TCHARITY REALLY DOES BEGIN AT HOME:


 

FLORIDA TAX REVIEW 
Volume 11 2011 Number 2  
 

97 
 

IN DEFENSE OF COLLEGE SAVINGS PLANS: USING 529 PLANS TO 
INCREASE THE IMPACT OF DIRECT FEDERAL GRANTS FOR HIGHER 

EDUCATION TO LOW- AND MODERATE-INCOME STUDENTS 
 

by 
 

Caroline Waldner 
 

 
INTRODUCTION ............................................................................................ 98 
I. FEDERAL FINANCIAL AID SCHEME ....................................................... 101 

A. Background .................................................................................... 101 
B.  Current 529 Plans ......................................................................... 105 

1. Structure ............................................................................. 105 
2. Previous Evaluations of Effectiveness ................................ 107 

II. COLLEGE SAVINGS AS A USEFUL ELEMENT OF FEDERAL  
FINANCIAL AID .................................................................................. 110 
A. Lower-Income Students Under-Invest in Higher Education ........... 111 

1. Low- and Moderate-Income Students May Have Lower 
Returns from Higher Education ......................................... 117 

2. Hyperbolic Discounting ...................................................... 118 
3. Debt Aversion ..................................................................... 118 

B. Lower-Income Students Respond More to Grants Than Other  
Types of Aid ................................................................................ 120 

C. 529 Plans Have the Same Effect as Grants ..................................... 121 
III. POLICY PROPOSALS TO MAXIMIZE THE IMPACT OF 529 PLANS ..... 122 

A. Structure Plans to Induce Savings .................................................. 122 
B. Reduce Regressivity ........................................................................ 125 
C. Reduce Uncertainty ......................................................................... 127 

IV. AN ILLUSTRATION ............................................................................... 129 
CONCLUSION .............................................................................................. 135 
 
  



98 Florida Tax Review  [Vol. 11:2  

INTRODUCTION 
 

A high school graduate with no college education can expect to earn 
on average $638 per week. Those holding a bachelor’s degree earn almost 
twice as much at an average of $1,121 per week.1 This will amount to almost 
$1 million more in earnings over a lifetime. These abstract numbers, 
however, do not tell the entire story. Consider that at $638 a week, a person 
with only a high school diploma will earn on average $33,176 per year. This 
falls below the average annual family budget by almost $7,000 per year 
when including only necessities.2 In fact, this annual salary is lower than 
average family budgets in seven of the eight cities sampled.3  

In addition to these individual monetary benefits, higher education 
also has significant non-monetary benefits as well. College graduates 
overwhelming report being happier and healthier than those who only 
attended high school, and most often reported being in “excellent” or “very 
good health.”4 This disparity increased with age — for those over the age of 
65, 70 percent of college graduates identified themselves as being in good 
health compared to “only 45 [percent] of high school graduates.”5  

These benefits can be seen in society as a whole as well. Some of the 
societal benefits of increased enrollment in higher education include 
reductions in unemployment, poverty, dependence on social welfare 
programs, and crime. College graduates also have healthier lifestyles, more 
positive perceptions of personal health, and lower smoking rates than non-
graduates.6 Completing post-secondary education is also correlated with 
higher rates of civic participation, including volunteer work, blood donation, 
                                                      

1. U.S. Bureau of Labor Statistics, Usual Weekly Earnings of Wage and 
Salary Workers News Release (Jan. 21, 2010), http://www.bls.gov/ 
news.release/archives/wkyeng_01212010.htm. This difference means that a typical 
bachelor’s degree holder will earn approximately 61 percent more over his or her 
working life than someone holding only a high school diploma. See generally Sandy 
Baum & Jennifer Ma, College Board, Education Pays:  The Benefits of Higher 
Education for Individuals and Society, 11 (2007), http://www.collegeboard.com/ 
prod_downloads/about/news_info/cbsenior/yr2007/ed-pays-2007.pdf. 

2. Sylvia A. Allegretto, Economic Policy Institute, Basic Family Budgets: 
Working Families’ Incomes Often Fail to Meet Living Expenses Around the U.S. 
(Aug. 30, 2005), available at http://www.epi.org/publications/entry/bp165/. 
“Necessities” in the study included food, housing, transportation, taxes, child care, 
and healthcare.  

3. Id. The only city to have a family budget below this amount was Casper, 
Wyoming, the most rural city in the study.  

4. Inside Higher Ed, The (Non-Monetary) Value of a College Degree (Sept. 
13, 2007), http://www.insidehighered.com/news/2007/09/13/collegeboard 
(summarizing a survey conducted by the National Center for Health Statistics).  

5. Id.  
6. Baum & Ma, supra note 1, at 2.  



2011] In Defense of College Savings Plans 99 
 
voting, and openness to the opinions of others. There is even evidence that 
the earnings of workers with lower education levels are positively affected 
by increased numbers of college graduates in the workforce.7 
 Recognizing the numerous positive externalities and internalities 
associated with higher education, the federal government has chosen to 
heavily subsidize this area in order to increase enrollment. The government 
currently spends roughly $180 billion dollars per year subsidizing higher 
education costs through a variety of educational incentives ranging from 
direct grants to deductions for interest paid on student loans.8 Despite this 
tremendous effort, there has been little increased enrollment in higher 
education in recent years.  

The federal government began financing higher education in the 
1960s.9 This support came first in the provision of direct financial aid, 
primarily targeted at lower-income students and families. More recently, the 
government has expanded its role by delving into the world of tax incentives. 
While some of these incentives may still reach those in lower- and middle-
income households (specifically the Hope and Lifetime Learning credits), 
they also include broader tax incentives and advantages that reach even those 
in the highest income ranges. None of these incentives can reach the low-
income taxpayers who are generally in the zero bracket, and thus are not 
influenced by deductions, exclusions, and non-refundable credits, which are 
the forms current tax incentives take.10  

                                                      
7. Id.  
8. See College Board, Trends in Student Aid 2009, at 6 (2009), 

http://inpathways.net/2009_Trends_Student_Aid.pdf. Further, federal tax 
expenditures are currently estimated at $35 billion annually. Staff of the Joint 
Comm. on Tax’n, Estimates of Federal Tax Expenditures for Fiscal Years 2007-
2011, at 31 tbl.1 (Comm. Print 2007), available at http://www.jct.gov/s-3-07.pdf.  
This estimate includes tuition tax credits, the deductions for tuition and interest on 
education loans, and education savings accounts.  The Stafford Loan program, 
including subsidized, unsubsidized, and consolidated loans, currently costs roughly 
$7.05 billion annually. The Pell Grant Program is funded with $18.2 billion 
annually.  See Memorandum from Deborah Kalcevic & Justin Humphrey on CBO 
March 2007 Baseline Projections for the Student Loan and Grant Programs (Mar. 2, 
2007), available at http://www.cbo.gov/budget/ factsheets/2007b/studentloans.pdf; 
see also Sima J. Gandhi, Viewing Education Loans Through a Myopic Lens 11 
(Brookings Inst., The Hamilton Project Discussion Paper 2008-05, 2008), available 
at http://ssrn.com/abstract=1607774. 

9. See Higher Education Act of 1965, Pub. L. No. 89-329, § 401, 79 Stat. 
1219, 1232 (1965). 

10. As part of the American Recovery and Reinvestment Act of 2009, the 
American Opportunity Tax Credit replaced and revised the Hope Credit for tax years 
2009 and 2010.  This credit is partially refundable (40 percent of the maximum of 
$2,500 credit, or $1,000) and thus can reach those with little or no tax liability.  



100 Florida Tax Review  [Vol. 11:2  

This paper will focus on one such tax incentive — the tax preferred 
college savings plan, or 529 plan — and its current and potential impact on 
higher-education participation rates among low- and moderate-income 
students.  

This article will synthesize and describe the existing literature 
attempting to explain the low participation in higher education among low-
to-moderate-income students, despite economic models that indicate that 
participation should be equivalent across income levels. Building on the 
recommendations stemming from the literature, which suggests that grants 
are the best way to increase those participations rates, I will argue that 529 
college savings plans can, with important changes from their current 
structure, be a similar but more powerful tool than direct grants. In 
conclusion, the paper will attempt to illustrate one possible revenue-neutral 
plan for the federal government making use of 529 plans. The example will 
show that a 529 plan with an initial contribution from the government 
offered to every low- and moderate-income child, and matching grants 
annually, can increase enrollment among this group of students by almost 
five percentage points. This increase would far outpace any seen over the 
previous two decades, and yet is obtained using a very small portion of the 
total federal financial aid program. 

Such an increase would have positive impacts on social equality and 
mobility and offer substantial monetary benefits to those affected. The 
increased enrollment would also have widespread societal benefits like 
raising all wages, reducing healthcare costs, and increasing volunteerism and 
civic participation.11 

This article offers a revenue-neutral proposal for amending 529 plans 
simply to show the additional impact that the same amount of money could 
have if targeted differently. This paper argues, however, that these 529 plans 
are the most efficient way to offer financial aid incentives because they lead 
to the largest effective grant. I would therefore advocate for an increased 
amount of federal funding to be directed at these plans, either as increased 
spending on educational incentives or redirected from other programs. 
However, it is beyond the scope of this paper to analyze all of the financial 
aid programs and whether or how much of that funding would be better 
allocated to 529 plans.12 
                                                                                                                             
President Obama has proposed making this tax credit permanent but as of this 
writing it applies only for these two years.   

11. Baum & Ma, supra note 1, at 1-2.  
12. It would be too simplistic to say that all federal funding for education 

could or should be distributed through these 529 plans.  It may be that no amount of 
incentive can entice those of very low-income to save for education or otherwise.  
After all, it is impossible to save money that you do not have.  The best incentive to 
enroll in education for these households would therefore be an upfront grant at the 
time of enrollment, like the Pell Grant.  It would therefore not be wise to redirect all 



2011] In Defense of College Savings Plans 101 
 

Part I will offer a brief description of the federal financial aid 
scheme as it exists today, including an overview of 529 plans as currently 
structured. Part II explains why grants are the most effective form of 
financial aid for increasing enrollment in higher education, and how 529 
plans operate in largely the same manner. Part III offers a number of 
proposals for maximizing the impact that 529 plans can have on higher 
education attainment. Part IV provides an illustration of how a 529 plan 
might be structured around these proposals and demonstrates the impact that 
such a remodeled 529 plan could have on enrollment rates.   

 
I. FEDERAL FINANCIAL AID SCHEME 

  
A. Background  
 
 The federal financial aid scheme was adopted in response to the 
widely held beliefs that (1) the high cost of higher education was acting as a 
barrier to entry, (2) higher education produces a number of positive 
externalities, and (3) a failure to equalize opportunity perpetuates income 
inequality because of the associated market rewards.13 All three of these 
beliefs led to the aim of decreasing the cost in order to increase enrollment.  
 The initial push of federal financial aid focused on direct aid targeted 
at lower-income students under the rationale that moderate- and high-income 
students could afford to attend college without help from the federal 
government. This remained true for roughly twenty years through the Pell 
Grant and Stafford Loans systems, both of which are need-based and 
primarily awarded to students from families with income below $40,000.14 
The formula for determining eligibility for these types of aid is highly 
progressive.15 

                                                                                                                             
of that type of funding into a 529 matching program.  A similar argument could be 
made for student loans; one aspect of Human Capital Theory, discussed infra, is that 
there is no liquidity problem for obtaining college education.  This is largely true 
because of the availability of student loans.  Eliminating government support of 
student loans could therefore change that entire calculus.  Again, in-depth analysis 
into such issues is beyond the scope of this paper.  Suffice it to say that there are 
certainly existing government expenditures that would be more effective if spent on 
matching grant 529 plans.   

13. See Deborah H. Schenk & Andrew L. Grossman, Failure of Tax 
Incentives for Education, 61 Tax L. Rev. 295, 296-97 (2008).  

14. Susan Dynarski, Hope for Whom? Financial Aid for the Middle Class 
and its Impact on College Attendance 1 (Nat’l Bureau of Econ. Research, Working 
Paper No. 7756), available at http://www.nber.org/papers/w7756.pdf.  

15. See id.  Progressive as used here indicates that these types of aid are 
given more heavily to students of lower income and less to students of relatively 
moderate income, and generally not at all to students of high-income households.  



