. Dorman, August 2016
85
Figure 3.1: Percentage Change in State Funding per Student, Inflation Adjusted,
2008-2015
Source: Center on Budget & Policy Priorities
35.5%
5.4%
3.9%
-2.0%
-4.0%
-6.8%
-7.0%
-8.3%
-9.5%
-10.2%
-11.1%
-13.4%
-14.8%
-15.4%
-16.3%
-16.5%
-16.7%
-17.2%
-17.7%
-20.2%
-20.7%
-21.0%
-21.1%
-22.0%
-22.1%
-22.3%
-22.4%
-22.6%
-22.8%
-23.0%
-23.1%
-23.2%
-23.4%
-23.5%
-24.5%
-25.0%
-25.2%
-26.7%
-26.8%
-27.6%
-28.4%
-30.8%
-32.1%
-32.2%
-33.5%
-35.8%
-36.6%
-37.9%
-42.0%
-47.0%
-60.0% -40.0% -20.0% 0.0% 20.0% 40.0%
NORTH DAKOTA
WYOMING
ALASKA
MONTANA
ILLINOIS
MARYLAND
NEW YORK
NEBRASKA
INDIANA
ARKANSAS
CALIFORNIA
MAINE
SOUTH DAKOTA
VERMONT
UTAH
WISCONSIN
CONNECTICUT
MASSACHUSETTS
COLORADO
HAWAII
KANSAS
MINNESOTA
RHODE ISLAND
NEW JERSEY
TENNESSEE
IOWA
GEORGIA
OHIO
TEXAS
WEST VIRGINIA
MICHIGAN
MISSISSIPPI
NORTH CAROLINA
OKLAHOMA
VIRGINIA
FLORIDA
MISSOURI
DELAWARE
NEW HAMPSHIRE
KENTUCKY
WASHINGTON
NEVADA
IDAHO
NEW MEXICO
OREGON
PENNSYLVANIA
ALABAMA
SOUTH CAROLINA
LOUISIANA
ARIZONA
Texas Tech University, Brigham T. Dorman, August 2016
86
Households observe a college education as a good investment if the present value
of the expected future income is greater than the cost of attendance. The reduction in state
funding, and the increase in household financial responsibility may be cause for an increase
in demand for minimizing household expected family contribution (EFC). Wealthier
households will be more likely to demand asset re-positioning because they have a higher
propensity to save.
Investigating need-based financial aid and the re-positioning of non-qualified assets
to home equity undertakes the notion that homeowners may be able to continue
consumption smoothing more easily during the college years when current income may be
constrained. In other words, households that carry the responsibility to save for retirement
and cover college expenses may have a larger “savings gap,” thus a greater demand to
reduce this gap (Stiglitz, Tyson, Orszag, & Orszag, 2000).
Preparing for and managing college expenses includes planning for federal
financial aid (including student loans) in addition to saving and investing. This paper is
similar in nature to studies on asset location (Poterba, Shoven, & Sialm, 2000; Shoven,
1999; Trout, 2013; Zhou, 2009). Specifically, we investigate how re-locating non-qualified
assets to home equity can minimize EFC. First, we build a hypothetical scenario a financial
planner may encounter when working with a pro-saving client with a college funding goal.
Second, we construct the FAFSA model and test the change in the hypothetical households
EFC score as non-qualified assets are converted into home equity. Third, we employ a
logistic regression to evaluate the relationship between HOWS and DUAL on the demand
for EFC minimization. Next, we discuss the key topics needing consideration when
analyzing the complexity of the asset re-positioning strategy, and identify situations in
Texas Tech University, Brigham T. Dorman, August 2016
87
which this strategy can be useful. Findings indicate that re-positioning non-qualified assets
to home equity may be an option for households with large amounts of non-qualified assets,
assuming EFC is less than the cost of attendance (COA). We also find evidence that
HOWS have a greater probability of demanding EFC reduction, compared to their
counterparts. Results suggest that re-positioning non-qualified savings to home equity can
reduce the savings gap.
Background and Literature Review
President Johnson believed, much like many Americans do today, that a strong
postsecondary education system is a gateway to opportunity and economic growth.
Providing economic resources to low and middle income families, as well as postsecondary
institutions, combats poverty. On November 8, 1965 financial assistance for a college
education was signed into law through the Higher Education Act of 1965 (HEA). HEA
provides means by which grants, loans, tax credits, and scholarships are extended to
prospective college students, and undergraduate enrollment grows as a result. In its
infancy, HEA recommissioned every four years until 1980 after which the re-evaluation
periods were extended to every six years.
The 1992 amendment of the bill (HEA ’92) re-classified home equity as a qualified
asset, thus being removed from the financial aid calculation. This legislative change
increased both aid eligibility, primarily for wealthier households, and college attendance
(Dynarski, 2002). Reducing financial barriers to post-secondary education improves
enrollment and completion rates (Dynarski & Scott-Clayton, 2013). HEA ’92 also
redirected the focus of public policy on student aid from grants to student loans (Dynarski,
2002; Hannah, 1996).
Texas Tech University, Brigham T. Dorman, August 2016
88
States are challenged with budget allocation puzzles as funding demands increase
with rising tuition and declining state appropriation. Delaney & Doyle (2014) demonstrate
that higher education funds are the most volatile categories in a state’s budget, having
greater marginal increases during economic expansions, while experiencing greater
marginal decreases through economic contractions. Peaking in 1979, by 2000 higher
education funding fell by approximately 25% ($23.4 billion). As a result, the
socioeconomic gap widened due to a growing supply of an under-skilled labor force.
Rising tuition made a college degree less affordable. This means that increases of
household income has struggled to keep up with the rate of college tuition. As a result of
declining state funding and rising operating costs, states seek alternative funding sources
such as grants, private donations, and household savings and income (Archibald &
Feldman, 2008; Hearn, 2006). However, research supports the notion that public financial
support is significantly important for institutional success and social improvement
(Mumper, 2003; Ryan, 2004; Titus, 2006).
Private donations act as a substitute to state funding. Charitable deductions are an
attractive tax incentive for the wealthy and one of the top ten most expensive tax breaks
offered by the federal government, costing $44.3 billion in 2015 and projected to cost more
than $600 billion by 2024 (Office of Management and Budget, 2015). Private donations
are triggered by recognition, such as name displays or supporting remarkable research,
implying that lower level institutions may encounter stiff competition when it comes to
institutional success and graduate quality. Lower level institutions are the lifeline of higher
education for lower socioeconomic students, thus an education quality gap increases as
Texas Tech University, Brigham T. Dorman, August 2016
89
private donations are catered to more prestigious institutions (Cheslock & Gianneschi,
2008).
State financial aid has morphed from grants to student loans based more on merit
(merit-based) compared to financial need (need-based) (Toutkoushian & Shafiq, 2010). In
the 1980s, state support came primarily from need-based funding, but by 2000, a significant
amount of need-based aid had shifted to merit-based financial support32 (Mortenson, 2000).
Merit based aid sprang into existence in the early 1990s. It awards students based on their
academic achievements rather than the need for financial support. Studies debate the
impact of need-based and merit-based aid on institutional outcomes since economically
disadvantaged households have a greater demand for need-based aid (De La Rosa, 2006).
Titus (2009) finds a positive relationship between state need-based aid and undergraduate
degree completion, but no relationship between merit-based aid and the generation of
undergraduate degrees. However, Dynarski & Scott-Clayton (2013) point out that meritbased
aid improves postsecondary attendance persistence more than need-based aid.
