The Economic Value of the Rutgers Future Scholars Program Clive

The Economic Value of the Rutgers Future Scholars Program
Clive R Belfield
[email protected]
Queens College, City University of New York
Flushing NY 11367
March 2010
1. Introduction
Research evidence on the advantages of education is copious and compelling (Haveman & Wolfe, 1984;
Belfield & Levin, 2007). Persons with more education have higher earnings, better health, are more
economically independent, and less likely to be involved in the criminal justice system. These private
benefits yield gains to taxpayers as well: tax revenues are increased and government spending on
health, criminal justice, and welfare is reduced. In addition, there are social benefits from education:
firms have access to a more skilled workforce; and there are fewer victims of crime. Importantly, the
benefits of education are particularly powerful for economically disadvantaged students: those at the
margin of dropping out of school or struggling in college; those residing in neighborhoods with highpoverty or with few educational opportunities; and those who do not have the financial resources or
family supports to allow them to invest in their own education (Levin et al., 2005).
Programs that raise high school graduation rates and college progression rates may therefore be
viewed as public investments: helping students to reach higher levels of attainment will not only benefit
them but also the local economy and local taxpayers. At issue is the economic magnitude of these
benefits and what programs should be put in place to raise educational opportunities for students in the
lowest quintiles of socioeconomic status.
This paper calculates the potential economic value of the Rutgers Future Scholars Program
(RFSP). A total of 200 promising 8th grade students from Rutgers host cities of Camden, New Brunswick,
Newark, and Piscataway, New Jersey, are chosen to participate RFSP. The programmatic goal is: to
increase the numbers of academically ambitious high school graduates who are from low-income
backgrounds; to help these students get admitted to college; and then to provide free tuition for those
admitted to Rutgers University. Thus, the program is directed at students who are talented and
motivated, yet at-risk due to the level of economic disadvantage their families currently face. Such a
program might therefore be anticipated to yield high returns to the participants, taxpayers, and society.
At this stage, our calculation of the economic value of the RFSP is necessarily ex ante: as of early 2010,
the RFSP has only just begun and so it is only possible to predict – based on the best available evidence –
its expected consequences. Nevertheless, the extent of the research is such that it is possible to predict
what will happen if the RFSP is successful in its goals of raising attainment for disadvantaged students.
Our analysis is structured as follows. First, based on the projected outcomes for those living in
severe financial disadvantage, as is the case for Rutgers Future Scholars, we predict their education
levels with and without the program. Next, we calculate the anticipated benefits over the lifetime from
participation in RFSP. These calculations are performed from the perspective of the private individual
1
and the taxpayer in New Jersey. We then aggregate these benefits into a single present value at 8th
grade. This amount may be interpreted as the economic value of the program and should be compared
to the cost of implementing it. To check the robustness of this economic value three separate models
are estimated: a lower bound, a conservative model, and a likely model. Following convention, results
from the conservative model are described in detail. Finally, we offer conclusions about the interpreting
these economic values.
2. The Rutgers Future Scholars Program
Currently, the RFSP has two cohorts of participants who began in 8th grade and are anticipated to
graduate from Rutgers, or another school of higher education by 2017-18. Each cohort has
approximately 200 students who have been identified as academically ambitious. The participants are
evenly split male/female and are almost all minority students (48% Hispanic, 40% African American, and
12% other). The scholars are drawn from a range of schools across four districts in New Jersey: Camden,
Newark, New Brunswick, and Piscataway. These districts report high school proficiency levels
significantly below the average for New Jersey in 2008-09. The proportions of students scoring at least
proficient in Language, Arts and Literacy is 84% for New Jersey; for the four districts, the proportions are
73% (Piscataway); 69% (New Brunswick); 55% (Newark); and 54% (Camden). For high school math, the
proportions are 73% (New Jersey); 76% (Piscataway); 59% (New Brunswick); 43% (Newark); and 24%
(Camden). Therefore, the RFSP students are academically ambitious but are typically being educated in
a school environment where the average academic standard is relatively low.
The RFSP incorporates a comprehensive range of educational and informational services to help the
participants meet program goals. These include: summer academies held on campus; academic,
social/cultural, and career development workshops; tutoring and enrichment; exposure to athletic and
cultural events; educational and study materials; internships, job-shadowing, and job-placement
assistance; and mentoring. These supports are intended to help the participants in multiple ways, but
the focus here is on how the program raises educational attainment.