102 Florida Tax Review  [Vol. 11:2  

 Over the past twenty years, however, there has been a substantial 
increase in the cost of both public and private higher education. This rise in 
price has outpaced increases in both inflation and family income and has 
priced many more families out of higher education obtainment.16 In 
inflation-adjusted dollars, the price of a four-year private education rose by 
$5,500 over the past decade, while four-year public universities have 
increased by $2,200.17 

To counter this trend, many of the new federal financial aid 
programs have been targeted less toward lower-income students and instead 
more toward moderate-to-high-income families. Many of the incentives 
directed toward these classes of students are tax incentives such as the 
deduction for interest on student loans, tax preferred savings vehicles, and 
more recently, higher education tax credits.  

At the same time, those programs targeted at lower-income families 
have lost much of their power. The maximum Pell Grant, when established 
in the 1970s, covered approximately 77 percent of the cost of a four-year 
public education for those who received it. It now covers only 30 percent of 
that cost.18 These changes, combined with the fact that low-to-moderate-
income students are much more price-sensitive to college tuition, means that 
there are far less of these students obtaining higher education than high-

                                                                                                                             
This thus results in a redistribution of wealth from higher-income students and 
households to lower-income students and households. A progressive tax system is 
one where average rates go up as income increases, so that those with a higher 
income are paying a larger percentage of that income towards taxes. This is 
contrasted with a proportional system where the tax rate remains constant for every 
level of income. For example, in a proportional system a taxpayer with $50,000 of 
income would pay $5,000 (10 percent) of taxes and someone with income of 
$100,000 would pay $10,000 (10 percent still) of taxes. In contrast, under a 
progressive system, someone with income of $50,000 might pay $5,000 (10 percent) 
of taxes while a taxpayer with income of $100,000 might pay $20,000 (20 percent) 
of taxes. Because the many complex features of the tax code that affect the effective 
tax rate that people pay, some scholars have advocated thinking of a progressive tax 
system more simply as one where after-tax income is more equally distributed than 
before-tax income. See, e.g., Thomas Piketty & Emmanuel Saez, How Progressive is 
the U.S. Federal Tax System? A Historical and International Perspective, 21 J. Econ. 
Persp. 3, 5 (2007).  

16. See College Board, Trends in College Pricing 2006, at 7 (2006), 
http://www.collegeboard.com/prod_downloads/press/cost06/trends_college_pricing_
06.pdf 

17. College Board, Trends in College Pricing 2008, at 9 (2008), 
http://professionals.collegeboard.com/profdownload/trends-in-college-pricing-2008. 
pdf.  

18. Francine Knowles, Durbin, Roosevelt Students Tout Hike in Pell 
Grants, Chicago Sun-Times (Mar. 30, 2010), http://www.suntimes.com/business/ 
2129674,CST-NWS-Pell30.article. 



2011] In Defense of College Savings Plans 103 
 
income students. The rate of enrollment among those from households with 
income below $18,000 per year has never been above 30 percent, while the 
rate for students from households with annual income of $60,000 (still 
considered moderate income) is consistently above 50 percent.19 Further, the 
percentage of high school graduates enrolling in college in the lowest 
quintile stayed roughly the same from the 70s to the 90s, at about 43 percent. 
At the same time, however, the percent of those in the highest quintile rose 
by approximately ten percentage points, from about 70 percent to about 80 
percent.20  

A number of commentators have called these recent tax incentives 
for higher education complete failures — they cost the federal government 
billions of dollars annually but incentivize essentially no one to obtain a 
higher education.21 In fact, after these tax incentives took effect in 1998, 
there was no increase in the percentage of high school graduates enrolled in 
post-secondary education (Table 1). 

 
TABLE 1 

POPULATION AGE 18-24 ENROLLED IN POST-SECONDARY SCHOOLS 22 
 

Year Percent of High School Graduates  
1992 41.9 
1993 41.4 
1994 42.4 
1995 42.4 
1996 43.5 
1997 45.2 
1998 45.3 
1999 43.8 
2000 43.3 
2001 44.1 
2002 45.0 

   
 This is true not only of the population as a whole, but for lower- and 
middle-income students as well (Table 2). Enrollment rates for these students 
showed fluctuations over the time period but no clear increase over previous 
rates of enrollment (Table 2).  
 
                                                      

19. See infra Table 2.  
20. Charles F. Manski, Income and Higher Education, 14 Focus 14, 17 

(1992),   http://www.irp.wisc.edu/publications /focus/pdfs/foc143c.pdf. 
21. See Schenk & Grossman, supra note 13, at 298-99. 
22. U.S. Census Bureau, Current Population Survey tbl.A-5, The Population 

14 to 24 Years Old by High School Graduate Status, College Enrollment, 
Attainment, Sex, Race, and Hispanic Origin: October 1967 to 2002 (Jan. 4, 2004), 
available at http://www.census.gov/population/ socdemo/school/tabA-5.pdf. 



104 Florida Tax Review  [Vol. 11:2  

TABLE 2 
PERCENTAGE OF POPULATION, AGE 18-24,  

ENROLLED IN COLLEGE BY FAMILY INCOME23 
 

Year/Income $18,000 $35,000 $60,000 
1992 29.4 38.2 52.6 
1997 26.1 40.1 58.4 
2002 29.2 32.8 52.6 
2005 29.9 40.2 53.8 

 
The problem is that these tax incentives are too opaque and targeted 

at the wrong people. One incentive that has been particularly criticized is the 
529 college savings plan. These plans are: regressive, in that the benefit is 
tied to the taxpayer’s tax bracket; expensive, costing the federal government 
approximately $1 billion annually; and used almost entirely by high-income 
households.24 Since high-income students are presumed to need no incentive 
to obtain higher education, many believe that these plans are an entirely 
misdirected use of federal funds.25 Despite the fact that many believe that 
high-income families need no incentive to save money or to enroll in higher 
education, almost 60 percent of the funds in 529 accounts are held by those 
in the top 5 percent of household income.26 In contrast, virtually none of the 
balance of these accounts is held by those in the first quartile.27 

 
  

                                                      
23. Data from 1992-2002 from Schenk & Grossman, supra note 13, at 362, 

slightly modified to reflect changes in reporting in the 2005 reported data.  2005 data 
compiled from U.S. Census Bureau, Current Population Survey tbl.14, Enrollment 
Status of Dependent Primary Family Members 18 to 24 Years Old, by Family 
Income, Level of Enrollment, Attendance Status, Type of School, Sex, Race, and 
Hispanic Origin: October 2005, available at http://www.census.gov/population/ 
www/socdemo/school/cps2005.html. The income listed on the table represents the 
median of the range used by the Census Bureau. 

24. Staff of the Joint Comm. on Tax’n, supra note 8. One billion dollars is 
the estimate provided for 2010.  

25. See William G. Bowen, Martin A. Kurzweil & Eugene M. Tobin, 
Equity and Excellence in American Higher Education 97 (2005) (“Children from 
families [in the highest income brackets] almost surely would go to college in any 
case.”); see also Schenk & Grossman, supra note 13, at 351 (“Higher income 
taxpayers, []have a strong propensity to save, especially for college, and need no 
incentive to do so . . . .”).   

26. See Schenk & Grossman, supra note 13, at 351; infra Figure 1.  
27. See infra Figure 1. 



2011] In Defense of College Savings Plans 105 
 
B. Current 529 Plans 
 

1. Structure  
 
Qualified tuition plans, commonly referred to as “529 plans” for the 

code section which authorizes them, were first introduced into the tax code in 
1996.28 Section 529 allows all fifty states and the District of Columbia to 
administer such plans and provides rules that the plans must follow in order 
to receive favorable tax treatment.29 The plans are then administered by the 
states. Forty-three of the states allow investment by out-of-state residents 
while five make them available only for in-state residents.30 Most states 
contract out management and record-keeping functions, and all states 
contract out investment services to third-party financial services 
companies.31 These plans can, but do not always, come with fees attached for 
enrollment, maintenance, and administration.32 The state tax benefits can 
vary widely but are generally more favorable than federal tax treatment. 
Most are both deductible when money is contributed and excludable upon 
withdrawal under state income taxes.33  
                                                      

28. IRC § 529. There are actually two different types of plan authorized 
under § 529 of the code:  college savings plans and prepaid tuition plans. In 2008, 88 
percent of assets were in college savings plans with only 12 percent in prepaid 
tuition plans. There has been a steady shift away from prepaid tuition plans (which 
pre-date college savings plans) to college savings plans. While the tax treatment of 
the two types of plans are highly similar, this paper will only specifically discuss 
college savings plan and this is what is meant by 529 plans.  

29. The only rule of real consequence here is that each plan must have one 
and only one “designated beneficiary” at all times.  

30. The plans are available in the District of Columbia and every state 
except Wyoming. In addition, Tennessee and Washington offer only prepaid tuition 
plans and not savings plans.  Dep’t of Treas., An Analysis of Section 529 College 
Savings and Prepaid Tuition Plans: A Report Prepared by the Department of 
Treasury for the White House Task Force on Middle Class Working Families 1 
(Sept. 9, 2009), http://www.treasury.gov/resource-center/economic-policy/ 
Documents/09092009TreasuryReportSection529.pdf  

31. Id. at 2.  
32. Plans with no fees attached are offered in Arizona, California, 

Connecticut, Florida, Georgia, Idaho, Illinois, Iowa, Kentucky, Louisiana, Michigan, 
Minnesota, Oklahoma, Tennessee, and Vermont. See generally Your Guide to 
Saving for College, http://www.savingforcollege.com (last visited Jan. 26, 2011).  

33. A total of thirty-two states offer a full or partial income tax deduction 
for contributions to the state’s 529 plan or have no income tax. Of those, five states 
offer a full income tax deduction, twenty states a partial income tax deduction, and 
seven states have no income tax. Those states that offer a partial income tax 
deduction often allow a carry forward of excess contributions for up to five years. 
See Saving For College, Compare 529 Plans, Compare by Features, 
http://www.savingforcollege.com/compare_529_plans/index.php?plan_question_ids[



106 Florida Tax Review  [Vol. 11:2  

 Investors in these accounts cannot fully control how the funds are 
invested, but generally are able to choose between several options. This 
choice usually allows variations based on the time until college enrollment 
for the beneficiary and personal preferences for risk.34 Some states offer 
investment options that are guaranteed to grow with the rate of tuition for in-
state colleges.35  

 The primary federal tax benefit stems from the exclusion from 
federal income tax of withdrawals from a qualified account used for qualified 
higher education expenses. The earnings accrued on the accounts are also 
tax-free. Contributions are made from after-tax income for federal tax 
purposes.36 529 plans grew significantly after 2001 when they were revised 
to make distributions tax-free at a federal level as long as used for qualified 
expenses.37 Prior to that time, the benefit of such plans was that contributions 
could grow tax-deferred; the earnings on such accounts were only taxable 
once distributed. After the benefit changed from deferral to exemption, the 
assets in 529 plans grew from $14 billion to $130 billion within six years.38  

Section 529 plans also allow for beneficial gift tax treatment.39 Up to 
five years of tax-exempt giving can be compressed into a one-year period. As 
of 2010, the § 2503(h) gift tax exclusion was $13,000 annually.40 Based on 
this amount, a donor could deposit up to $65,000 into the account in one year 
without being subject to the gift tax. The donor could then not take 
advantage of the annual exclusion as to that beneficiary for the next four 
years, but this allows a larger lump sum and therefore a larger accumulation 
of interest over that time. 
  

                                                                                                                             
]=437&plan_question_ids[]=85&mode=Compare&page=compare_plan_questions&
plan_type_id=; http://www.finaid.org/savings/529plans.phtml (last visited Jan. 26, 
2011). 

34. Schenk & Grossman, supra note 13, at 303-04.  
35. See, e.g., Montana Family Education Savings Program, CollegeSure® 

CD, http://montana.collegesavings.com/Montana/collegesure_cd.asp (last visited 
Jan. 26, 2011).  

36. See generally What is a 529 Plan?, http://savingforcollege.com/ 
intro_to_529s/what-is-a-529-plan.php (last visited Jan. 26, 2011).  