The demand for need-based financial aid is increasing. Figure 3.2 shows that
household income fails to keep up with rising tuition, and income is one of the primary
sources of funding college expenditures. Qualifying for need-based financial assistance
requires students to apply through the Free Application for Federal Student Aid (FAFSA)
provided by the U.S. Department of Education. The FAFSA is a mechanism in reducing
the price barrier to higher education. This need-based aid application provides students

32 Need-based aid is based on financial need, while merit-based aid is based on academic accomplishment.
Literature debates the relationship of merit-based aid on need-based aid. Doyle (2010) examines this
relationship through the punctuated equilibrium theory (Baumgartner & Jones, 1993) and finds that the rise
in merit-based aid has had little impact on the decline of need-based aid.
Texas Tech University, Brigham T. Dorman, August 2016
90
with access to federal, state, and institutional financial resources to help cover
postsecondary expenses.33
Figure 3.2: Tuition vs. Income
Source: National Center for Education Statistics and U.S. Census Bureau
Calculating financial aid eligibility is complex since it is a function of parental,
student, and third party financial support. In an effort to assist low and middle income
households in affording a college education, financial aid eligibility is negatively correlated
with parental and student financial resources, as well as third party support – as household
assets or third party scholarships increase, need-based eligibility declines.
Grants, scholarships, and current parent income are the largest sources of nonborrowed
funds to pay for college. In 2015, 47% of households relied on current income

33 Expenses that qualify are classified as tuition and fees, room and board, books and supplies,
transportation, and other related expenses (i.e. electronic devices).
Texas Tech University, Brigham T. Dorman, August 2016
91
to help cover postsecondary expenses. However, the use of current income for college
expenditures has declined by 11% since 2011. This decline may be a result of the inability
of household income to keep up with the rising cost of tuition. Grant and scholarship use
remains stable since 2011, with approximately 45% of households.
Financial aid is the largest source of borrowed funds to pay for college. The purpose
of financial aid is to increase college enrollment and completion rates, and evidence shows
that college attendance rises with financial assistance (Dynarski, 2003). In 2015, 38% of
borrowed college funds were from federal government loans, up from 33% in 2014. The
awarded financial aid is determined by the difference between the COA and the EFC.
Thus, minimizing EFC reduces the current financial burden on the household. According
to the National Center for Education Statistics and the 2011-2012 National Postsecondary
Student Aid Study (NPSAS:12), in 2012 70% of households with undergraduate students
applied for federal aid, a substantial increase from 58.5% in 200834. The only households
that experienced a decline in FAFSA applications, thus a reduction in federal borrowing,
were households with annual income greater than $105,000. Sallie Mae (2015) reports that
households with a plan to pay for college borrow 40% less than non-planning households,
but only four in ten households with prospective college students are planning for
postsecondary education expenses.
Although financial aid may assist with college expenses, it may discourage saving
by imposing a “financial aid tax” (Dick & Edlin, 1997; Edlin, 1993). A positive EFC is
commonly recognized as a “financial aid tax” and is a function of household adjusted gross

34 Authors calculations using data from the High School Longitudinal Study of 2009 (HSLS:09), with a
sample of 2,396 households, shows that 75% of households filed their FAFSA application in 2011. Thus,
the 2012 results of 70% from the NPSAS:12 demonstrates a 5% decline in FAFSA applications between
2011 and 2012..
Texas Tech University, Brigham T. Dorman, August 2016
92
income, non-qualified assets35, marginal tax rates, parent age, household size (including
the number of dependents enrolled in postsecondary institutions), state of residency, and
“other financial information.” As ones EFC increases, either primarily due to a rise in
adjusted gross income or the value of non-qualified assets, the need-based aid award
declines, signaling that increasing ones income or savings “taxes” the household.
Home equity and retirement accounts are the smallest sources of borrowed funds
to pay for college. Figure 3.3 shows that few households use home equity towards
postsecondary expenditures, and its use has declined since 2008. However, 75% of
households that either own their home outright or are making mortgage payments filled out
the 2011 FAFSA36. Home equity makes up a significant part of household wealth, and in
terms of market value, is the difference between the market value of the home and the
remaining balance on the mortgage. More common uses of home equity are for long term
care, retirement income, bequests, emergencies, or future health care expenses (Davidoff,
2010; Lusardi & Mitchell, 2007; Skinner, 2007).

35 Non-qualified assets consist of all assets, with the exception of retirement assets (i.e. 401k’s and IRAs),
life insurance, personal property, non-qualified annuities, and home equity. It is important to note that college
specific assets (i.e. 529 plans) are included.
36 Authors calculations using data from the High School Longitudinal Study of 2009 (HSLS:09), produced
by the National Center of Education Statistics. From the nationally represented study, 2,396 households are
investigated. Due to data limitations we are unable to determine if the sample households either hold equity
in their home or are underwater.
Texas Tech University, Brigham T. Dorman, August 2016
93
Figure 3.3: Household Home Equity Use for College Expenditures
Source: (Sallie Mae, 2012, 2015b)
Studies look at how households manage mortgage debt and decisions to re-locate
economic resources to home equity. Most notably, Campbell (2006) notes that households
make irrational mortgage decisions, while other research shows that some households with
children save less when faced with the financial aid tax (Wolpaw Reyes, 2007; Long 2004).
Amromin, Huang, & Sialm (2007) examine the benefit (consequence) of prepaying a
mortgage as opposed to maximizing tax deferred account contributions. Using data from
the Survey of Consumer Finances they find that prepaying a mortgage may cost between
11% and 17% of tax deferred growth. They also find that prepaying a mortgage is because
the household is averse to financial market participation. However, Adelman, Cross, &
Shrider (2010) show, also using the Federal Reserve’s Survey of Consumer Finances data,
that mortgage prepayments are related to liquidity risk and the propensity to save.
McCollum, Lee, & Pace (2015) discuss how mortgage prepaying has grown in popularity
in America over the last decade and show that 30% of mortgage holders have made balance
prepayments. They also show that households with higher credit scores are more likely to
make balance prepayments. Finally, Barbiarz & Yilmarez (2009) demonstrate that college
0%
1%
2%
3%
4%
5%
$0
$3,000
$6,000
$9,000
$12,000
$15,000
2008 2009 2010 2011 2012 2013 2014 2015
Percentage of Households Using
Home Equity Towards College
Cost
Average Amount of Home Equity
Contributed Toward College cost
Average Home Equity Amount % of Households
Texas Tech University, Brigham T. Dorman, August 2016
94
saving households may reduce their EFC, and improve need-based aid, by maximizing taxdeferred
contributions and paying down the mortgage with non-qualified assets.
This paper contributes to the literature of mortgage prepayments and asset relocation
by discussing the potential net benefit from re-positioning non-qualified assets
into home equity for purposes of financial aid maximization. The idea of re-positioning
non-qualified assets into home equity is attractive for multiple reasons. College planning
households are sandwiched with the demand to save for retirement and college. However,
saving for retirement, or other non-educational goals, in non-qualified asset locations
“taxes” the household with higher EFC scores, thus reducing the amount of need-based aid
the prospective college student may qualify for. Submitting a FAFSA application signals
that the household demands minimizing their EFC and maximizing need-based aid. If
households who have the potential or ability to re-position their non-qualified assets are
not doing so, they may be failing to optimally minimize their EFC. This paper argues that
financial planners may be able to assist households in optimally maximizing need-based
aid by presenting the option to re-position non-qualified assets into home equity. Findings
suggest that households may be able to reduce the burden of college affordability by repositioning
non-qualified savings into home equity, but the transaction is not for every
household, and may pose challenges related to asset allocation and taxes.