To calculate the economic value of the RFSP it is necessary to predict how attainment levels
would increase for participants, i.e. how many more students would complete high school and progress
through to attend and complete college, and to compare this increase to the current attainment levels
of students in the four districts. High school graduation rates vary significantly depending on the source,
with state administrative rates typically much higher than those found by academic researchers. The
best estimates for New Jersey calculate that the state’s high school graduation rate is 85%, which is one
2
of the highest in the country (Swanson, 2004). However, the graduation rates across the RFSP districts
are much lower on average, at: 86%, Piscataway; 74%, New Brunswick; 68% in Newark; and 55% in
Camden.1 Thus, students in the four districts have approximately a 70% probability of graduation from
high school. Given income levels in the four districts and the link between incomes and college
attendance, it is possible to predict terminal educational attainment for these students. Specifically,
students in these districts are estimated to have attainment at rates of: 30% high school dropout; 45%
high school graduate; 7% some college; 7% Associate degree; and 11% Bachelor degree.
If the RFSP is successful, it is expected that more students will graduate from high school and
more will progress on to college. For RFSP participants, two progression rates are plausible. The first is
that participants have attainment levels equivalent to students in the highest quintile of socioeconomic
status, i.e. students for whom the financial burden of college is manageable. These students are
categorized in terms of highest level of attainment at: 0% high school dropouts; 5% high school
graduates; 15% some college; 25% Associate degree; and 55% Bachelor degree. The second progression
rate is that 50% of the participants obtain two years of college – equivalent to attaining an Associate
degree – and 50% complete a Bachelor degree. This is reasonable if the program heavily subsidizes
college enrollment. Thus, it is not assumed that the RFSP is effective for all students; merely that it
raises the overall probabilities of reaching higher levels of postsecondary attainment.2 These two
progression rates are used below to calculate the economic value of the RFSP relative to the status quo
educational attainment.
1
These graduation rates are derived from the New Jersey Department of Education website, adjusted for the
states’ estimate of the high school graduation rate. Alternative estimates of the graduation rate in New Jersey are
lower than those used here, so the change in educational attainment is likely to be conservatively estimated.
However, information is not available on the academic standing of the participants relative to their peers in the
districts, so only the average student path is applied. This may understate the gains if the program targets
students on the margin of high school graduation; it may overstate the gains if the participants are positively
selected on ability. Finally, no attainment beyond a BA degree is assumed; this is likely to understate the gains
from RFSP.
2
Note also that the program includes one district – Piscataway – where students are already performing at or
above state norms.
3
3. Benefits of the RFSP
3.1 Earnings Benefits
The earnings gains from additional education have been identified in hundreds of studies (Card, 1999).
Although it is not possible to prove causality, there is also an extensive methodological literature on
whether published estimates of these earnings gains are biased upward or downward.3 In reviewing this
literature, both Card (1999) and Rouse (2007) conclude that at least some of these biases do not
significantly distort the results obtained from simple earnings functions (in part because the upward and
downward biases offset) and that adjusting for prior ability does not greatly influence the relationship.
Thus, it is possible to be confident that these earnings advantages from education are genuine.
Table 1 shows the lifetime earnings by education level for males and females using data from
the Current Population Survey (2008). These are present values at 8th grade and are calculated using
three models (see Table 1 notes for details).4 The earnings figures are weighted by race as per the racial
composition of the RFSP participants. As shown, there are substantial earnings advantages for those
with more education. This is the case across all the models but for brevity results are reported from
Model [2] throughout, which is designated as a conservative model. Male [female] high school
graduates earn $176,000 [$173,000] more than dropouts; for those with a BA degree, the lifetime
earnings gap is $621,000 [$563,000]. Overall, the earnings advantages from higher levels of attainment
are substantial.5
Table 2 translates these differences into economic values for the RFSP. The earnings values in
Table 1 are mapped onto probabilistic attainment levels for students in the four target districts (as
described in Section 2). Given likely attainment levels, a male [female] in the target population but who
does not have access to the RFSP would have a lifetime earnings of $686,000 [$489,000]. For the RFSP
participants, higher educational attainment and so higher incomes are predicted. If the RFSP
3
These biases might include: ability bias, i.e. observed wage premiums are actually a return to innate ability not
education; selection bias, in that differences in earnings are conditional on individual choices about college
(Arcidiacono, 2004) or about where to work after college (Dahl, 2002); imputation bias in matching individuals to
earnings profiles (Bollinger & Hirsch, 2006); geographic clustering by education (Black et al., 2009); and sampling or
measurement error (Schmitt & Baker, 2006; Hamermesh & Donald, 2008).