37. IRC § 529(e)(3)(B); Economic Growth and Tax Relief Reconciliation 
Act, Pub. L. No. 107-16, 115 Stat. 38, (2001); see also Christopher E. Houston, 
Section 529 Plans: Opportunities and Uncertainties, 2002 A.L.I.-A.B.A. 
Sophisticated Est. Plan. Techs. 63, 65. 

38. Dep’t of Treas., Analysis of Section 529, supra note 30, at 3. For 
comparison, there are an estimated $2 trillion in assets in 401(k) accounts.  

39. IRC § 529(c)(2)(B). This result is achieved through an election that 
allows the taxpayer to treat a contribution as if it was made over a five-year period 
rather than in one year.  

40. Rev. Proc. 2009-50, 2009-45 I.R.B. 617. 



2011] In Defense of College Savings Plans 107 
 

2. Previous Evaluations of Effectiveness  
 
529 plans have been largely criticized as being both ineffective and 

highly regressive. Because the primary benefit from 529 plans is the 
exclusion from income, the benefit from these plans is tied to the taxpayer’s 
tax bracket and is thus worth more for high-income (high-bracket) taxpayers. 
The regressivity of the program is compounded by the fact that high-income 
taxpayers generally are willing and able save more than middle- and lower-
income taxpayers. The plans can also offer no (federal) incentive to the 
significant number of taxpayers in the zero bracket who can make no use of 
an exclusion (or deduction).41  
 There are a number of additional elements of 529 plans that make 
them particularly attractive for high-income taxpayers. First, the limits on 
such plans are quite high, generally allowing enough investment to cover 
four years of private higher education plus additional years of graduate 
study, varying by state. Further, while each beneficiary can have only one 
account in his or her name in each state, there is no limit to the number of 
total accounts in each beneficiary’s name. Therefore a family could 
potentially hold one account in each state open to non-residents for each 
beneficiary up to the limit of approximately $300,000.42 This would result in 
a total investment of roughly $12.9 million per beneficiary.43 The benefit of 
investing more than could possibly be spent on qualified higher education 
expenses is that when a non-qualified distribution from the account is made 
(i.e., not for higher education expenses), it can be included in the 
beneficiary’s income and thus is taxed at a lower rate than would apply to the 
account owner. Non-qualified withdrawals have also still benefited from 

                                                      
41. It has been estimated that 43.4 million individuals paid no income tax in 

2006, amounting to 32 percent of the returns filed.  Scott A. Hodge, Number of 
Americans Paying Zero Federal Income Tax Grows to 43.4 Million, Tax Found. 1, 2 
(2006), http://www.taxfoundation.org/files/ff54.pdf. The number of those 
individuals actually in the zero bracket is no doubt lower than that number given that 
some of these people eliminated their tax liability through the use of credits, and the 
number of these individuals making use of the educational credits is not known.  It is 
clear that the percentage of people who cannot make use of these tax incentives is 
not insignificant.  

42. The limit currently ranges from $224,465 to $368,600 and exceeds or 
equals $300,000 in a majority of states. There are forty-three states offering plans 
that are open to non-residents. See Dep’t of Treas., Analysis of Section 529, supra 
note 30, at 2.  

43. Calculations derived using the estimate of a $300,000 limit in forty-
three plans. Note that a resident in one of the five states with accounts open only to 
residents could have plans in forty-four states — all forty-three open to non-residents 
plus in his or her resident state.  



108 Florida Tax Review  [Vol. 11:2  

deferral. The 10 percent penalty is often not sufficient to offset the benefit of 
tax deferral and eventual taxation at this lower rate.44  
 Besides the fact that it appears that high-income tax payers can use 
529 plans as a tax shelter, reaping tax benefits even when not used for 
educational savings, it also seems that they provide no actual incentive to 
higher-income individuals. High-income individuals tend to save for 
education without any tax incentive to do so.45 Those in the highest tax 
brackets also generally have enough current income to pay for education 
such that they do not need a financial incentive to encourage enrollment.46 
 There is also an element of uncertainty surrounding 529 plans that 
reduce any incentive that these plans might have for middle- and lower-
income families.47 Families, especially at lower levels of income, are often 
uncertain whether their children will attend college, what type of college 
they will attend, the amount of tuition that will need to be paid, and the 
amount of financial aid that they will receive. There is further uncertainty 
and complexity surrounding the interaction of 529 plans with other benefits, 
particularly availability of need-based federal financial aid. This uncertainty 
arises because assets of both the parents and the child are counted in 
determining the amount of aid the student is otherwise eligible to receive, 
and a 529 plan is an asset that is considered in the determination. These 
ambiguities might well keep an interested taxpayer from taking advantage of 
529 plans.48  

                                                      
44. Assume the withdrawal is $10,000.  Also assume that the account owner 

is in the 35 percent bracket and the beneficiary, son of the account owner, is in the 
15 percent bracket. Without even considering the benefit of deferral, the account 
owner would owe taxes of $3,500 ($10,000 * .35) on a taxable account, while the 
beneficiary would owe taxes of only $2,500 [($10,000*.15) + ($10,000 * .10)] when 
withdrawing from a 529 plan with a penalty.  

45. Schenk & Grossman, supra note 13, at 350. (“[I]nvestment decisions [of 
taxpayers in the 33 percent and 35 percent tax brackets] with respect to education 
will not be affected by the availability of a 529.”) 

46. Bowen, Kurzweil & Tobin, supra note 25, at 197 (“Children from 
families who elect to participate in the savings plan almost surely would go to 
college in any case.”); see also SallieMae, How America Pays for College, at 6 
(2009),  available  at  http://www.salliemae.com/NR/rdonlyres/52D9FB57-D14A-
46EA-A6D9-ECB284D13FD/11499/GCR1979_2009_PAYS_survey_final_091609.pdf,  
(finding that families of higher-income students cover, on average, over two-thirds 
of costs out of current income as compared to only 19 percent for lower-income 
families).  

47. See Schenk & Grossman, supra note 13, at 346 (arguing that “it is . . . 
extremely unlikely that taxpayers in either [the 10 percent bracket] or the 15 percent 
bracket will be able to overcome the ambiguities associated with 529 accounts”).  

48. Id.  



2011] In Defense of College Savings Plans 109 
 

Assuming that this incentive then is aimed almost entirely at those 
who do not need to be incentivized, this is an extremely inefficient use of 
government funds if the goal is to increase college enrollment. 
 Empirical data relating 529 participation rates to income, while 
limited, confirms this expected pattern of contribution. As illustrated in 
Table 3 and Figure 1, data from the Survey of Consumer Finances shows that 
participation rises rapidly with income; less than 1 percent of those in the 
lower half of the income distribution participate while over 30 percent 
participate in the top 5 percent of the income distribution.49 The asset 
balance in 529 plans also follows this trend; for the taxpayers with income in 
the first three quartiles, the average account balance is about $8,000. The 
average balance rises to over $100,000 for those in the 95-100 percentiles 
(Table 3 and Figure 1).  

TABLE 3 
EDUCATION SAVINGS PLANS ACCOUNT BALANCES HELD BY INCOME GROUP50 

                                                      
49. The survey suffers from a number of problems. First, it has a relatively 

small sample size. Second, both income and assets were self-reported and thus 
subject to measurement error. While the survey asked about 529s and Coverdells 
separately, the public data does not distinguish between the two. This is likely to 
skew the data somewhat because high-income individuals are more likely to invest 
in 529s than Coverdells because of Coverdells low contribution limits.  

50. Dep’t of Treas, Analysis of Section 529, supra note 30, at 30 tbl.7. 

Section 529 and Coverdell ESA Account Balances by Income Group 

Item Income Percentile Range 
0-25 25-50 50-75 75-90 90-95 95-100 

With Children       
Observations 1,439 1,479 1,428 966 440 1,808 
Percent w/     
529/CSA 

 
     0.4 

 
   1.2 

 
  8.6 

 
15.0 

 
27.8 

 
31.4 

With Children 
& 529 

      

Observations 1 20 112 148 117 352 
Average for:       
Income $27,766 $47,827 $80,005 $120,177 $176,284 $548,077 
529/CSA                     
Balance 

 
$  3,000 

 
$  8,794 

 
$  8,111 

 
$  15,482 

 
$  30,674 

 
$106,250 

Balance as % of 
Income 

 10.8   20.0 10.2 13.4 18.0 24.5 

 Number of 
Children 

 
  1.0 

 
   1.3 

 
  1.9 

 
  1.8 

 
  2.4 

 
  2.0 

Age Oldest Child 17.0  13.0 12.7 11.9 12.0 12.8 
Percent of Total:       
529/CSA 
Balance 

 
 0.0 

 
   1.1 

 
  7.0 

 
13.9 

 
18.5 

 
59.4 

Income  0.0    1.0 11.6 18.1 17.9 51.3 



110 Florida Tax Review  [Vol. 11:2  

 
 

FIGURE 1 
EDUCATION SAVINGS PLANS ACCOUNT BALANCES HELD BY INCOME 

GROUP51 
 

 
 
 

II. COLLEGE SAVINGS AS A USEFUL ELEMENT OF  
FEDERAL FINANCIAL AID 

 
There is no question that qualified tuition plans can be structured in a 

way that would make them less regressive, and thus a more tailored tool for 
lowering the cost of higher education for those most in need of the incentive. 
The question, though, is not how these plans can be structured to maximize 
the effectiveness of saving for college. Instead, the correct inquiry is whether 
a savings plan can ever be the most effective and efficient way for the 
government to spend a given amount of money to incentivize enrollment in 
higher education.  
  This part argues that students should decide whether to obtain higher 
education based on a fairly simple evaluation of the costs of education and 
the expected returns. Empirical evidence suggests that the returns from 
higher education are always greater than the costs and thus everyone, 
regardless of income, should make the decision to invest in higher education. 
This does not hold true, however, because low-to-moderate-income students 
systematically misestimate these values for a number of reasons, making 
them believe that higher education is not a worthwhile investment. The best 

                                                      
51. Id. 

59.4%18.5%

13.9%

7.0% 1.1% 0%

529 Account Balance Held by Each Income Percentile  
Range

95-100

90-95

75-90

50-75

25-50

0-25



2011] In Defense of College Savings Plans 111 
 
way to change that calculus and incentivize these students to enroll in higher 
education is to lower the upfront cost (or net price) of higher education by 
providing them with grants. A properly structured college savings plan can 
provide the same incentive as a grant but be of greater financial value at the 
time the college decision is made. 
 
A. Lower-Income Students Under-Invest in Higher Education  
 

The prevailing theory on the provision of financial aid generally 
stems from Gary Becker’s Human Capital Theory (HCT). This theory simply 
states that people use the same cost and return analysis in making decisions 
about training and education as they do in other areas of economic decision-
making. As applied to higher education, this means that so long as the 
investment in higher education will lead to more future income than it will 
cost, rational people will invest in higher education. In terms of the below 
graph, as long as area A (future income) is larger than areas B (forgone 
income) and C (tuition) then a rational person will make the investment. 
 

FIGURE 2 
STANDARD HUMAN CAPITAL THEORY52 

 

 
 

                                                      
52. Alex Usher, Grants for Students: What They Do, Why They Work, 

Educ. Policy Inst. 16 (2006), http://www.educationalpolicy.org/pdf/grantsforstudents 
.pdf. 



112 Florida Tax Review  [Vol. 11:2  

The most recent estimates show that college graduates earn roughly 
$900,000 more over a lifetime than those with only a high school diploma, 
have a far more stable work life, and enjoy a much more stable working 
career (Figure 4). The College Board estimates the present value of this 
amount to be around $450,000, well over the amount of loans the average 
person incurs to attend college.53 This number also exceeds the cost of 
borrowing to pay for four years of private education, including all living 
expenses.54  

 
 
 
  

                                                      
53. See Sandy Baum & Kathleen Payea, College Board, Education Pays:  

The Benefits of Higher Education for Individuals and Society, 11 (2004), 
http://www.collegeboard.com/prod_downloads/press/cost04/EducationPays2004.pdf. 