Data and Method
To investigate the topic of re-locating non-qualified assets to home equity for the
purpose of financial aid maximization, we break our analysis into two parts: 1) building a
model to determine the net benefit from repositioning non-qualified assets to home equity
and 2) investigating consumer demand for EFC minimization.
Texas Tech University, Brigham T. Dorman, August 2016
95
The FAFSA application is the gateway to federal need-based aid. First, we build a
hypothetical scenario a financial planner may encounter when working with a pro-saving
client with a college funding goal. Second, we construct the FAFSA model and test the
change in the hypothetical households EFC score as non-qualified assets are converted into
home equity. Third, we employ a logistic regression to evaluate the relationship between
HOWS on the demand for EFC minimization. Next, we discuss the capital gain and asset
allocation implications when analyzing the complexity of the asset re-positioning strategy.
We use data from the High School Longitudinal Study of 2009 (HSLS) dataset
provided by the National Center for Education Statistics (NCES), a nationally
representative longitudinal study representing over 23,000 high school freshman, their
parents, school administrators, and counselors from 944 high schools. It contains five
waves of data collection beginning in 2009 (base year) and focuses on understanding the
path high school students take into college and the work force. The HSLS fits well with
this study for two main reasons. First, to the best of our knowledge, this dataset has not
been used by prior research in the context of investigating FAFSA usage, thus providing
empirical insight from a different lens. Second, the HSLS provides further light and
knowledge with respect to why home equity building households did or did not file for
federal aid. In addition, the use of this dataset provides data relating to household
demographics, college enrollment, student high school academic performance, and 529
plan participation, including the time horizon of plan participation. Due to the scarcity of
datasets with these variables, this dataset suffices the focus of this study.
For the regression analysis, we are attentive to home equity building households
with non-qualified savings, which have prospective college students. We discard from the
Texas Tech University, Brigham T. Dorman, August 2016
96
sample those households that failed to provide information regarding their plans or actual
participation of filing a FAFSA application. Total sample consists of 4,761 households
that indicate they completed the FAFSA application for the 2013-2014 postsecondary
academic year. Table 3.1 presents the descriptive statistics of the sample. We differentiate
between those households that are homeowners, and have non-qualified assets (HOWS),
and non-HOWS. Of the total sample, 2,209 (46%) respondents qualify as HOWS, with the
remaining 54% representing the non-HOWS. It is important that we divide the sample as
such because in order for a household to demand asset re-positioning, they must have nonqualified
assets, that have the potential to be re-positioned, and own a home.
Texas Tech University, Brigham T. Dorman, August 2016
97
Table 3.1: Descriptive Statistics of Households that Completed the FAFSA (N = 4,761)
Variable Non-HOWS (N =
2,552) HOWS (N = 2,209) Total (N = 4,761)
Household Income
< $15,000 68 (72%) 27 (28%) 95 (2%) Between $15,000 & $35,000 233 (82%) 51 (18%) 284 (6%) Between $35,000 & $55,000 323 (69%) 142 (31%) 465 (10%) Between $55,000 & $75,000 428 (64%) 239 (36%) 667 (14%) Between $75,000 & $95,000 391 (58%) 280 (42%) 671 (14%) Between $95,000 & $115,000 314 (50%) 313 (50%) 627 (13%) Between $115,000 & $135,000 236 (48%) 255 (52%) 491 (10%) Between $135,000 & $155,000 180 (44%) 225 (56%) 405 (9%) Between $155,000 & $175,000 86 (41%) 123 (59%) 209 (4%) Between $175,000 & $195,000 51 (36%) 90 (64%) 141 (3%) Between $195,000 & $215,000 67 (37%) 116 (63%) 183 (4%) Between $215,000 & $235,000 10 (16%) 54 (84%) 64 (1%) > $235,000 165 (36%) 294 (64%) 459 (10%)
Student Gender
Male 1249 (53%) 1096 (47%) 2345 (49%)
Female 1303 (54%) 1113 (46%) 2416 (51%)
Student Race
White 1621 (51%) 1581 (49%) 3202 (67%)
Black 181 (60%) 119 (40%) 300 (6%)
Asian 207 (54%) 178 (46%) 385 (8%)
Hispanic 313 (67%) 151 (33%) 464 (10%)
American Indian 6 (40%) 9 (60%) 15 (0%)
Multi Race 215 (57%) 162 (43%) 377 (8%)
Other Race 9 (50%) 9 (50%) 18 (0%)
Parent Age
< 35 189 (79%) 51 (21%) 240 (5%) 35-40 463 (66%) 240 (34%) 703 (15%) 40-45 741 (54%) 631 (46%) 1372 (29%) 45-50 688 (47%) 786 (53%) 1474 (31%) 50-55 339 (47%) 389 (53%) 728 (15%) > 50 132 (54%) 112 (46%) 244 (5%)
Parent Education
Less than high school 79 (81%) 19 (19%) 98 (2%)
High school diploma 863 (67%) 416 (33%) 1279 (27%)
Associate’s degree 440 (61%) 281 (39%) 721 (15%)
Bachelor’s degree 725 (45%) 893 (55%) 1618 (34%)
Master’s degree 319 (43%) 428 (57%) 747 (16%)
Ph.D./M.D/Law/other 126 (42%) 172 (58%) 298 (6%)
Texas Tech University, Brigham T. Dorman, August 2016
98
Hypothetical College Funding Household
Financial planners that provide education planning services to their clients, and
assist in evaluating optimal need-based aid, should understand the parameters that make
up the EFC score. EFC score management and minimization is where the financial planner
can add value to a client seeking to optimize need-based aid since the COA is only changed
if the future college enrollee were to attend a different postsecondary institution. Even an
institutional change provides minimal flexibility to maximize need-based aid.
Since EFC scores are impacted by both parent and child characteristics, we divide
the assumptions in our hypothetical household between each. Table 3.2 shows the
foundational assumptions that are employed in our hypothetical household. We assume
the pro-saving household consists of a married couple, living in a randomly selected state:
Virginia, with a single dependent child looking to attend college full time – with “full time”
representing a five year graduation time horizon37 and a 100% probability of completing
their college degree.

37 The FAFSA consists of three different formulas depending on the status of the dependent. Formula A is
used when the child is a dependent student, formula B is used if the child is an independent student without
dependents, and formula C is used if the child is an independent student with dependents. This study only
employs formula A. Employing the other formulas is an opportunity for future research. The basic
assumptions are purposely built so that the household does not qualify for the simplified EFC formula, and
the state of residency was randomly selected.
Texas Tech University, Brigham T. Dorman, August 2016
99
Table 3.2: Foundational Assumptions of Hypothetical Household
Assumptions Dual Income
Parental Assumptions
Marital Status Married
Number of prospective college children 1
Total Household Size 3
Number of working parents 2
Age of Oldest Parent 45
Age of Youngest Parent 45
Household State of Residency Virginia
Parental Income
Father $50,000
Mother $20,000
Total Household Income* $70,000
Marginal Tax Rate* 15%
Total Additional Financial Information $0
Assets in Parents Name
Total Retirement Account $615,000
Taxable Accounts
Savings* $10,000
Brokerage* $50,000
Home Equity Prior to Repositioning $10,000
529 Plan, ESA, etc. $20,000
Life Insurance $100,000
Debt (other than mortgage) $0
Net Worth of Business/Farm $0
Mortgage Balance* $300,000
Interest Rate* 3.5%
Loan Term 360 months
Prospective College Student Assumptions
Gross Income $0
Total Untaxed Income $0
Total AGI $0
Marginal Tax Rate 0%
Total Additional Financial Information 0%
Assets in Childs Name
Total Retirement Account $0
Taxable Accounts
Savings $0
Brokerage $0
Home Equity $0
529 Plan, ESA, etc. $0
Life Insurance $0
Cost of College
Public Four-Year In-State $19,548
Public Four-Year Out-of-State $34,031
Years of College Attendance 5
Probability of Graduating 100%
*We change these variables in determining the net benefit of repositioning non-qualified
assets to home equity. Therefore, these are not fixed assumptions.