4
Present valuation means that these amounts have been adjusted for the fact that the benefits of the program
occur after the program has been completed. Each model presented here uses a different discount rate (3.5%7.5%) to weight the future streams of benefits.
5
Most likely, the earnings gains exceed those reported here. They only partially adjust for differences in labor
market participation. They do not adequately value non-labor market time, which is especially salient for female
labor market decisions. They account in only a limited way for the non-pecuniary characteristics of employment
(e.g. risk, fringe benefits). Finally, they do not account for the benefits to companies from having a more educated
pool of workers to draw on.
4
participants reach attainment levels equal to those of students in the top quintile of socioeconomic
status, their lifetime earnings would be on average $973,000 [$742,000]. This amounts to an additional
lifetime earnings of $286,000 [$253,000] over equivalent students who do not participate in RFSP.
Alternatively, if half of the RFSP participants obtain Associate’s degrees and half earn Bachelor’s
degrees, the lifetime earnings are estimated at $982,000 [$744,000]. This amounts to an additional
lifetime earnings of $296,000 [$255,000] over equivalent students who do not participate in RFSP.
Thus, there is a significant private gain for students who accumulate higher levels of education.
From society’s perspective, this gain may be especially valuable given that it is helping students who
may face many obstacles to achieving economic independence.
3.2 Tax Benefits: Federal and State
Earnings gains advantages translate into significantly larger tax payments. The federal and state income
tax implications of the RFSP are calculated using the National Bureau of Economic Research tax
calculator, TAXSIM9.6 For each income level, the federal income tax payments for a household in New
Jersey are calculated and summed into a present value. State income, sales, and property taxes are
calculated separately.7 All these taxes are summed to get a full estimate of the additional taxes paid as
incomes rise.
Table 3 summarizes the present value differences in federal tax payments by education level.
Valued at the time a student is in 8th grade, a male [female] high school dropout will pay $80,000
[$35,000] in federal taxes (Model [2]). A high school graduate will pay $134,000 [$70,000] and a student
who obtains a Bachelor’s degree will pay $306,000 [$194,000] in federal income taxes over the lifetime.
Thus, as with incomes, those with more education pay higher amounts of tax.
Weighting these differences by progression rates, Table 4 shows the net additional taxes paid by
participants in the RFSP. Assuming a progression rate equal to the top quintile of socioeconomic status,
6
The TAXSIM calculations cannot accurately include expense exemptions or mortgage interest tax relief. Also,
individuals are assumed to file their taxes singly rather than as a household. In an earlier simulation, Rouse (2007)
found that single filing produces more conservative results. Tax payments adjust for differences in labor market
participation because non-workers do not pay tax. Federal tax liabilities include Social Security payments.
7
State income tax in New Jersey is progressive and ranges from 1-3% of annual earnings. State taxes New Jersey
sales tax is 7%, with some exemptions, and is regressive. Estimates for property tax are difficult to calculate. Rouse
(2007) estimates that high school graduates pay at least an extra 1% of their earnings and college graduates an
extra 1.5% of their earnings in property taxes, relative to a high school graduate. To be conservative, it is assumed
that the progressive and regressive components counterbalance, such that state tax payments are a flat
proportion of income, at 6%. State tax collections will depend on migration rates across states. However, the vast
majority of school-aged residents gain adult employment in their local labor market; also, rates of out-migration of
college students from New Jersey are low and in the case of RFSP the participants have the strong incentive to
remain in-state for their college education. (Finally, based on data at www.higheredinfo.org, New Jersey is a net
importer of college graduates.)
5
the extra federal taxes amount to $106,000 [$72,000] for males [females]. If all RFSP participants gain
two or four years of college, the additional federal taxes paid will total $110,000 [$73,000].
State tax payments by more educated persons are also substantial. Table 5 shows the
differences in state tax payments by education level. A high school dropout will pay $30,000 [$19,000]
in state taxes over the lifetime. A high school graduate will pay $40,000 [$29,000] and a college
graduate will contribute $66,000 [$53,000].