54. The most expensive college for this school year was Sarah Lawrence 
College with tuition plus room and board costing $54,410. See Campus Grotto, Most 
Expensive Colleges for 2009-2010, http://www.campusgrotto.com /most-expensive-
colleges-for-2009-2010.html (last visited Apr. 25, 2009). Even factoring in an 
additional $10,000 of living expenses, the entire four years would cost $257,640.  
The College Board estimates that average cost of a four-year private education will 
cost $244,571 in 2019, including tuition, fees, room, and board. See Minnesota 
Higher Educ. Servs. Office, Start Today.  Invest in a Child’s Tomorrow, at 1 (2003), 
http://529professor.com/pdfs/mn_eb.pdf. 



2011] In Defense of College Savings Plans 113 
 

FIGURE 3 
EARNINGS FOR FULL-TIME WORKERS BY EDUCATION ATTAINMENT 

OVER 40-YEAR WORK LIFE 55 
 
 

 
  

                                                      
55. U.S. Census Bureau, The Big Payoff:  Educational Attainment and 

Synthetic Estimates of Work-life Earnings, at 4 (2002), http://www.census.gov/ 
prod/2002pubs/p23-210.pdf 

$1 1.2
1.6

2.1
2.5

3.4

4.4

$0.0

$1.0

$2.0

$3.0

$4.0

$5.0

Work Life Earnings
(in millions of dollars)



114 Florida Tax Review  [Vol. 11:2  

FIGURE 4 
UNEMPLOYMENT RATE OF THE POPULATON 25 YEARS AND OVER BY 

EDUCATIONAL ATTAINMENT56 
 

 
 
 
The College Board further estimates that the average college 

graduate has earned enough to compensate for borrowing for full tuition and 
forgone income by the age of 33, as shown in Figure 5 below.57  

 
 

  

                                                      
56. U.S. Bureau of Labor Statistics, Education Pays 2009, available at 

http://www.bls.gov/emp/ep_chart_001.htm. 
57. See Baum & Ma, supra note 1, at 11.  

7.1%
4.4%

3.0% 2.2% 1.8% 1.4% 1.3%

0%

10%

20%

Unemployment Rate



2011] In Defense of College Savings Plans 115 
 

FIGURE 5 
CUMULATIVE EARNINGS RELATIVE TO COSTS OF EDUCATION, 

INCLUDING LOAN REPAYMENT AND FOREGONE INCOME 58 
 

 
 

Based on this data, it seems empirically clear that the return from 
investment in higher education (area C on the HCT graph) is larger than the 
cost of obtaining that education (areas A and B).  

The only hindrance from obtaining higher education thus ought to 
arise from cash-flow shortages; even if a person realizes that higher 
education is a good investment (and empirical evidence has clearly shown 
that it is in almost all cases), not everyone can afford to pay for it. This cash-
flow problem is most cheaply and easily solved through the availability of 
student loans.59 In other words, if student loans are readily available then 
there can be no cash-flow problem and thus no barrier to obtaining higher 
education.60  

                                                      
58. Id. 
59. This theory does not suggest that these loans should be subsidized 

through below-market interest rates or deductions for interest paid, only that they 
should be readily available.  

60. In theory, student loans are universally available because of the robust 
market for private student loans.  Private student loans do not have caps on the 
interest that can be charged and also are not dischargeable in bankruptcy, leading to 
very high approval rates for student borrowers.  See, e.g., Deanne Loonin & Alys 
Cohen, Nat’l Consumer L. Ctr., Paying the Price: The High Cost of Private Student 
Loans and the Dangers for Student Borrowers 12-14 (Mar. 2008), 
www.studentloanborrowerassistance.org/uploads/File/Report_PrivateLoans.pdf.   
 Analyzing how this affects the returns to education as illustrated in Figure 2 
is beyond the scope of this paper and should not have a dramatic impact given that 
only 14 percent of undergraduate students use private loans, and many of them could 
be using federal loan programs. The Project on Student Debt, Private Loans: Facts 
and Trends (Aug. 2009), http://projectonstudentdebt.org/files/pub/private_loan_ 
facts_trends_09.pdf. 



116 Florida Tax Review  [Vol. 11:2  

 Human capital theory and the estimated returns from higher 
education thus suggest that financial aid incentives should not be needed to 
induce students to obtain a higher education; this should apply equally to 
students of all income levels.61 This theory though does not square with 
reality. In fact, lower- and middle-income students attend college at much 
lower rates than high-income students: 90 percent of graduates from families 
earning $80,000 or more are attending college compared to only 60 percent 
of other graduates, and around 30 percent of those at the lowest income 
levels.62 Even among only high-achieving students, virtually all students 
from the top quarter of families in terms of income enroll in post-secondary 
education compared to only 75 percent of those in the lowest quartile.63 
Empirical evidence further shows that reducing the upfront costs of college 
has a dramatic impact on enrollment in lower-income students.64 This 
evidence thus suggests that lower-income students are systematically 
undervaluing the return from education and are not making the rational 
decision to invest in higher education.65 The behavioral economic theories of 
myopia and debt aversion appear to be the best explanations for this 
phenomenon. 
  

                                                      
61. Increasing the number or percentage of students obtaining a higher 

education should have no impact on the returns to college education, although early 
proponents of this theory suspected that it would. See Herbert L. Smith, 
Overeducation and Underemployment: An Agnostic Review, 59 Soc. Educ. 85, 97 
(1986) (quoting Finis Welch, an early proponent of Human Capital Theory, as 
saying that “one of the most important phenomena of our time is that rates of return 
to investments in schooling have failed to decline under the pressure of rapidly rising 
educational levels” and noting that “this observation is still relevant today”).  

62. Lawrence Gladieux, Low-Income Students and the Affordability of 
Higher Education, in America’s Untapped Resource: Low-Income Students in 
Higher Education 17, 20 (Richard D. Kahlenberg ed., 2004). 

63. High-achievement is measured based on high (“top”) standardized test 
scores. Baum & Ma, supra note 1, at 2. 

64. See Usher, supra note 52, at 23 (reviewing the research on price 
elasticity to changes in net price and finding that “[o]ne constant across all research 
findings is that grants/reductions in net price are much more effective among low-
income students than among middle- or high-income students”).  

65. See supra note 12. One other explanation is that lower-income people 
are more prone to over-estimating the costs of higher education, something that has 
been proven through empirical evidence. However, it seems that to the extent that 
that is the problem, the most efficient solution would be an information approach, 
not lowering the cost.  For that reason, I have left this explanation out of the 
analysis.   



2011] In Defense of College Savings Plans 117 
 

1. Low- and Moderate-Income Students May Have Lower Returns 
from Higher Education 

 
There is some evidence that those coming from lower-income 

families cannot expect to have these same high returns from investment in 
education. One study in Canada actually found that the rate of return to 
education was approximately 30 percent for the top quintile of university 
graduates, but was negative for the bottom quintile.66 If this is true, the 
returns to higher education for lower-income students might look more like 
the following: 

 
FIGURE 6 

HUMAN CAPITAL THEORY – LOWER RETURNS67 
 

 
 

However, this picture does not seem to hold in the U.S. The College 
Board found that higher education provides more than adequate returns to 
cover costs for all income levels, racial and ethnic groups, and both 
genders.68 Even assuming that the returns from education are lower for 
lower-income students (but still positive), the solution would be the same as 

                                                      
66. See Daniel Boothby & Geoff Rowe, Rate of Return to Education: A 

Distributional Analysis Using the LifePaths Model (Human Resources Development 
Canada, Working Paper No. W-02-8E, 2002), available at http://www.s3ri.soton.ac. 
uk/qmss/documents/RateofReturn_to_Education-DistributionalAnalysisusing_Life 
Paths.pdf. 

67. Author’s illustration. 
68. Baum & Ma, supra note 1, at 12.  



118 Florida Tax Review  [Vol. 11:2  

if those returns were simply misevaluated — lower the upfront costs in order 
to make the investment worthwhile and encourage enrollment.69 

  
2. Hyperbolic Discounting  

 
In calculating rates of return, each person must use his or her own 

discount rate — the rate at which he or she values money in the present more 
than money in the future. A higher discount rate makes one less likely to 
make an investment — someone with a higher discount rate values the 
returns on the investment less than someone with a lower discount rate, even 
if both accurately estimate what that return will be.  
 A number of empirical studies have shown that myopia, or the 
tendency to have a high discount rate, increases as income decreases.70 This 
outlook makes low- and moderate-income people less likely to make any 
investment where the returns come only in the future. A high discount rate is 
also related to the amount of uncertainty related to an investment. Since low- 
and middle-income students are far less likely to attend and complete college 
than higher-income students, this uncertainty means that the discount rate as 
related to education returns may be even higher than the already hyperbolic 
discount rate of lower-income people in general.71 

A degree in higher education takes a substantial amount of time to 
acquire. Further, the College Board estimates that it takes on average until 
the age of 33 for that particular investment to begin showing positive returns. 
That long time horizon, combined with the uncertainty of obtaining the 
degree and the returns from that degree once obtained, makes those with 
shorter temporal preferences less likely to pursue a degree in higher 
education. 

  
3. Debt Aversion  

 
Empirical evidence also suggests that lower-income students are 

more averse to debt than are higher-income students. This tendency is 
                                                      

69. Another issue with this line of argument is whether lower-income 
students could possibly know that their expected return may be lower than that of 
higher-income students. While some argue that lower-income students would not 
know this information and thus not include it in their calculus, others argue that 
lower-income students are well aware that their performance levels tend to be lower 
than those of more affluent students. Lower-income students could thus assume from 
that knowledge that they would experience lower returns because of lesser 
performance.  See supra Figure 2. Either way, the answer must be the same — lower 
the upfront costs to make the investment worthy.   

70. See, e.g., Gary S. Becker & Casey B. Mulligan, The Endogenous 
Determination of Time Preference, 112 Q.J. Econ. 729 (1997).  

71. Id. at 745 (income), 742 (certainty).  



2011] In Defense of College Savings Plans 119 
 
generally tied to the behavioral economic theory of loss aversion. Loss 
aversion means that a loss generates more disutility than a gain generates 
utility. Some studies have shown that a loss can generate twice the disutility 
that a gain can generate utility — in other words, the pain someone incurs 
from a $100 loss is more than the pleasure he gets from a $100 gain, and 
equal to the pleasure from a $200 gain.72 Loss aversion then is presumed to 
manifest itself as debt aversion because taking on debt is a constructive out-
of-pocket expense equivalent to a loss.73 Since this myopic loss aversion is 
most commonly associated with low-income people, many presume that low-
income students are also debt averse and thus under-invest in higher 
education because of a reluctance to take on debt in order to finance that 
education.  

The most sophisticated study on debt aversion stems from England 
and used multivariate analysis to attempt to disaggregate the view on debt 
from actual decisions. The study found a very significant relationship 
between debt and social class.74 
 Empirical evidence suggests that this debt aversion can be seen in 
lower-income students but is not entirely conclusive. In one study Tom 
Mortensen analyzed Federal Reserve data on perceptions about borrowing 
for higher education and found that low-income individuals were less 
inclined to borrow and concluded that loans were not a viable option for 
increasing participation of these students in higher education.75 More recent 
studies have found that the significant factor is not income, but race and 
ethnicity, determining that minority students are less likely to borrow to 
finance unmet need for all income levels.76 There is still a strong correlation 
between minority status and low-income status such that most of those 
averse to borrowing are low-income. 
  

                                                      
72. Gandhi, supra note 8, at 14.  
73. Id.  
74. See Claire Callender, Access to Higher Education in Britain: The 

Impact of Tuition Fees and Financial Assistance, in Cost-sharing and Accessibility 
in Higher Education: A Fairer Deal? 105, 126 (2005).  

75. See The Project on Student Debt, The Student Debt Dilemma: Debt 
Aversion as a Barrier to College Access 4, available at http://projectonstudentdebt. 
org/files/pub/DebtDilemma.pdf; Tom Mortensen, Attitudes of Americans Toward 
Borrowing to Finance Educational Expenses 1959-1983 14, 22 (ACT Student 
Financial Aid Research Report Series No. 88-2, 1988).  

76. Alisa F. Cunningham & Deborah A. Santiago, The Inst. for Higher 
Educ. Policy & Exelencia in Educ., Student Aversion to Borrowing: Who Borrows 
and Who Doesn’t 17, http://www.ihep.org/assets/files/publications/s-z/Student 
AversiontoBorrowing.pdf. 