Texas Tech University, Brigham T. Dorman, August 2016
100
Determining the Net Benefit
To determine the net benefit of repositioning non-qualified assets to home equity
for purposes of minimizing EFC, we employ equation 1. The net benefit is equal to the
sum of the EFC reduction for each academic year, less the loss of the mortgage interest
deduction, plus the student loan interest deduction (SLID). SLID is added back into the
formula because it is an above the line tax deduction that offsets the loss of the mortgage
interest deduction when a student loan38 is used as financial aid.
= [∑ −

=0
] − [( −
) ] + [1]
EFCB and EFCA represent the EFC values before (B) and after (A) repositioning
the non-qualified assets. Determining the mortgage interest deduction loss, we subtract the
mortgage interest deduction after (MIDA) asset repositioning from that of before (MIDB).
We discuss other important variables that impact this transaction later, such as capital gains
taxes and the loss of equity premium.
The heart of equation 1 is determining the household EFC score. The process of
determining a college student’s need-based aid begins with the student and parents filling
out the FAFSA. If the student anticipates attending a top tier postsecondary institution, the
CSS Profile aid form must be filed as well. This paper focuses only on households that are
required to fill out the FAFSA, thus students seeking to attend postsecondary education
institutions that require the CSS Profile are excluded39
.

38 For purposes of this study, we classify a student loan as a Stafford, subsidized loan and is only taken when
the student does not qualify for a ell grant.
39 For a list of CSS Profile participating institutions visit College Board (i.e.
https://profileonline.collegeboard.org)
Texas Tech University, Brigham T. Dorman, August 2016
101
The Department of Education uses equation 2 to determine the awarded needbased
financial aid. It is the difference of the cost of attendance (COA) and the expected
family contribution (EFC) score.
= − , where [2]
=
((−)+(∗(− )))

+ (( −
) ∗ ) + ( ∗
). [3]
COA represents total expenses for attendance40, and EFC denotes a households
minimum financial resources41 that are expected to be allocated to the child’s
postsecondary expenditures. Simple mathematical processes demonstrate that as EFC
decreases, financial aid increases42, given COA is unchanged. Thus, households cannot
control COA, other than selecting a different institution, but have complete discretion on
the variation in EFC in order to maximize need-based aid.
Determining a prospective college students EFC score, demonstrated in equation 3,
comprises of two parts: parental financial support and student financial responsibility. The
framework for parental support includes evaluating household available income43
,
contributions from qualified and non-qualified assets, and the number of children enrolled
in college44. The following variables are used in the formula:

40 Cost of Attendance consists of tuition, room & board, required materials, supplies, travel, and other
personal expenses.
41 Household financial resources are evaluated differently between the parents, the child, and third party
support. Parental assets and income are weighted less significantly than the child’s income and assets. Third
party financial support may consist of scholarships earned, which decrease the child’s income dollar for dollar
in the EFC calculation.
42 Need-based aid is a first come first serve system. Prospective college students are encouraged to apply
through the FAFSA as early as possible at the beginning of the year in order to maximize their aid. Therefore,
each prospective college student may not receive the full amount of financial aid calculated.
43 As of 2015, parental income reported is two years prior to the academic enrollment year. For example, if
a prospective college student plans to attend in the 2018-2019 academic year, the parent’s 2016 income is
reported on the FAFSA.
44 Parental expected financial contribution is distributed proportionally among each child enrolled in a
qualified postsecondary education institution. For example, if the household has three children enrolled in
Texas Tech University, Brigham T. Dorman, August 2016
102
TIp = Total Income of the Parent(s)45
TAp = Total Allowances for the Parent(s)46
ACRp = Asset Conversion Rate for the Parent(s)47
NWp = Parent(s) Net Worth (excluding qualified assets)48
APAp = Asset Protection Allowance for the Parent(s)49
ES = Number of Household Members Enrolled in College

college, the total parent contribution is divided evenly among each child for EFC calculation purposes. Thus,
an increase in the number of college enrolled children reduces each child’s EFC score.
45 Parental total income is calculated by taking the parents AGI and adding back in any untaxed income &
benefits such as retirement contributions, then subtracting all “additional financial information,” which
includes items such as education credits, child support, and taxable income from need-based employment.
46 Parent total allowance is subtracted from the parent(s) total income to determine the amount of parental
income is available to support college expenditures. This allowance is calculated by summing up any income
taxes paid, social security allowances, income protection allowances, and employment expense allowances.
47 Asset conversion rates are set by the U.S. Department of Education. The Asset conversion rate for the
2015-2016 academic year is 0.12. This studies incorporates this rate into its evaluation.
48 Tax-advantaged savings designed for educational purposes are included in net worth (i.e. ESAs, 529 plans).
49 The asset protection allowance at the parent level is determined by the age of the parent (older parent, if
married, remarried, or unmarried & living together).
Texas Tech University, Brigham T. Dorman, August 2016
103
Student financial responsibility is a function of the student’s income50, any
qualified and non-qualified assets owned in the student’s name, and third party awards (i.e.
scholarships earned). The following variables that represent the student’s portion of the
EFC calculation are:
TIs = Total Income of the Student51
TAs = Total Allowances for the Student52
AVs = Assessment Value of Students Available Income53
NWs = Student Net Worth (excluding qualified assets)54
ARs = Assessment Rate of NWs
55

50 Distributions from 529 plans held in the name of a grandparent are considered income at the child’s level.
51 Total income of the student is calculated by adding together their taxable income and any untaxed income
or benefits, then subtracting any “additional financial information,” which may include items such as
scholarships. Untaxed income includes any 529 plans in the child’s name.
52 Total allowances for the student is the sum of income taxes paid, social security tax allowance, income
protection allowance set by the U.S. Department of Education, and allowances if parental adjusted available
income is negative.
53 The assessment value of a student’s available income is fixed by the U.S. Department of Education. The
2015-2016 academic year has an assessment value of 0.5. This study employs this assessment value in
determining the student’s income contribution.
54 Similar to the parent’s net worth, education specific tax advantaged accounts are included.
55 The student assessment rate on their net worth is set by the U.S. Department of Education. For the 2015-
2016 academic year, the assessment rate is 0.2. This study employs this rate.
Texas Tech University, Brigham T. Dorman, August 2016
104
Logistic Regression
The second part of our analysis evaluates the relationship between EFC
minimization demand and households with non-qualified savings (HOWS) by employing
the following logistic regression:
FAFSA Application (FASA) = α + β 1 HOWS + β 2 DUAL + β 3 Age
+ β4 FEMALE + β5 INCOME + β6 RACE + ɛ.
Where FASA is a dichotomous variable indicating whether or not the household
submitted a FAFSA application (a representation of EFC minimization demand); HOWS
is a dummy variable representing the home ownership (plus savings) status of the
household; DUAL is another dummy variable signifying whether or not the household is a
dual income, two parent household or not; Age represents the age of the older parent;
FEMALE represents the gender of the student; INCOME is a dummy variable indicating
the households income level; and RACE denotes the ethnicity of the student. This study
posits that HOWS have greater probabilities of submitting a FAFSA application – meaning
they have a higher demand for EFC minimization, compared to their counterparts. It is
important to note that the data does not provide details on home ownership contracts, thus
we make the assumption that all households have the ability to either pay down their
existing mortgage or take out a second mortgage with the savings.