Correspondingly, Table 6 reports the net effect of the RFSP, contingent on improvements in
educational progression. State taxpayers benefit by between $15,000 and $17,000 in present value
terms for each RFSP participant.
3.3 Crime Benefits for the State
Low attainment and poor cognitive scores are key determinants of both juvenile and adult crime. Better
quality education is likely to reduce juvenile crime by improving the ‘social bond’ with school; also, being
in school reduces the time available to commit crimes. Lower juvenile crime also leads to lower adult
crime (via path-dependence). Adult crime should also be lower because the opportunity cost of crime
(earnings) is higher for persons with more education.8 Reduced criminal activity should therefore
reduce expenditures on policing, the criminal justice system, and incarceration.
For male juveniles aged 15-19, Levitt and Lochner (2001, Table 7.5) estimate that being a high
school graduate by age 18 is associated with reductions in property crime by 9%, violent crime by 17%,
and drug-related crimes by 10%. For male adults, Lochner and Moretti (2004) find education is
negatively associated with five major types of crime (violent, property, sexual assault, drug offenses, and
murder) and incarceration. For females, the educational effects are generally negative but not
statistically significant because females commit relatively few crimes. Research by Merlo and Wolpin
(2009) on black males in the NLSY97 data show that the education–crime link is stronger for more
disadvantaged and minority students. Of those black males who were in school at age 16, 18% had
committed a crime, 12% had been arrested, and 4% had been incarcerated by age 22; of those not in
school at age 16, the respective arrest, crime, and incarceration figures are 31%, 27%, and 16%. These
8
For a general review, see Farrington & Welsh (2007). On the school social bond, see Sprott et al. (2005). On the
association between education and juvenile crime, see Luallen (2006) and Hjalmarsson (2008). It is also possible
that education is associated with psychological attributes such as social control or time preference (see Wilson &
Daly, 2006).
6
relationships are applied to calculate the probabilities of being either an offender or chronic offender
across the different levels of education from dropout to college graduate.9
A number of studies have calculated the lifetime economic consequences of being either an
offender or a chronic offender.10 Recently, Belfield and Levin (2009) calculated the costs for the state of
California, including both juvenile and adult crime and separating out the costs to the taxpayer (policing,
criminal justice system, and incarceration) and the costs to society (victims of crime). Following
research convention, estimates are reported separately for offenders and chronic offenders and by sex.
All four estimates are applied here, adjusted according to local prices and criminal justice system
expenditures in New Jersey. Economic values are estimated for the additional spending on policing, the
criminal justice system, and on incarceration associated with lower levels of education, i.e. only the
crime burden that is paid by the state. In addition, it is necessary to include the marginal excess tax
burden from collecting taxes to pay for government services.11 Notably, the costs to the victims of crime
in terms of reduced quality of life (and property loss) are not counted – even though both are far larger
than the taxpayer burden.12 Based on comparisons of state and federal spending on crime, the burden
of crime is paid primarily by New Jersey citizens: federal government expenditures on crime are only
one-half of the size of expenditures by state and local crime prevention and justice system agencies.13
Table 7 reports the resource implications from the RFSP for the New Jersey state government.
These figures are calculated as the product of the probabilities of being an offender or chronic offender
times the resource implications per person, weighted by the respective progression rates. The said
quantities are present values over ones’ lifetime up to age 30.14 The economic magnitudes are
substantial for males; however, (given their much lower rate of criminal activity, the fiscal burdens and
so savings for females are very small). The expected state fiscal burden per male non-RFSP student is
9
Of all offenders, one-in-seven is a chronic offender, responsible for almost half of the total crime burden
(Farrington & Welsh, 2007).
10
Belfield & Levin (2009); Delisi & Gatling (2003) ; Fass & Pi (2002); and Cohen & Piquero (2009).
11
The marginal excess tax burden (METB) measures the economic distortion imposed by raising taxes to pay for
health, crime, and welfare services. Strictly, the value of the METB depends on the level of government at which
taxes are collected and the price elasticity of demand of the taxed good. Fullerton (1991) estimates the marginal
excess burden from labor taxes at 7-25 cents per dollar; Allgood & Snow (1998) estimate 13-28 cents as the
marginal welfare cost per dollar of a lump-sum grant. To be conservative, 7% is applied as the value for the METB.