120 Florida Tax Review  [Vol. 11:2  

B. Lower-Income Students Respond More to Grants Than Other Types 
 of Aid 
 

The failure of lower-income students to properly evaluate the cost of 
and returns from higher education suggests a particular role for grants in the 
financial aid system. There are two aspects of grants that distinguish them 
from other modes of financial aid: first, they are paid directly to the student 
(or to the institution on behalf of the student); second, they do not require 
repayment.  

 The unique aspect of grants in the financial aid system is that they 
reduce both out-of-pocket costs and “net price,” two factors that bear most 
heavily on the cost-benefit ratio when making the higher education decision. 
“Net price” or “net tuition” is the idea that the price of a year of higher 
education to the student is not the full price of tuition or the amount of 
money that will be paid by the student over time for tuition, but is actually 
the amount of tuition reduced by guaranteed payments made to, or on behalf 
of, the individual student.  

Grants are the only type of federally sponsored financial aid that 
immediately reduces both net price and out-of-pocket expenses. Tax 
incentives (such as deductions and credits) and subsidized loans both reduce 
the cost of higher education, but only in the future, thus not changing the 
immediate calculus. Tax incentives generally have a time lag because they 
are administered through the tax system rather than concurrently with 
provision of financial aid or the payment of tuition. Further, the credits are 
often taken by the parents of dependent students while it is the student 
making the cost and returns evaluation.77 In these instances, the credit cannot 
change that ratio.  

Grants alone can make what is perceived as an otherwise unworthy 
investment worthy by lowering the “net price” of higher education (areas A 
plus B in Figure 2) to less than the return (area C in Figure 2). This reduction 
in net price essentially offsets the hyperbolic discount rate or debt aversion 
so that even lower-income students subject to these behavioral economic 
biases will perceive the returns from higher education as exceeding these 
reduced costs.  

Empirical evidence supports the theory that grants most effectively 
correct for the irrational failure to invest in higher education. Studies into this 
phenomenon largely began after taking note of changes in patterns of 
financial aid and enrollment in the 1970s and 1980s. The 1970s were marked 

                                                      
77. See Jeffrey Taylor, Marcia B. Harris & Susan Taylor, Nat’l Ass’n of 

Colleges & Employers, Parents Have Their Say . . . About Their College-Age 
Children’s Career Decisions (2004) (finding that 91.8 percent of parents felt that 
they should be either neutral or have very little influence on their children’s college 
and career decisions).  



2011] In Defense of College Savings Plans 121 
 
by high amounts of direct government aid to low-income and minority 
students and were also a period of relatively high participation rates for those 
groups. The 1980s saw a marked decline in both indicators.78  

The important findings from these studies are that grants (or 
reductions in net price) increase enrollment, particularly for lower-income 
students, and that loan subsidization does not have the same effect.79 
 The fact that low-to-moderate-income students seem to have 
systematically higher discount rates and may be subject to debt aversion 
suggests that these students are less likely to attend higher education even 
when not credit-constrained. These biases influence these students to 
subjectively value higher education at less than its objective value. Therefore 
low-to-moderate-income students will under-invest in higher education 
unless they are given a subsidy — a grant — which increases this subjective 
rate of return. In other words, grants can have a much larger effect on the 
higher education choice of low-to-moderate-income students than on high-
income students who already perceive higher education to be a good 
investment.80  
 
C. 529 Plans Have the Same Effect as Grants  
 

Recall from above that the salient features of grants are that they do 
not require repayment and that they be paid directly to the student for 
educational purposes or to the institution on behalf of the student. The 
important point in time at which to look is when the college decision is 
made, the time when the costs of higher education are compared to the 
returns. At that time, the 529 plan must be paid toward education expenses 
and must be used for the particular designated beneficiary.81 This means that 
the 529 acts to reduce the net tuition on that day — it is an available amount 
of money that is guaranteed to be paid to the student to reduce the amount of 
tuition that needs to be funded from elsewhere.  For example, imagine two 
students, Anna and Ben, both deciding whether to attend school X. School X 
has tuition of $20,000 and both students receive a grant of $5,000. Anna, 
however, is the designated beneficiary of a 529 account worth $5,000. On the 
day that the college decision is made, the net price to Ben is $15,000 but the 

                                                      
78. Edward P. St. John, Refinancing the College Dream: Access, Equal 

Opportunity, and Justice for Taxpayers 100, 113 (Johns Hopkins University Press 
1997).  

79. Gandhi, supra note 8, at 23.  
80. See Usher, supra note 52, at 23 (making this same argument, but finding 

no debt aversion and relying largely on lower returns for lower-income students).  
81. This is what distinguishes 529 plans from general savings or income — 

the latter may or may not be used toward education and thus are not earmarked in the 
same way as 529 plans.   



122 Florida Tax Review  [Vol. 11:2  

net price to Anna is only $10,000 because the 529 account must be paid 
toward the tuition (or else the penalties will be incurred). The 529 plan then 
has the same effect of reducing net price as a grant and therefore offers the 
same efficient incentive to low-income taxpayers. 
 

 III. POLICY PROPOSALS TO MAXIMIZE THE IMPACT OF 529 PLANS  
 
A. Structure Plans to Induce Savings 
 

In order for 529 plans to be a more efficient incentive than outright 
grants, they must induce those taxpayers who can save for college to do so in 
order to increase the “grant” at the time the enrollment decision is made. For 
example, say the government has $500 to use toward each student to 
incentivize his or her college decision. The above analysis shows that the 
most efficient use of that $500 is to give it directly to the student to reduce 
net price at the time of enrollment. However, this sum of money would be a 
better incentive if the government could use it to encourage the family to 
pledge an additional $500 toward the student’s education.  If that can be 
done, the amount of the grant received by the student is no longer the $500 
of federal aid, but instead is $1,000.   

There are ways to structure 529 plans to increase participation and 
the amount of savings, especially by lower-income families.  On this point it 
is best to look at different ways that the states have experimented with 
increasing participation and the effectiveness of those programs.82  

One successful program involves opening an account in the name of 
all children with an initial contribution by the government to the fund. Maine 
currently offers such a program, the Harold Alfond College Challenge. The 
program, now in its initial phases, was funded by a large grant from a Maine 
philanthropist and provides a one-time grant (“scholarship”) into a Maine 
529 account in the name of any newborn resident of Maine. The paperwork 
required to open the account and receive the grant is reviewed with the 
mother along with other hospital paperwork before discharge after giving 
birth. It is also available on the internet and in government offices and the 
program is widely advertised on TV, radio, and in newspapers. In the first 
two months of the program roughly 1,000 people began the process to open 
the account.83 Given that there are around 14,000 births in Maine per year, 
this would indicate that almost half of the newborns in Maine will open an 

                                                      
82. See Appendix B for a summary of the state programs offering some sort 

of initial grant or matching grant for contributions.  
83. Press Release, Western Maine Health, Over 1,000 Families Have 

Requested Harold Alfond College Challenge Grant Information, http://www.wmhcc. 
org/wmh_body.cfm?id=5439. 



2011] In Defense of College Savings Plans 123 
 
account and receive the grant.84 In the national study only roughly 10 percent 
of respondents with children indicated having a 529 account.85 This is also a 
significant increase over previous participation; only 8,000 accounts had 
previously been opened in Maine since the program began in 1999. Those 
administering the program expect that number to increase as the program 
becomes more widely known. The program has not yet released information 
on effects by income level, but is a clear indication that this upfront grant can 
greatly increase participation. 

A more ideal program along these lines would open an account in 
the name of every child with this initial contribution, but only make the 
program available to children in families below a certain income threshold. 
This presents two administrative difficulties: first, determining the 
appropriate income level where the cut-off should be, and second, dealing 
with changing levels of income over the life of the child. The first has been 
given much attention by states experimenting with incentives that target only 
low- and moderate-income families. While none of these involve income 
limits for an initial contribution, most of the states offering a matching grant 
program have an income limit that attempts to track the line between 
moderate-income households in need of an incentive to save, and high-
income households without need for such an incentive. This income limit is 
generally between $50,000 and $80,000.86  

There is also evidence stemming from a number of programs that 
offering a matching grant increases both participation and the amount of 
saving by participants. Increased savings is the variable that makes these 
plans more effective than a traditional grant. One study focused on the Maine 
NextGen Matching Grant Program (before the introduction of the initial 
contribution, discussed above). The program was available to lower-income 
households, defined as those with income below $50,000 adjusted gross 
income (indexed to CPI). Of the 6,414,529 NexGen accounts in the state, 
1,335 had received at least one matching grant, meaning that they met this 
income threshold. This indicates that at least 20 percent of the accounts were 
held by those with household AGI below $50,000. This is a stark 
improvement over the national data discussed previously where those in the 
lower half of the income distribution held only 2.8 percent of the 529 

                                                      
84. Press Release, Fin. Auth. of Maine, Harold Alfond College Challenge 

Celebrates Six Months of Awards (Aug. 1, 2008), http://www.famemaine.com/ 
blog/post/harold_alfond_college_challeng.aspx. 

85. There were 7,560 total respondents and 750 indicated having a 529 
plan.  See supra Table 3. 

86. See Margaret Clancy, Lisa Reyes Mason & Soda Lo, Ctr. for Soc. Dev., 
Wash. Univ. St. Louis, State 529 Matching Grant Program Summary (2008), 
available at http://csd.wustl.edu/Publications/Documents /529_Summary.pdf. 



124 Florida Tax Review  [Vol. 11:2  

accounts;87 $50,000 is roughly the 55th percentile in terms of income 
distribution.88 The data further shows that at least 2 percent of the 
participants have household AGI of less than $20,000 compared to around .1 
percent in the national survey.89 

The results of the study on Maine’s Matching Grant Program also 
support the fact that lower- and moderate-income people are incentivized to 
save by the program. The report states that “the most important results of this 
study are the simple facts that low-to-moderate income individuals save in 
NextGen, and save through the Matching Grant Program,” and that “income 
level is not statistically associated with saving performance.”90 The study 
further finds that about 80 percent of those who receive a matching grant go 
on to receive continued matching contributions. The “very positive impact on 
low- and moderate-income families” eventually formed the basis for a 
proposal of a similar matching grant program in Missouri.91 

The fact that a matching grant program encourages participation as 
compared to deductions is also evidenced by a comparison of the Kansas and 
Louisiana 529 programs. The programs are similar in most respects, but at 
the time of the study Louisiana offered a progressive matching grant system 
(increasing the amount of the match as income decreased) while Kansas 
offered only the deduction. In Louisiana, participants in the 0-50 income 
percentiles held almost 23 percent of the total accounts. Participants of the 
same income level in Kansas held only about 8 percent of the total 529 
accounts.  

Both of these studies compare the use of matching grants to the use 
of deductions and find significantly increased participation among low-to-
moderate-income households with use of matching grants. While a 
refundable credit provides an economically equivalent benefit to a matching 
grant, empirical evidence has shown that people are more incentivized by the 
matching grant than the equivalent credit. The main study in this area found 
that participants gave around 20 percent more to charities when the incentive 
                                                      

87. Of the 750 accounts in the survey, twenty-one were held by those in the 
0-50 percentiles. See supra Table 3. 

88. Dep’t Treas., Analysis of Section 529, supra note 30, at 11 (noting less 
than 1 percent participation); Margaret Clancy, Chang-Keun Han, Lisa Reyes Mason 
& Michael Sherraden, Ctr. For Soc. Dev., Wash. Univ. St. Louis,  Inclusion in 
College Savings Plans: Participation and Saving in Maine’s Matching Grant 
Program 5 (2006), http://csd.wustl.edu/Publications/Documents/RP06-03.pdf.  

89. Only one of the 750 survey participants was in the 0-25 percentile, 
which corresponds to a maximum income of roughly $22,500.  

90. Clancy, Han, Mason & Sherraden, supra note 88, at 40. 
91. Press Release, George Warren Brown Sch. of Soc. Work, Wash. Univ. 

St. Louis, Center for Social Development’s Research Informs Missouri’s Legacy 
Initiative (Jan. 22, 2007), http://gwbweb.wustl.edu/newsroom/PressRelease/Pages/ 
MissouriTreasurerSarahSteelman.aspx. 