Results
This study investigates the benefit of re-positioning non-qualified assets for the
purpose of minimizing household EFC. First, we develop a hypothetical pro-saving
household that anticipates their child to enroll in a non-CSS Profile higher education
institution for five years. Second, we model the U.S. Department of Education EFC
Texas Tech University, Brigham T. Dorman, August 2016
105
formula with our hypothetical household, and calculate the change in the household EFC
score when increasing the asset repositioning amount by $10,000 increments. Next, we
calculate the net and marginal benefit of repositioning non-qualified assets for different
income levels and asset amounts. Finally, we perform a logistic regression to evaluate the
relationship between equity building households that have non-qualified assets (HOWS).
We then discuss the complex nature of asset re-positioning for EFC minimization,
including tax and asset allocation propositions financial planners may want to consider
when evaluating such a transaction with a client who has a college funding goal. We make
the following hypotheses’:
1. Re-positioning non-qualified assets to home equity reduces household EFC scores.
2. The net benefit of asset re-positioning to home equity is most beneficial for middle
income households56
.
3. Equity building households, with non-qualified assets, have a higher demand for
need-based aid, compared to their counterparts.
Table 3.3 shows that regardless of the amount of non-qualified assets held by the
household, the EFC score has the same dollar-for-dollar change for every $10,000 that is
either re-positioned from or contributed to non-qualified accounts. In other words, for
every dollar that is either re-positioned from, or contributed to non-qualified accounts, the
household EFC score decreases or increases by $0.0564, respectively, ceteris paribus. This
is regardless of COA. For an in-state tuition scenario, and given the assumptions for the
hypothetical case, the household will reach an EFC ≥ COA by the fourth academic year

56 FAFSA applications require that income is reported as household income level two years prior to the
academic year.
Texas Tech University, Brigham T. Dorman, August 2016
106
when continuing to save in non-qualified accounts, resulting in ineligibility of need-based
aid for the final two years of postsecondary participation.
Table 3.3: Change in Household EFC Score for Every $10,000
Dual Income, Two Parent Household
Public Four-Year In-State (COA = $19,548)
Repositioning (reducing non-qualified assets)
Age of Oldest
Parent Academic Year EFC
Before
EFC
After
Δ in
EFC
%Δ in
EFC
Cumulative
Δ in EFC
Value
45 1 $18,041 $17,477 -$564 -3.13% -$564
46 2 $17,443 $16,879 -$564 -3.23% -$1,128
47 3 $16,839 $16,275 -$564 -3.35% -$1,692
48 4 $16,230 $15,666 -$564 -3.48% -$2,256
49 5 $15,621 $15,057 -$564 -3.61% -$2,820
Mean $16,835 $16,271 -$564 -3.36%
Contributing (increasing non-qualified assets)
Age of Oldest
Parent Academic Year EFC
Before
EFC
After
Δ in
EFC
%Δ in
EFC
Cumulative
Δ in EFC
Value
45 1 $18,041 $18,605 $564 3.13% $564
46 2 $18,571 $19,135 $564 3.04% $1,128
47 3 $19,095 $19,659 $564 2.95% $1,692
48 4 $19,614 $20,178 $564 2.88% $2,256
49 5 $20,133 $20,697 $564 2.80% $2,820
Mean $19,091 $19,655 $564 2.96%
Public Four-Year Out-of-State (COA = $34,031)
Repositioning (reducing non-qualified assets)
Age of Oldest
Parent Academic Year EFC
Before
EFC
After
Δ in
EFC
%Δ in
EFC
Cumulative
Δ in EFC
Value
45 1 $18,041 $17,477 -$564 -3.13% -$564
46 2 $17,443 $16,879 -$564 -3.23% -$1,128
47 3 $16,839 $16,275 -$564 -3.35% -$1,692
48 4 $16,230 $15,666 -$564 -3.48% -$2,256
49 5 $15,621 $15,057 -$564 -3.61% -$2,820
Mean $16,835 $16,271 -$564 -3.36%
Contributing (increasing non-qualified assets)
Age of Oldest
Parent Academic Year EFC
Before
EFC
After
Δ in
EFC
%Δ in
EFC
Cumulative
Δ in EFC
Value
45 1 $18,041 $18,605 $564 3.13% $564
46 2 $18,571 $19,135 $564 3.04% $1,128
47 3 $19,095 $19,659 $564 2.95% $1,692
48 4 $19,614 $20,178 $564 2.88% $2,256
49 5 $20,133 $20,697 $564 2.80% $2,820
Mean $19,091 $19,655 $564 2.96%
Texas Tech University, Brigham T. Dorman, August 2016
107
Results from Figure 3.4 demonstrate that the net benefit increases with lower
reported household income and larger amounts of non-qualified assets repositioned. Under
our hypothetical scenario, with reported income at the highest level of the 15% marginal
tax bracket (i.e. income just under $75,000), the household loses almost $4,000 of
mortgage interest deductions, but receives the greatest net benefit, compared to other
income levels, as long as they reposition at least $26,000 of non-qualified assets. When
this similar household repositions $150,000 they lose nearly $14,000 of mortgage interest
deductions, but this loss is offset by the reduction in EFC and the student loan interest
deduction, resulting in a net gain. Households that report income in the 25% marginal tax
bracket or greater experience larger losses of mortgage interest deductions, reducing the
net benefit. The net benefit diminishes once reported household income rises above
$100,000.
Texas Tech University, Brigham T. Dorman, August 2016
108
Figure 3.4: Base Case of 30 year, $300k Mortgage at 3.5%
Reposition Amount (Thousands)
Income
$10
$20
$30
$40
$50
$60
$70
$80
$90
$100
$110
$120
$130
$140
$150
$160
$170
$180
$190
$200
$210
$220
$230
$240
$250
$40k
$50k
$60k
$70k
$80k
$90k
$100k
$110k
$120k
Texas Tech University, Brigham T. Dorman, August 2016
109
Figure 3.5 shows that the greatest marginal increase in the net benefit is experienced
at approximately the reported income level of $70,000. The results practice a marginal
increase at a declining rate once the household re-positions greater than $230,000 of nonqualified
assets. The marginal benefit is reduced at the asset repositioning levels of
$170,000 and $120,000 for reported income of $80,000 and $90,000, respectively.
Households that report income of approximately $60,000 experience net gains in asset
repositioning when re-locating between $90,000 and $100,000, but the marginal benefit is
substantially less compared to higher income levels.
Figure 3.5: Marginal Change for Asset Re-positioning to Home Equity by Income Level
and Repositioning Amount
($3,000)
($2,000)
($1,000)
$0
$1,000
$2,000
$3,000
$4,000
$5,000
Marginal Change
Reposition Amount
$60,000
$70,000
$80,000
$90,000
Texas Tech University, Brigham T. Dorman, August 2016
110
The purpose of repositioning non-qualified assets is to aid households in
minimizing EFC. As the loss of mortgage interest deduction rises, the net benefit of
minimizing EFC declines. Homeowners look forward to deducting mortgage interest from
their taxable income each April 15. To evaluate the impact of different mortgage sizes and
changes of interest rates on the marginal benefit of repositioning to home equity, we use
$70,000 of reported income as a benchmark. Figure 3.6 shows that when interest rates rise,
the marginal benefit of repositioning non-qualified assets is reduced. Higher interest rates
imply greater losses of the mortgage interest rate deduction. Figures 3.7 and 3.8 show that
increasing (decreasing) the mortgage balance reduces (increases) the marginal benefit.
Households with smaller mortgage balances and report lower levels of marginal income
taxes demonstrate larger marginal increases as more assets are repositioned.
Figure 3.6: Marginal Change for Asset Re-positioning to Home Equity When Increasing
Mortgage Interest Rate
($3,000)
($2,000)
($1,000)
$0
$1,000
$2,000
$3,000
$4,000
$5,000
Marginal Change
Repositioning Amount
300k Mortgage
5% Mortgage Interest
Texas Tech University, Brigham T. Dorman, August 2016
111
Figure 3.7: Marginal Change for Asset Re-positioning to Home Equity When Increasing
Mortgage Amount
Figure 3.8: Marginal Change for Asset Re-positioning to Home Equity When Decreasing
Mortgage Amount
($2,000)
($1,000)
$0
$1,000
$2,000
$3,000
$4,000
$5,000
Marginal Change
Repositioning Amount
500k Mortgage
300k Mortgage
($1,000)
$0
$1,000
$2,000
$3,000
$4,000
$5,000
Marginal Change
Repositioning Amount
200k Mortgage
300k Mortgage
Texas Tech University, Brigham T. Dorman, August 2016
112
For the second half of this study, we evaluate the relationship between equity
building households, that have non-qualified assets, and FAFSA application behavior. We
do this by regressing whether or not respondents are classified as home owners with [nonqualified]
savings (HOWS) against if the responding household demands lowering their
EFC score via a submitted FAFSA application (FAFSA). We employ the following logistic
regression:
FAFSA Application (FAFSA) = α + β 1 HOWS + β 2 DUAL+ β 3 AGE + β4
FEMALE + β 5 INCOME+ β6 RACE + ɛ
The results from Table 3.4 imply that equity building households, that have nonqualified
assets, have a greater demand for lowering household EFC scores. These
households have a 16% greater probability of submitting a FAFSA application, compared
to their counterparts. In addition to this positive relationship between HOWS and FAFSA,
the regression results demonstrate that dual income, two parent households have a 50%
higher probability of demanding an EFC score reduction, compared to their counterparts.
However, income is negatively related to FAFSA application submission.
Texas Tech University, Brigham T. Dorman, August 2016
113
Table 3.4: Logistic Regression on FAFSA Application Behavior (N = 4,761)
Dep. Var. = FAFSA Application Submitted Odds Ratio Std. Err. P Value
Own Home + 529 (Non) 1.159 0.082 0.037 **
Dual Income Households (Non-Dual) 1.504 0.106 0.000 ***
Parent Age 1.003 0.006 0.599
Female (Male) 1.535 0.104 0.000 ***
Income 0.916 0.010 0.000 ***
Race (White)
Asian 0.979 0.122 0.863
Black 1.910 0.322 0.000 ***
Am. Indian 0.501 0.270 0.200
Hispanic 0.912 0.106 0.429
Multi-Race 0.838 0.103 0.150
Other Race 0.631 0.321 0.365
Intercept 3.774 3.373 0.137
***,**,* show significance 0.01, 0.05 & 0.10 levels respectively
Filling out the FAFSA is an effort to borrow for postsecondary education
expenditures, given that the EFC is greater than the Pell Grant eligibility threshold. Data
from the HSLS reports that HOWS fail to submit a FAFSA application because they feel
they can afford college without any financial aid, they feel ineligible, and they may not
seek EFC minimization because they dislike debt. Figure 3.9 shows that, of the 1,088
respondents that indicated why they chose not to complete the FAFSA, 56% of HOWS
neglected to complete the FAFSA because they feel they can afford college without a need
to minimize EFC. Nearly 55% of HOWS indicated that they did not submit the FAFSA
because they felt they were ineligible or unqualified, and 50% of HOWS ignored the
FAFSA because they did not want to avoid debt.
Texas Tech University, Brigham T. Dorman, August 2016
114
Figure 3.9: Why HOWS Fail to Complete the FASFA (N = 1,088)
Source: HSLS:09
Discussion
In this section we discuss factors associated with the choice to pay down the
mortgage with non-qualified assets for EFC minimization purposes. The purpose of this
section is to help financial professionals and households with the decision of implementing
an asset repositioning recommendation. We discuss tax information (i.e. capital gain and
tax loss harvesting), asset allocation, asset location, and equity premium implications.
Taxes
Repositioning non-qualified assets may require the sale of appreciated assets.
Given capital gains taxes will reduce any net gains, and that retirement assets are qualified,
households may find it beneficial to maximize tax deferred contributions and convert nonappreciating
assets first, such as cash. In the case of cash, tax implications are essentially
non-existent because there is no reduction in the net gain through capital gains. High basis
50%
56% 55%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Wish to Avoid Debt Can Afford without
Financial Aid
Feel Ineligible
Texas Tech University, Brigham T. Dorman, August 2016
115
assets should be re-located before low basis assets. However, if a household has low basis
non-qualified assets, they should consider other options such as donating and increasing
their charitable contributions. Charitable donations avoid capital gains taxes and reduce
reportable income and EFC.
Not all assets available to be sold are appreciated assets. Thus, it is worthy to note
that tax loss harvesting (Arnott, Berkin, and Ye, 2001) may help in repositioning
opportunities. Harvesting can reduce capital gains taxes and provide for a reduction in
future income tax liabilities.
Asset Allocation, Location, and the Equity Premium
Because of the large amount of assets needing to be re-located to home equity to
experience a reasonable net benefit, asset allocation and asset location become a critical
issue. Asset allocation is an important predictor of portfolio performance and should be a
function of risk aversion (Brinson, Hood, & Beebower, 1995; Ibbotson & Kaplan, 2000;
Riley & Chow, 1992). In the decision of re-locating non-qualified assets, households may
fail to recognize the impact on their portfolio asset allocation. Research argues that a home
should be considered a bond-like asset (Reichenstein, 1998). Thus, households may be
better off repositioning bonds, as opposed to stocks into home equity. However, this may
be a challenge for some households since research argues that optimal asset location is
founded upon positioning bonds in tax deferred accounts (Dammon, Spatt, & Zhang,
2004), which are classified as qualified assets for financial aid eligibility purposes. Paying
down a mortgage is synonymous to increasing the bond portion of a portfolio.
Texas Tech University, Brigham T. Dorman, August 2016
116
The notion of repositioning bonds may be open for debate since some households
with emergency funds may not use cash, but debt instruments. That means that households
seeking to maximize the use of emergency funds may place bonds in taxable accounts for
purposes of liquidity and greater returns. Thus, households may reposition emergency
funds into home equity. This permits households to open a Home Equity Line of Credit
(HELOC), of which they can simply access through check writing services for easy access
to the emergency funds. With college years demonstrating a short time horizon, HELOCs
allow households to bring these repositioned assets out of home equity and return them to
the financial markets.
Transferring equities into home equity may trigger a loss in equity premium (Fama
& French, 2002). Households with lower levels of risk aversion expect higher returns
because they are taking greater risk. Thus, repositioning equities may result in an increase
in risk aversion, signaling an imbalance between the household’s level of risk aversion and
their portfolio risk composition.
Although the existing research places a strong emphasis on locating tax-advantaged
securities in tax-disadvantaged accounts, and tax-disadvantaged securities in taxadvantaged
accounts, the discussion of optimal asset location may be incomplete due to
the lack of research on investigating when and if it makes sense to place tax-disadvantaged
assets in tax-disadvantaged accounts (i.e. bonds in taxable accounts).