2011] In Defense of College Savings Plans 125 
 
was structured as a matching grant rather than a subsidy rebate.92 For this 
reason, the ideal program should use a matching grant in order to induce the 
most savings.93 
 
B. Reduce Regressivity  
 

The role of federal financial aid is to increase enrollment in higher 
education. Lower- and middle-income students are most responsive to and in 
need of such an incentive, and grants are most effectively and efficiently 
targeted at those groups. In order to achieve that goal through 529 plans, the 
plans need to be made not regressive but progressive means of federal 
financial aid. There are a number of simple changes that can and should be 
made to 529 plans to achieve such progressivity. Many of these proposals 
have been experimented with for state tax purposes but should be 
implemented for federal tax purposes as that is the primary benefit of the 529 
plans.  

First, there should be an overall contribution limit. This is probably 
most easily implemented by allowing only one account total (as opposed to 
one per state) per designated beneficiary. It may also be that the contribution 
limit, which now hovers around $300,000, should be lowered.  

Further, there should be an income limit which an account owner 
must fall below in order to participate in a 529 plan.94 Empirical evidence 
shows that low-income families are most incentivized by direct financial aid, 
but middle-income families can be responsive to such incentives as well. 
Middle-income families are also more able to save for higher education and 
take advantage of this particular incentive. Thus it is probably the case that 
direct financial aid, Pell Grants in particular, should remain focused on the 
lowest-income families while 529 plans may be a better vehicle for those 
who are somewhat better off but not high-income. 529 plans can still be 

                                                      
92. Catherine C. Eckel & Philip J. Grossman, Rebate Versus Matching: 

Does How we Subsidize Charitable Contributions Matter? 87 J. Pub. Econ. 681 
(2003).  

93. See infra note 96 for a discussion of the terminology of match vs. 
refund.  

94. Consistent with the existing 529 matching programs targeted at lower- 
and middle-income families, the eligibility should be means-tested based on the AGI 
of the child’s family for the previous one to three years.  Note that the relevant 
income is that of the child’s family and not that of the owner/donor of the account.  
This is a more accurate measure of the financial resources available to the child.  
Eligibility must be re-established every year.  



126 Florida Tax Review  [Vol. 11:2  

available to even the very lowest-income families but with recognition that 
they likely will not be able to make much use of them.95 

The most important change is that the contributions should be 
structured as a matching grant.96 Either a matching grant or a refundable 
credit must be used in order to prevent the tax incentive aspect of 529 plans 
from being regressive. An exclusion is always regressive because it is tied to 
the taxpayer’s tax bracket. Excluding an item of income of $1,000 is worth 
$350 to a wealthier taxpayer in the 35 percent tax bracket, but worth only 
$150 to a lower-income taxpayer in the 15 percent tax bracket.  

A non-refundable credit is not necessarily regressive but is limited 
by a taxpayer’s tax liability. If two taxpayers are both entitled to a $1,000 
deduction and one has tax liability of $3,000 while the other’s is only $800 
then the non-refundable credit will be worth the full $1,000 for the first 
(probably higher-income) taxpayer but worth only $800 to the second 
(probably lower-income) taxpayer, making the credit regressive.  

A refundable credit is not limited in this way. As to the $800 
taxpayer, he or she would be entitled to a $200 payment, or refund. The 
refundable credit is therefore worth $1,000 to both of the taxpayers. A 
matching grant would be a payment into the account equal to some 
percentage of the payments made by the taxpayer.97 Therefore both the 
refundable credit and the matching grant would solve the regressivity 
problem of the current exclusion.  

Given the tax-neutral choice between a matching grant and 
refundable credit, empirical evidence suggests that people will contribute 
more money when the incentive is a matching grant rather than a refundable 
credit, even where the economic impact is exactly the same. A credit would 
be implemented as follows: the taxpayer would pay $2 into the account and 
the government would remit $1 to the taxpayer. The account would have a 
$2 balance and the government and taxpayer would each have $1 less. A 
matching grant would have the following effect: the taxpayer would pay $1 
into the account and the government would match that with a $1 payment 
                                                      

95. It is probably also the case that we do not want to force these taxpayers 
to use 529s because we do not want to force the very limited funds of these families 
into one particular use.  

96. Match here is used to mean that the funds are deposited directly into the 
account (or linked account in the beneficiary’s name) even though theoretically 
either a refundable tax credit or a spending grant program could be structured either 
way (to be paid to the donor or the account).  For purposes of this paper, a “match” 
or “matching grant” is payment made to the account while a “refund” is a payment to 
the donor.  

97. In actuality, the matching grants, or contributions by the government, 
are almost universally held in a separate but linked account.  This allows the state or 
the plan to be the legal owner of the funds and eases administration and return in the 
event that the funds are not used for qualified expenses.  



2011] In Defense of College Savings Plans 127 
 
(depending on the match ratio). The account balance is again $2 and both the 
government and taxpayer have $1 less, the exact economic equivalent. 
Despite this equivalence, empirical evidence is replete with studies showing 
that taxpayers respond more to matching grants than credits, though these are 
not in the educational context.98 

  
C. Reduce Uncertainty  
 

Middle- and lower-income families may be dissuaded from taking 
advantage of the benefits of 529 plans, even as restructured, if the 
uncertainty surrounding the child’s college choice and financial aid remain 
too high. There are a number of changes that can be made that would reduce 
this uncertainty and allow these families to use these accounts in the event 
that their children do attend college.  

First, there needs to be a mechanism by which the family is not fully 
penalized if the student fails to attend college. Such a penalty could be an 
overriding deterrent for families where even high-achieving students are 
attending college only 75 percent of the time.99 Given that most of this 
decision is made well before the family knows anything about the child’s 
performance or desires (imagine the Maine program where the mother is 
asked to enroll in the program at the time she gives birth), this is a huge level 
of uncertainty. Current 529 plans do not deal with this uncertainty at all — 
the only way to withdraw funds in that scenario is to incur the penalty.100 
The portion of the Maine plans contributed by the state, which makes the 
initial contribution, simply reverts back to the state if the child does not 
                                                      

98. A real but easily corrected problem with a matching grant is churning.  
Churning results when a deposit is made into an account just long enough to receive 
the match and then both the grant and the match are immediately withdrawn.  The 
problem of churning is generally dealt with in one of two ways: either the grant is 
recaptured upon early withdrawal, or matches are disallowed during the calendar 
year (or two, etc.) in which a withdrawal is made from the account.  See, e.g., Notice 
of Hearing on Proposed Administrative Rulemaking, State of Kan., State Treasurer’s 
Office (2009), http://www.kansasstatetreasurer.com/prodweb/pdfs/hearing_notice_ 
09.pdf (including a discussion of this proposed change to the regulations). This is 
also the approach used with the Saver’s Credit.  IRC § 25B(d)(2)(A).  Since this 
latter anti-churning rule only extends the time period during which churning can 
occur, the recapture of the match option is the better one and should be adopted.  
This rule would essentially require that should the expenses be withdrawn for any 
non-qualified reason, the matches that had been deposited into the account would be 
reclaimed by the government.  

99. Supra note 63.  
100. As noted earlier, while it is possible that a high-income taxpayer may 

still be better off in this scenario than if they had saved in a taxable account, this is 
unlikely to ever be true for the lower-income taxpayers who will be eligible to use 
these 529 plans.  



128 Florida Tax Review  [Vol. 11:2  

enroll in higher education or to the extent not used for that purpose. It 
probably would not work to allow the family to withdraw and reclaim the 
funds with no penalty like the state can do. The problem would be that this 
could be used against the student — essentially the family could pressure the 
student not to attend college in order to regain use of those funds. In other 
words, in order for the family’s portion of the contribution to act as a grant, it 
must be earmarked for education. The loss of the state funds alone would not 
act as a deterrence or penalty as to the family’s portion.  

An option would be to allow the 529 plans to be rolled over into 
another such account for a different beneficiary, to the extent that that 
account was below the limit. In the case where there is no such account or all 
such accounts are at the contribution limits, the family could have the choice 
to withdraw the fund with penalty, or to roll it into another tax-preferred 
account such as a 401(k) or IDA. 

Another major cause of uncertainty that leads marginal taxpayers to 
avoid use of 529 plans is the effect that these assets will have on financial 
aid. Most low-to-moderate-income students will be eligible for federal 
financial aid, and accumulating assets can act to reduce that financial aid. As 
it currently stands, 529 plans count as an asset of the parent for purposes of 
determining the financial aid eligibility of the student. This means that 
roughly 5.6 percent of the value of the account is treated as being available to 
pay for college.101 In order to eliminate this disincentive to save, these plans 
should be excluded from consideration when determining need-based 
financial aid.  
 The last change that ought to be made to reduce the uncertainty is 
that withdrawals should be allowed without penalty to the extent that the 
student receives financial aid, either government or institutional, and does 

                                                      
101. The calculation of the 5.6 percent is a bit complicated.  All federal 

financial aid programs use the Free Application for Federal Student Aid (FAFSA).  
The FAFSA operates by calculating the “Expected Family Contribution,” or the 
amount that the family should reasonably be expected to pay toward the student’s 
college education. This in turn indicates to both the government and the institution 
the amount of need the student has (essentially the tuition at the institution minus the 
EFC). In calculating the EFC, the FAFSA determines the family’s Adjusted 
Available Income, which is the family income plus 12 percent of the “unprotected” 
assets (i.e., those that are not specifically protected like the family home and 
retirement accounts), of which a 529 plan is one. Once the AAI is calculated, the 
EFC is 47 percent of that amount. Thus in the end 47 percent of 12 percent of the 
529 is included in the EFC, or 5.6 percent. The FAFSA must be filled out each year 
and so this calculation will be repeated each year based on the value of whatever 
amount remains in the 529 plan. Note that these are maximum amounts, but they 
begin to apply at low levels of income, generally below $30,000. See U.S. Dep’t of 
Educ., Info. for Fin. Aid Profs., The EFC Formula, 2010-2011 19, http://www.ifap. 
ed.gov/efcformulaguide/attachments/111609EFCFormulaGuide20102011.pdf. 



2011] In Defense of College Savings Plans 129 
 
not need to use the funds from the 529 account. This would be only a slight 
expansion from the current rules which allow withdrawal without penalty 
when the student receives a scholarship and thus does not need to use the 529 
plan. The withdrawal would be subject to taxation but not the 10 percent 
penalty. One possibility would be that the amount no longer needed for 
education expenses would be returned to the state and withdrawn by the 
family penalty-free in proportion to overall contributions to the plan. 
Alternatively, the state may want its own contributions to be returned before 
the family withdrew any funds without penalty. 
 

IV. AN ILLUSTRATION  
 
 Building on the evidence taken from state experimentation and 
studies about matching grants, it seems that the best way to structure a 529 
plan to encourage participation is with an initial contribution by the 
government with no initial contribution requirement on the individual’s 
behalf. In order to induce additional contributions by the owner, the 
matching grant is the most effective option. Empirical evidence is fairly clear 
that a matching grant encourages greater contribution than a refundable 
credit.102 This empirical evidence further shows that at least a one-to-one 
match encourages more contribution than any lower level of match.103 
Therefore, the lowest level of match should be one-to-one.  

It is crucial that the program address the fact that low-income 
households are less likely to save (or save as much) in response to savings 
incentives, but are more likely to respond to any program that lowers the net 
price of higher education because they have the lowest enrollment rates. In 
order to compensate for those facts, the match should be higher for low-
income families and then reduced to one-to-one for higher-income families. 
Similarly, the initial contribution by the government should be higher for 
lower-income families and reduced as income increases until it is phased 
out.104 
 The federal tax benefits of 529 plans currently cost the government 
around $1 billion annually.105 Given this ideal structure of an initial 
contribution and at least one-to-one matching grant, this section will make a 
number of assumptions and attempt to estimate the impact that 529 plans 

                                                      
102. Eckel & Grossman, supra note 92.  
103. Id.  
104. This design is essentially a compromise between increasing the size of 

grants to the lowest-income students and offering a savings incentive that would be 
used more often by middle-income students’ families. This phasing out also prevents 
the cliff effect and impact on marginal rates when the taxpayer first becomes 
ineligible.  