Texas Tech University, Brigham T. Dorman, August 2016
117
Conclusion
College expenditures have risen and state postsecondary education funding has
declined, placing an increased financial burden on households with college goals. This
challenge increases the demand for household EFC minimization. HEA ’92 reclassified
home equity from non-qualified to a qualified asset that no longer penalizes households for
building home equity. EFC minimization, an act to maintain current consumption during
college years, is achieved through strategic asset re-positioning. This study investigates
the benefit of re-positioning non-qualified assets to home equity in order to minimize
household EFC, and maximize need-based aid. Findings suggest that converting nonqualified
assets may be beneficial because it decreases household EFC, reducing the
savings gap for households with both retirement and college savings goals. However, relocating
assets to home equity reopens the challenge of optimizing the household’s asset
allocation.
Financial advisors may question if they should recommend such a strategy to
clients. Given no costs, financial advisors should always recommend this strategy, but the
complexity of this transaction demonstrates that there are numerous costs that should not
be ignored. For example, the cost of repositioning includes, but is not limited to, capital
gains taxes, the loss of equity premium, and expenses related to employing a HELOC. This
study suggests that, given the costs involved, asset repositioning for the purpose of EFC
minimization may be useful in the following cases:
A. Homeowners that have the ability to report (two years prior) income at the
marginal tax rate of 15%, and have substantial non-qualified assets of high
basis.
Texas Tech University, Brigham T. Dorman, August 2016
118
B. Low interest rate economic environments.
C. Homeowners who may be eligible for tax loss harvesting.
D. Homeowners that have emergency savings eligible to be repositioned to home
equity at a low cost57
.
Future Research Considerations
This study provides insight to areas of potential future research. For example, if a
household maximizes their federal aid, and needs to consider additional borrowing
measures, should they use home equity as their second source, rather than a private loan?
Private loans, according to Sallie Mae are the second largest borrowing sources next to
federal loans, but often times carry substantially higher interest rates, depending on the
credit worthiness of the borrower. Future research may also consider evaluating the
breakeven point at which the expected future rate of return becomes sub-optimal to that of
asset re-positioning into home equity. Furthermore, additional research is warranted in
light of the discussion on asset location.

57 Because costs associate with HELOCs can vary substantially by institution we ignore costs for this study.
HELOCs can be opened with a non-mortgage servicing provider, but the client may experience lower fees if
taken with the mortgage servicing financial institution. Although it is common for lenders to not charge fees
or closing costs on HELOCs, clients and financial professionals must be aware that is not the case with all
HELOC providers. Fees charged on HELOCs may be related to, but not limited to, appraisals, administrative
processing, attorney services, or even title services. Since appraisal costs may constitute a large part of
potential fees, conscientious participants of HELOCs may consider seeking an automated valuation or a
broker’s price opinion, if it is accepted by the lender. Interest must be factored into the decision to pay down
the mortgage and take out a HELOC, as well as other administrative fees that may go undetected, such as
those for account maintenance, transaction fees, or possible account inactivity fees.
Texas Tech University, Brigham T. Dorman, August 2016
119
References
Adelman, S. W., Cross, M. L., & Shrider, D. G. (2010). Why Do Homeowners Make
Mortgage Curtailment Payments ? Journal of Housing Research, 19(2), 195–212.
Amromin, G., Huang, J., & Sialm, C. (2007). The Tradeoff Between Mortgage
Prepayments and Tax-deferred Retirement Savings. Journal of Public Economics,
91(10), 2014–2040.
Archibald, R. B., & Feldman, D. H. (2008). Explaining Increases in Higher Education
Costs. The Journal of Higher Education, 79(3), 268–295.
Arnott, R. D., Berkin, A. L., & Ye, J. (2001). Loss Harvesting: What’s It Worth to the
Taxable Investor?. The Journal of Wealth Management, 3(4), 10-18.
Babiarz, P., & Yilmazer, T. (2009). The Impact of College Financial Aid Rules on
Household Portfolio Choice. National Tax Journal, 635-655.
Baumgartner, F. R., & Jones, B. D. (1993). Agendas and Instability in American Politics.
Chicago: University of Chicago Press.
Brinson, G. P., Hood, L. R., & Beebower, G. L. (1995). Determinants of Portfolio
Performance. Financial Analysts Journal, 51(August 1986), 133–138.
Campbell, J. Y. (2006). Household Finance. Journal of Finance, 61, 1553–1604.
Cheslock, J. J., & Gianneschi, M. (2008). Replacing State Appropriations with
Alternative Revenue Sources: The Case of Voluntary Support. The Journal of
Higher Education, 79(2), 208–229.
Dammon, R. M., Spatt, C. S., & Zhang, H. H. (2004). Optimal Asset Location and
Allocation with Taxable and Tax-Deferred Investing. Journal of Finance, 59(3),
999–1037.
Davidoff, T. (2010). Home Equity Commitment and Long-term Care Insurance Demand.
Journal of Public Economics, 94(1-2), 44–49.
De La Rosa, M. L. (2006). Is Opportunity Knocking?: Low-Income Students’
Perceptions of College and Financial Aid. American Behavioral Scientist, 49(12),
1670–1686.
Delaney, J. A., & Doyle, W. R. (2014). State Spending on Higher Education: Testing the
Balance Wheel over Time. Journal of Education Finance, 36(4), 343–368.
Dick, A. W., & Edlin, A. S. (1997). The Implicit Taxes from College Financial Aid.
Journal of Public Economics, 65(3), 295–322.
Doyle, W. R. (2010). Does Merit-Based aid “Crowd Out” Need-Based aid? Research in
Higher Education, 51(5), 397–415.
Dynarski, S. M. (2002). Loans, Liquidity, and Schooling Decisions. Kennedy School of
Government Working Paper.
Texas Tech University, Brigham T. Dorman, August 2016
120
Dynarski, S. M. (2003). Does Aid Matter? Measuring the Effect of Student Aid on
College Attendance and Completion. The American Economic Review, 93(1), 279–
288.
Dynarski, S. M., & Scott-Clayton, J. (2013). Financial Aid Policy: Lessons from
Research (No. w18710). National Bureau of Economic Research.
Edlin, A. S. (1993). Is College Financial Aid Equitable and Efficient? Journal of
Economic Perspectives, 7(2), 143–158.
Fama, E. F., & French, K. R. (2002). The Equity Premium. The Journal of Finance,
57(2), 637–659.
Hannah, S. B. (1996). The Higher Education Act of 1992: Skills, Constraints, and the
Politics of Higher Education. The Journal of Higher Education, 67(5), 498.
Hearn, J. C. (2006). Alternate Revenue Sources. In D. M. Priest & E. P. St. John (Eds.),
Privatization and Public Universities (pp. 87–108). Bloomington: Indiana
University Press.
Ibbotson, R. G., & Kaplan, P. D. (2000). Does Asset Allocation Policy Explain 40, 90, or
100 Percent of Performance? Financial Analysts Journal, 56, 26–33.
Lusardi, A., & Mitchell, O. S. (2007). Baby Boomer retirement security: The roles of
planning, financial literacy, and housing wealth. Journal of Monetary Economics,
54(1), 205–224.
Long, M. (2004). The Impact of Asset-tested College Financial Aid on Household
Savings. Journal of Public Economics, 88(1), 63-68.
McCollum, M. N., Lee, H., & Pace, R. K. (2015). Deleveraging and Mortgage
Curtailment. Journal of Banking and Finance, 60, 60–75.