105. Staff of the Joint Comm. on Tax’n, supra note 8.   



130 Florida Tax Review  [Vol. 11:2  

could have on college enrollment among low- and moderate-income groups 
given a reasonable scenario.  
 According to the U.S. Census Bureau, in 2006 there were 
approximately 4.2 million births in the U.S.106 Additional data (see 
Appendix A) shows that about 52 percent of these births were in households 
with an income below $50,000 per year.107 That means that around 2.2 
million children were born into households below that level of income in 
2006; approximately 550,000 to households with income between $35,000 
and $50,000 and 1.63 million to households with an income below $34,999.  

Consistent with the federal financial aid plan and the existing 
matching programs for 529 plans, this illustration will treat a household with 
income below $35,000 as low-income, and those with income between 
$35,000 and $50,000 as moderate-income. In order to offer the best incentive 
and increase enrollment rates the most, an initial contribution of $300 and a 
later match with a three-to-one ratio will be given to students from 
households of low-income.108 For those households of moderate income, the 
beneficiary will receive an initial grant of $100 and one-to-one matches for 
later contributions.109 

Looking at the existing plans and participation rates suggests that 
participation might be somewhere in the 30 percent range. Recall that the 
Louisiana Matching Grant Program had about 20 percent participation for 
these income ranges, without an initial contribution from the government. 
The Maine plan may have had up to a 50 percent participation rate, but that 
was not controlled for income. It makes sense then to assume that the 
participation rate when limited to this income would be above that seen in 
Louisiana (with no initial grant) but below that observed in Maine.  

If the participation rate were around 30 percent, and assuming it is 
equal for households of low- and moderate-income, then participation could 
be expected for about 489,000 children of low-income and 165,000 of 
                                                      

106. U.S. Census Bureau, Statistical Abstract of the United States: 2011 tbl. 
78, Live Births, Deaths, Marriages, and Divorces: 1960 to 2007, available at 
http://www.census.gov/compendia/statab/2011/tables/11s0078.pdf. 

107. Jane Lawler Dye, U.S. Census Bureau, Fertility of American Women: 
2006, 6 tbl. 3, available at http://www. census.gov/prod/2008pubs/p20-558.pdf.  

108. There should be no requirement of initial (or any) contribution by the 
family.  The earlier Maine program as well as the OK SEED program both saw large 
increases in participation when this requirement was eliminated and in interviews 
most account owners indicated that the lack of upfront requirement was a decisive 
factor in opening an account. See Lisa Reyes Mason, Margaret Clancy, Margaret 
Sherraden & Chang-Keun Han, Ctr. for Soc. Dev., Wash. Univ. St. Louis, Saving for 
College in Maine’s Matching Grant Program: Account Owner Experiences 13 
(2006), http://csd.wustl.edu/ Publications/Documents/RP06-04.pdf. 

109. As discussed above, the determination of eligibility should be based on 
the average AGI of the previous three years of the family of the beneficiary.  



2011] In Defense of College Savings Plans 131 
 
moderate income. Given these participation numbers, the $1 billion could 
fund almost $1,530 per plan initially if it were all used for these upfront 
grants.  

Some of the funds, however, should be reserved for matching grants 
to encourage additional investment by the owner of the plan. Recall that in a 
study of Maine’s matching grant program, 80 percent of participants who 
received a first match continued to make contributions to receive subsequent 
matches; this data was using only low-to-middle-income taxpayers.110 
Assuming that this rate would be somewhat lower for those only of low-
income, because of the difficulty of saving any money, it seems reasonable 
to estimate roughly 60 percent of low-income households will continue to 
participate. The 80 percent observed in the Maine program will be assumed 
to be the continued participation rate among moderate-income households. 
This would mean that 18 percent of low-income families (60 percent of the 
original 30 percent) and 24 percent of moderate-income families (80 percent 
of the original 30 percent) of those in the targeted income group could be 
expected to maximize the matching portion of the account.  

Assuming these participation rates, a roughly $1 billion budget, 
and initial grants of $300 and $100 respectively, the government could 
offer the following (very basic) program: low-income households would 
receive the $300 initial grant and a three-to-one match of up to $150 per 
year; moderate-income families would receive the $100 initial grant and a 
one-to-one match of up to $50 per year.111 If the account owner (or 
friends, family, etc.) contributed the full matching amount of $50 per year, 
and the account grew at a rate of 6 percent annually and had earnings 

                                                      
110. The matching amount for that program was $100 per year at the 

beginning of the study and $200 per year by the end.  Mason, Clancy, Sherraden & 
Han, supra note 108, at 2. 

111. Creating and incentivizing the use of these accounts will certainly raise 
a number of administrative issues, many of which have already been dealt with by 
the states experimenting in this area or discussed previously in this paper. As noted 
several times, the contributions from the family/donor and the matches from the state 
will almost certainly be kept in different accounts to allow for separate ownership 
until the point of distribution. The funds can then be paid directly to the institution or 
paid as reimbursement upon proof of spending for qualified expenses. The fact that 
all of these 529 accounts are held as part of a larger plan of the state allows for 
administration of even the smallest accounts. Providers therefore do not need to 
avoid small, unprofitable accounts because these have been shown to be adequately 
supported by the larger accounts of the plan. See Margaret Clancy, Peter Orszag  & 
Michael Sherraden, Ctr. for Soc. Dev., Wash. Univ. St Louis, College Savings Plans: 
A Platform for Inclusive Savings Policy? (2004), http://www.cfsinnovation. 
com/system/files/imported/managed_documents/clancy_et_al_2004.pdf. 



132 Florida Tax Review  [Vol. 11:2  

taxed at a 20 percent rate, then on the day the child made his or her 
college decision, the plan for each household would look as follows:112  
 

TABLE 4 
ACCOUNT FOR FULL PARTICIPATING ACCOUNT FOR FULL 

PARTICIPATING MODERATE-INCOME HOUSEHOLD  
LOW-INCOME HOUSEHOLD 

 
  Family     Family 
 Gov Contri- Account   Gov Contri- Account 
Year Match bution Balance            Year Match bution Balance 
0 $100    $100    0 $300    $300  

1 $50  $50  $205    1 $150  $50  $514  

2 $50  $50  $315    2 $150  $50  $739  

3 $50  $50  $430    3 $150  $50  $975  

4 $50  $50  $550    4 $150  $50  $1,221  

5 $50  $50  $677    5 $150  $50  $1,480  

6 $50  $50  $809    6 $150  $50  $1,751  

7 $50  $50  $948    7 $150  $50  $2,035  

8 $50  $50  $1,094    8 $150  $50  $2,333  

9 $50  $50  $1,246    9 $150  $50  $2,645  

10 $50  $50  $1,406    10 $150  $50  $2,972  

11 $50  $50  $1,573    11 $150  $50  $3,314  

12 $50  $50  $1,749    12 $150  $50  $3,673  

13 $50  $50  $1,933    13 $150  $50  $4,050  

14 $50  $50  $2,126    14 $150  $50  $4,444  

15 $50  $50  $2,328    15 $150  $50  $4,857  

16 $50  $50  $2,539    16 $150  $50  $5,291  

17 $50  $50  $2,761    17 $150  $50  $5,745  

                                                      
112. This does not assume the current benefit of deferral but instead the 

account balance reflects a tax on the earning of 20 percent. The benefit of the 
accounts as presented here is not deferral but instead is the match. Thus, an 
account’s earnings should be taxable to the owner of the account, which in most 
cases is the parent. The tax (as assumed here) should be able to be paid out of the 
account balance in order to ease what might otherwise be a liquidity problem. Note 
too that many of these taxpayers will almost certainly be in the zero bracket. Also, 
recall from above that the contributions from the state and federal government likely 
should and would be kept in a separate but linked account, in the beneficiary’s name 
but technically owned by the plan or the state.  



2011] In Defense of College Savings Plans 133 
 
 While the balance of these accounts is not insignificant, the impact is 
greatly increased if the account also gets the benefit of a state matching 
program as well. As an illustration, assume that this is a plan in Arkansas 
which provides a two-to-one match for low-income households and a one-to-
one match for moderate-income households, with a maximum $500 match 
per year.113 
 

TABLE 5 
ACCOUNT FOR FULL PARTICIPATING ACCOUNT FOR FULL 

PARTICIPATING MODERATE-INCOME HOUSEHOLD, LOW-INCOME 
HOUSEHOLD, INCLUDING  STATE MATCH INCLUDING STATE MATCH 

 

Year  
Gov 
Match 

State 
Match 

Family 
Contri-
bution  

Account 
Balance   Year  

Gov 
Match 

State 
Match 

Family 
Contri-
bution  

Account 
Balance 

0 $100      $100    0 $300      $300  

1 $50  $50  $50  $255    1 $150  $100  $50  $614  

2 $50  $50  $50  $417    2 $150  $100  $50  $944  

3 $50  $50  $50  $587    3 $150  $100  $50  $1,289  

4 $50  $50  $50  $765    4 $150  $100  $50  $1,651  

5 $50  $50  $50  $952    5 $150  $100  $50  $2,030  

6 $50  $50  $50  $1,148    6 $150  $100  $50  $2,428  

7 $50  $50  $50  $1,353    7 $150  $100  $50  $2,844  

8 $50  $50  $50  $1,568    8 $150  $100  $50  $3,281  

9 $50  $50  $50  $1,793    9 $150  $100  $50  $3,738  

10 $50  $50  $50  $2,029    10 $150  $100  $50  $4,218  

11 $50  $50  $50  $2,276    11 $150  $100  $50  $4,720  

12 $50  $50  $50  $2,536    12 $150  $100  $50  $5,247  

13 $50  $50  $50  $2,807    13 $150  $100  $50  $5,799  

14 $50  $50  $50  $3,092    14 $150  $100  $50  $6,377  

15 $50  $50  $50  $3,391    15 $150  $100  $50  $6,983  

16 $50  $50  $50  $3,703    16 $150  $100  $50  $7,618  

17 $50  $50  $50  $4,031    17 $150  $100  $50  $8,284  
 

                                                      
113. Assume for the time being that the account owner still only contributes 

up to the level of the federal match.  



134 Florida Tax Review  [Vol. 11:2  

 A significant number of studies have found that an increase in direct 
aid leads to an increased enrollment of roughly 3.6 to 4 percentage points, or 
correspondingly, that an increase in net price leads to decreased enrollment 
of the same level.114 One such study looked at the elimination of the Social 
Security Student Benefits Program, which provided aid to students who had 
suffered the death of a parent. This study found, consistent with previous 
studies, that $1,000 in grant aid leads to an increase in enrollment of 3.6 
percentage points.115  

Since the aid provided by these 529 accounts will be almost entirely 
redirected from higher-income students to these low- and moderate-income 
students, almost all, if not all, of this amount should be treated as an increase 
in aid. Even assuming that only $7,500 (low-income) and $3,500 (moderate-
income) represent additional aid, and assuming the 18 percent (low-income) 
and 24 percent (moderate-income) participation rates, this basic scenario 
could lead to an increase in enrollment of up to 4.8 percentage points (low) 
and 3 percentage points for these income classes, increasing participation 
from around 40 percent to as high as 44 percent.116  

There is also reason to believe that these classes of students are more 
price-elastic than higher-income students, and the Dynarski study on the 
impact of aid did not address differential impact based on income. To the 
extent that that is true, the increase in enrollment may be toward the higher 
end of the spectrum (4 percent) since it is targeted only toward low- and 
moderate-income students. This would represent increased enrollment of 3.6 

                                                      
114. Susan Dynarski, Does Aid Matter? Measuring the Effect of Student 

Aid on College Attendance and Completion 16 (John F. Kennedy Sch. of Gov’t, 
Faculty Research Working Paper Series, No. RWP01-034, 2001), available at 
http://web.hks.harvard.edu/publications/getFile.aspx?Id=21. See also Larry L. Leslie 
& Paul T. Brinkman, The Economic Value of Higher Education 124-25, 155 (1988) 
(stating that a $1,000 increase in net price decreases attendance by 3-5 percent); 
Charles F. Manski & David A. Wise, College Choice in America 119-28, 123 tbl.7.4 
(1983) ($1,000 in aid increases enrollment by 3.8 percent); Thomas J. Kane, College 
Entry by Blacks since 1970: The Role of College Costs, Family Background, and the 
Returns to Education, 102 J. Pol. Econ. 878, 892-93 tbls.3 & 4 (1994) (finding that a 
$1,000 increase in price decreases enrollment by 3.7 percent).  