Mortenson, T. G. (2000). Poverty, Race, and the Failure of Public Policy: The Crisis of
Education Access in Higher. Academe, 86(6), 38–43.
Mumper, M. (2003). The Future of College Access: The Declining Role of Public Higher
Education in Promoting Equal Opportunity. The Annals of the American Academy of
Political and Social Science.
Office of Management and Budget. (2015). Analytical Perspectives: Budget of the U.S.
Government, FY 2016. US Government Printing Office, Washington, DC.
Poterba, J. M., Shoven, J. B., & Sialm, C. (2000). Asset Location for Retirement Savers
(No. w7991). National Bureau of Economic Research.
Reichenstein, W. (1998). Calculating a Family’s Asset Mix. Financial Services Review,
7(3), 195–206.
Riley, W. B., & Chow, K. V. (1992). Asset Allocation and Individual Risk Aversion.
Financial Analysts Journal, 48(6), 32–37.
Texas Tech University, Brigham T. Dorman, August 2016
121
Ryan, J. F. (2004). The Relationship Between Institutional Expenditures and Degree
Attainment at Baccalaureate Colleges. Research in Higher Education, 45(2), 97–
114.
Sallie Mae. (2012). How America Pays for College.
Sallie Mae. (2015a). How America Pays for College.
Sallie Mae. (2015b). How America Saves for College: Sallie Mae’s National Study of
Parents with Children Under Age 18.
Shoven, J. B. (1999). The Location and Allocation of Assets in Pension and Conventional
Savings Accounts. National Bureau of Economic Research Working Paper Series,
No. 7007.
Skinner, J. (2007). Are You Sure You’re Saving Enough for Retirement? Journal of
Economic Perspectives, 21, 59–80.
Stiglitz, J. E., Tyson, L. D., Orszag, P. R., & Orszag, J. M. (2000). The Impact of Paying
for College on Family Finances. Upromise, Inc., Massachusetts.
Titus, M. A. (2006). No College Student Left Behind: The Influence of Financial Aspects
of a State’s Higher Education Policy on College Completion. The Review of Higher
Education, 29(3), 293–317.
Titus, M. A. (2009). The Production of Bachelor’s Degrees and Financial Aspects of
State Higher Education Policy: A Dynamic Analysis. The Journal of Higher
Education, 80(4), 439–468.
Toutkoushian, R. K., & Shafiq, M. N. (2010). A Conceptual Analysis of State Support for
Higher Education: Appropriations Versus Need-based Financial Aid. Research in
Higher Education, 51(1), 40–64.
Trout, R. K. (2013). An In-Depth Look at the Tax Consequences of Asset Location. AAII
Journal, 35(3), 26–28.
Wolpaw Reyes, J. (2007). College Financial Aid Rules and the Allocation of Savings.
Education Economics, 16(2), 167-189.
Zhou, J. (2009). The Asset Location Puzzle: Taxes Matter. Journal of Economic
Dynamics and Control, 33(4), 955–969.
Texas Tech University, Brigham T. Dorman, August 2016
122
CHAPTER VI
CONCLUSION
College affordability has become the second most important financial goal for
many households. Prior studies suggest that reducing the financial barriers to a college
education increases post-secondary expectations, affordability, and improves enrollment
and completion rates (Dynarski & Scott-Clayton, 2013; Elliott, 2009). Therefore,
politicians, and households with prospective college students, are attracted to 529
education savings plans. However, planning for the cost of a college education becomes
complex when accounting for the probability that the child(ren) will complete college, and
given its rising cost, the challenge of a short time horizon, compared to that of retirement
saving.
529 education savings plans are the dominate college savings vehicle. Dynarski
(2004) and Hoxby (1998) suggest that the popularity of 529 plans remains for high income
households because of the tax deferment even if the assets are not used for college, while
Gokhale & Kotlikoff (2003) show that it is sub-optimal for low income households to
participate in tax deferred savings vehicles. This suggests that current public policy for
529 plans experiences efficiency loss (Dynarski & Scott-Clayton, 2006) and underserves
low and middle income households (Cramer & Schreur, 2013). The goal of public policy
is to increase postsecondary enrollment by offering 529 plans. Chapter II demonstrates that
forgoing federal tax revenues to support 529 plan participation may be an inefficient policy
for improving college enrollment due to the lack of empirical evidence to support
normative policy. Such findings support the argument that educational tax subsidies are
sub-optimal (Dynarski & Scott-Clayton, 2016).
Texas Tech University, Brigham T. Dorman, August 2016
123
Chapter III addresses the dangers of unrealistic optimism (Weinstein, 1980) and
how the risk that a child will fail to attend, or complete, college impacts optimal account
choice for college saving households. Prior research suggests that parents overestimate
their child’s cognitive ability (Miller, 1986). Such overconfidence may crowd out the
probability of incurring 529 tax shocks through non-qualified withdrawals. With college
completion rates below 50%, households seeking 529 plan participation should consider
the true probability of their child to complete college when evaluating account choice for
college savings.
Chapter IV demonstrates that households may be able to reduce the financial
burden of a college education by managing their expected family contribution score. Since
the 1992 amendment of the higher education act, home equity no longer penalizes
households attempting to maximize federal need-based financial aid. Thus, chapter IV
indicates that re-locating taxable assets into home equity, and paying down the mortgage,
may be a viable option that financial professionals can offer specific households seeking
advice on college affordability.
Research on the topic of financial planning for college appears to suggest that
current public policy on college affordability has room for improvement. Its focus on
educational tax subsidies that may be an inefficient method to reduce the financial burden
on college saving households, improve college affordability, and increase postsecondary
enrollment. The asymmetric growth in tuition and household income demonstrates that the
challenge of college affordability is increasing. This dissertation suggests that policy
leaders and financial professionals can better aid households in reducing the burden of
college affordability.
Texas Tech University, Brigham T. Dorman, August 2016
124
References
Cramer, R., & Schreur, E. (2013). Personal Savings and Tax Reform: Principles and
Policy Proposals for Reforming the Tax Code. New American Foundation.
Dynarski, S. M. (2004). Who Benefits From the Education Saving Incentives? Income,
Educational Expectations, and the Value of the 529 and Coverdell. National Tax
Journal, 57(2), 359–383.
Dynarski, S. M., & Scott-Clayton, J. (2013). Financial Aid Policy: Lessons from
Research (No. w18710). National Bureau of Economic Research.
Dynarski, S. M., & Scott-Clayton, J. (2006). The Cost of Complexity in Federal Student
Aid: Lessons from Optimal Tax Theory and Behavioral Economics. National Tax
Journal, LVIV, 319–356.
Dynarski, S. M., & Scott-Clayton, J. (2016). Tax Benefits for College Attendance (No.
w22127) (Vol. 53).
Elliott, W. (2009). Children’s College Aspirations and Expectations: The Potential Role
of Children’s Development Accounts (CDAs). Children and Youth Services Review,
31(2), 274–283.
Gokhale, J., & Kotlikoff, L. J. (2003). Who Gets Paid to Save? Tax Policy and the
Economy, 17(MIT Press), 111–139.
Hoxby, C. M. (1998). Tax Incentives for Higher Education. Higher Education, 12(MIT
Press), 49 – 82.
Miller, S. A. (1986). Parents’ Beliefs About Their Children’s Cognitive Abilities.
Developmental Psychology, 22(2), 276–284.
Weinstein, N. D. (1980). Unrealistic Optimisim About Future Life Events. Journal of
Personality and Social Psychology, 39(5), 806–820.