115. Dynarski, supra note 114, at 16. 
116. This assumes a linear impact of the aid such that the third $1,000 has 

the same impact as the first, an assumption also made in these studies.  Thus, this 
number is calculated as 3.6 percent (7.5) =  27 percent (.18) = 4.86 percent, and for 
moderate-income students 3.6 percent (3.5) = 12.6 percent  (.24) = 3.02 percent. 
Note also that this should not have an impact on college prices — when Pell Grants 
were introduced many argued that the increase in aid would just be directly offset 
with an increase in tuition.  This never materialized and is not suspected to be 
problem. See id.  



2011] In Defense of College Savings Plans 135 
 
percentage points for low-income students and 2.7 percentage points for 
moderate-income students. 

  
CONCLUSION  

 
Federal financial aid has recently moved away from focusing on 

giving direct aid to low- and moderate-income students and at the same time 
has seen a stalling of enrollment rates. There is also a continued and wide 
gap between the enrollment rates of lower-income students and higher-
income students. This paper has shown that the most effective way to 
increase the enrollment rates among low-to-moderate-income students is to 
decrease net price by offering a grant at the time of enrollment. A 529 plan 
can not only affect this same decrease in net price and thus have the same 
incentive, but it can also induce families to save on their own — thus giving 
the student a grant in the amount not only of the government’s share, but of 
the family’s contributed share as well. This can be accomplished in a 
revenue-neutral manner by eliminating the current tax benefits on 529 
accounts, thus giving the government $1 billion annually for the program. 
The government can thus offer an initial grant into a 529 account for each 
child born into a family with low-to-moderate income, as well as annual 
matching grants, both dependent on income. Based on evaluations of similar 
current programs, this could lead to an increase in enrollment among these 
students of almost 5 percentage points, a number which could be much larger 
if expanded beyond this revenue-neutral analysis. This increase would be 
more than any seen in the previous two decades and was achieved using only 
$1 billion of federal funds already allocated to 529 plans — a paltry sum 
compared to either the $180 billion of total federal financial aid or the $35 
billion of tax incentives for education. 
 
 
  



136 Florida Tax Review  [Vol. 11:2  

APPENDIX A ― BIRTHS BY INCOME FOR 2006 
 

Income 
Minimum 

Income 
Maximum 

Number of 
women 

Births per 
1,000 

Women Births % 

0 $10,000  
           

15,889,965  33.2 
         

527,547  12.61% 

$10,000  $14,999  
             

2,308,705  89.5 
         

206,629  4.94% 

$15,000  $24,999  
             

5,324,892  85.2 
         

453,681  10.85% 

$25,000  $34,999  
             

5,864,945  74.7 
         

438,111  10.47% 

$35,000  $49,999  
             

8,599,979  64.2 
         

552,119  13.20% 

$50,000  $74,999  
           

13,215,740  58.9 
         

778,407  18.61% 

$75,000  $99,999  
             

9,647,862  51.9 
         

500,724  11.97% 

$100,000  $149,999  
             

9,444,088  48.1 
         

454,261  10.86% 

$150,000  $199,999  
             

3,105,652  46.7 
         

145,034  3.47% 

 $200,000+  
 

             
2,770,679  45.7 

         
126,620  3.03% 

Total 
 

           
76,172,507  

 

      
4,183,133  100.00% 

Source: Jane Lawler Dye, U.S. Census Bureau, Fertility of American Women: 2006, 
6, available at http://www.census.gov/ prod/2008pubs/p20-558.pdf. 
 

 Source: Jane Lawler Dye, U.S. Census Bureau, Fertility of American Women: 2006, 
6, available at http://www.census.gov/ prod/2008pubs/p20-558.pdf. 

28.4

23.6718.61

22.83

6.5

Births By Income 

0-$24,999

$25,000-$49,999

$50,000-$74,999

$75,000-$149,999

$150,000+



2011] In Defense of College Savings Plans 137 
 

APPENDIX B ― STATE PLANS 

State 
 

Name of 
Plan 

Eligibility Match 
Criteria 

Funding Procedure 
for 
Granting 
Match 

Distribution 

Arkansas GIFT 
College 
Investing 
Plan  

Household 
Income below 
$60,000 

For income 
below 
$30,000, the 
match is 2 to 1.  
For income 
between 
$30,000 and 
$60,000 the 
match is 1 to 1.  
Both are 
capped at $500 
per year.  

State 
appropriation 
of $250,000 
for a pilot 
program.  

Match 
contributed 
to account 
opened by 
the owner.  

 

Colorado  Direct 
Portfolio 
College 
Savings 
Plan, 
Scholars 
Choice 
College 
Savings 
Program, 
Stable 
Value Plus 
College 
Savings 
Program 

Household 
income up to 
200% of the 
federal poverty 
line, and 
beneficiary 
under the age 
of 13 at the 
time of 
application.   

1 to 1 match 
for 
contributions 
up to $500 per 
year and 
limited to 5 
years of 
matches.  

Annual 
budgeting by 
the state.  

Match goes 
into separate 
account in 
the 
beneficiary's 
name that is 
owned by 
CollegeInve
st (the 
investment 
managemen
t company).  

Paid directly 
to institution, 
matches 
revoked if no 
qualified 
withdrawals 
by age 22.  

Kansas Learning 
Quest 

Household 
income below 
200% of the 
federal poverty 
level and 
available to 
first 300 
applicants in 
each of the 
four 
Congressional 
districts.  

1 to 1 match 
for 
contributions 
above $100 
and up to $600 
per year.  

Pilot program 
funded for 
three years by 
the state.  

Match goes 
into separate 
account in 
the 
beneficiary's 
name but 
tied to the 
same 
investment 
portfolio.  

Paid directly 
to institution 
or to 
beneficiary 
with proof for 
reimburseme
nt of qualified 
expenses.  

Louisiana START 
Saving 
Program 

All residents 
are eligible but 
the match is 
progressive 
based on 
household 

Progressive 
match based 
on income 
which ranges 
from 2% to 
14%.  

Subject to 
yearly 
appropriation
s.  

Credited 
directly to 
the 
accounthold
er.  

State recovers 
match and 
earnings 
accrued if 
withdrawal 
for non-



138 Florida Tax Review  [Vol. 11:2  

income, with 
the highest 
match (14%) 
available to 
those with 
income up to 
$29,999.  

qualified 
purpose.  

Maine NextGen 
College 
Investing 
Plan 

Families with 
AGI of 
$75,000 or 
less.  

State provides 
a $200 initial 
grant on 
accounts 
funded with at 
least $50.  The 
Harold Alfond 
Foundation 
makes 
available a 
$500 initial 
grant to ALL 
babies to open 
such an 
account, 
regardless of 
income.  (The 
annual 
contribution 
match was not 
renewed for 
2010).  

Funded by 
the user fees 
that are paid 
by non-
resident 
accountholde
rs.  

Matches go 
to separate 
account in 
beneficiary’
s name but 
owned and 
invested by 
FAME.  

Application 
process to 
apply for use 
of the funds.  

Michigan 
Discontinued 
for 2009-2010  

Michigan 
Educaiton 
Savings 
Program 

Available to 
those with AGI 
up to $80,000 
and beneficiary 
under the age 
of seven.  

One time 
matching grant 
of $1 for every 
$3 contributed, 
capped at a 
total match of 
$200.  

State 
appropriation 
from tobacco 
settlement 
fund.  

Matches go 
to separate 
account 
owned by 
the Savings 
Program 
and invested 
in 
institutional 
bonds.  

Paid directly 
to the 
institution 
and recovered 
by the state if 
not used by 
the age of 30 
or if no 
longer needed 
by 
beneficiary.  

Minnesota Minnesota 
College 
Savings 
Plan 

AGI up to 
$80,000 and 
must 
contribute 
$200 per year.  

15% match for 
those with 
income up to 
$50,000, 10 % 
match for 
those with 
income 
between 
$50,000 and 
$80,000.  Both 

Funded by 
annual 
appropriation
s where 
matches 
reduced 
proportionatel
y when not 
sufficient to 
cover all.  

Matches go 
to separate 
account 
owned by 
the state and 
invested in 
guaranteed 
return fund.  

No 
distributions 
can be made 
until the 
account has 
been open for 
3 years.  



2011] In Defense of College Savings Plans 139 
 

are capped at 
$400 per year.  

Nebraska College 
Savings 
Plan of 
Nebraska 

Available to 
residents and 
non-residents 
attending a 
Nebraska 
university. 

Can apply for 
additional 
contribution 
from a private 
fund.  

Privately-
funded 
endowment.  

Matches go 
to the 
account.  

 

Nevada The 
Upromise 
College 
Fund 

Prior-year AGI 
of $61,950 or 
less and 
Nevada 
resident.  

$300 annual 
matching 
contribution 
with a $1,500 
lifetime limit.  

   

New Jersey Any NJ 
College 
Savings 
Plan 

Eligible to 
anyone with a 
NJ College 
Savings plan 
with 
contributions 
requirements 
(not yet set).  

$1,500 
scholarship at 
NJ college for 
first semester if 
have met the 
contribution 
requirements.  

State 
appropriation
s.  

Directly to 
institution.  

 

North Dakota College 
SAVE 

ND residents 
with incomes 
below $80,000 
(joint) or 
$40,000 
(single) and 
beneficiary 
under the age 
of 12. 

One time 1 to 
1 match of up 
to $300.  

Funded by 
user fees.  

Matches go 
to separate 
account 
owned by 
Bank of 
North 
Dakota.  

Payment sent 
directly to 
institution. 

Oklahoma Oklahoma 
College 
Savings 
Plan 

Randomly 
selected 
children born 
in the state in 
2007.  

Received an 
initial $1,000 
contribution 
and are eligible 
for a .5 to 1 
($125 total) or 
1 to 1 ($250 
total) match 
based on 
income.  

Part of a 
study called 
SEED OK 
funded by the 
Ford 
Foundation.  

Matches and 
initial 
contribution 
go to 
beneficiary's 
account.  

 

Rhode Island CollegeBou
ndFund 

Eligibility 
based on 
previous-year 
AGI but 
income limits 
not yet set.  

1 to 1 match 
up to $500 
annually.  

Funded by 
national user 
fees (and 
reduced 
proportionall
y preserving 
1 to 1 match 
where 
necessary).  

Matches go 
to separate 
account in 
beneficiary's 
name and 
owned and 
invested by 
CollegeBou
nd. 

Sent directly 
to institution 
and will be 
revoked if not 
used within 
"reasonable" 
time of 
becoming 
eligible to 
withdraw.  



140 Florida Tax Review  [Vol. 11:2  

 
Sources: SavingforCollege.com, Compare 529 Plans, http://www.savingforcollege. 
com/compare_529_plans/index.php?plan_question_ids[]=438&mode=Compare&pa
ge=compare_plan_questions&plan_type_id= (last visited Jan. 24, 2011). 
 
Margaret Clancy, Lisa Reyes Mason, & Soda Lo, Ctr. for Soc. Dev., Wash. Univ. St. 
Louis, State 529 Matching Grant Program Summary (2008), available at 
http://csd.wustl.edu/Publications/Documents/529_Summary.pdf. 
 
 
 
 

Utah Utah 
Education 
Savings 
Plan 

Up to 200% of 
federal poverty 
level or 
eligible for 
TANF, AND 
must commit 
to saving $25 
per month in 
account.  

1 to 1 match 
up to $300 per 
year for a 
maximum of 
four years.  

Pilot program 
funded by the 
state.  

Matches go 
to separate 
account 
owned by 
the Plan and 
in the 
beneficiary's 
name.  

Paid directly 
to instution.  


	Introduction
	I. Federal Financial Aid Scheme
	A. Background
	B. Current 529 Plans
	1. Structure
	2. Previous Evaluations of Effectiveness


	II. College Savings as a Useful Element of
	Federal Financial Aid
	A. Lower-Income Students Under-Invest in Higher Education
	1. Low- and Moderate-Income Students May Have Lower Returns from Higher Education
	2. Hyperbolic Discounting
	3. Debt Aversion

	B. Lower-Income Students Respond More to Grants Than Other Types  of Aid
	C. 529 Plans Have the Same Effect as Grants

	III. Policy Proposals to Maximize the Impact of 529 Plans
	IV. An Illustration
	Conclusion