12
Also, there are tax losses from crime. Victims of crime are often unable to work for some periods; and the
criminals themselves are not participating in the formal labor market (Holzer et al. 2004).
13
Annually, the state spends $1.02 billion on incarceration and $0.58 billion on law and public safety
(http://www.state.nj.us/budget09/budget_summary.pdf). Federal expenditures on crime and incarceration are
from the Bureau of Justice Expenditure and Employment Extracts, 2006 (December 2008, NCJ224394).
14
The peak age for crime (18-19) and the decay rate of criminal activity are such that any crimes committed
beyond age 30 have only a small impact on these estimates. Again, however, this assumption is conservative in its
estimates of the net savings from RFSP.
7
$54,000 (model [2]); with increased levels of education from the RFSP, the burden is anticipated to be
reduced to $28,000 or $22,000 depending on the progression rate. As a result, the expected value of
lifetime fiscal savings to the state from RFSP are estimated at $26,000 or $31,000 per participant. For
female participants the effects are significantly lower, at between $1,000 to $2,000.
3.4 Health Benefits for the State
The associations between education and health behaviors and health status are strong.15 More
educated persons follow healthier diets, undertake fewer risky behaviors, and appear to internalize
health-related and medical information more efficiently. They are also likely to have jobs with private
health insurance. Total health spending in New Jersey is estimated at $48 billion annually, with 6% of
adults aged under 65 on Medicaid and 2% on other public programs (along with 19% uninsured).16
Hence, the participant and the taxpayer is likely to save on government health programs if its citizens
are educated to a greater level.
The analysis here follows Muennig (2007), who estimates lifetime costs associated with
Medicaid and Medicare for those under 65. Because eligibility for Medicaid is means-tested, education
is likely to reduce enrollment indirectly through income changes, as well as directly through behavioral
changes. Also, persons aged below the age of 65 may receive Medicare if they qualify for Social Security
Disability Income (SSDI). Although these programs are funded from federal sources, just under half
(49%) of funding for Medicaid is a state responsibility. Based on the proportions of federal and state
spending on health, additional state spending is also included. As with the crime analysis, these healthrelated savings do not include any economic value that the RFSP participants themselves place on
improved health or any savings they might make in health-related spending.
Using Muennig’s (2007) economic values across different education levels, Table 8 reports the
state-level cost savings associated with the RFSP, conditional on enhanced graduation and college
enrollment.17 That is, these values include direct state expenditures and the half-share that the state
must pay for Medicaid. For male [female] RFSP participants, the state health savings are
$22,000/$21,000 [$29,000]. These are substantial differences in taxpayer subsidies for health.
15
Cutler & Lleras-Muney (2009); Lleras-Muney (2005); and Backlund et al. (1999).
For New Jersey, state health expenditures are $1.70 billion, compared to $1.56 billion from federal sources. NJ
Department of Health and Senior Services, www.njleg.state.nj.us/legislativepub/budget_2010/~
health_senior_services10.pdf; see also, http://www.nj.gov/treasury/omb/publications/10citizensguide/~
pdf/citguide.pdf.
17
These cost estimates assume that costs are a direct function of the case-load and that lower case-loads translate
into lower total costs proportionately.
16
8
3.5 Welfare Benefits for the State
Higher earnings also lead to lower reliance on welfare systems and so there is a negative association
between education and welfare reliance (Waldfogel et al., 2007; Grogger, 2004). Here, three main,
national welfare programs are considered –Temporary Assistance for Needy Families; housing
assistance; and food stamps – and state welfare programs, which are assumed to be one-tenth of the
value of these federal programs. Again, this is a conservative estimate of the state’s contributions to
welfare programs because it does not include all welfare programs or income support programs.18
Similarly, no value is placed on the subjective well-being of persons who are economically independent.
For each national program, persons with low education are more heavily represented (and the
same relationship is assumed for the state programs). Using welfare receipt rates from Jayakody et al.
(2000) and cost estimates from Levin et al. (2005) and Waldfogel et al. (2007), Table 9 reports the net
savings in welfare expenditures from the RFSP for the state (see Table Notes on the distribution of the
burden by federal and state agencies). Compared to the values for taxes, crime, and health, the savings
in welfare spending are small for females ($1,000-$1,000) and trivial for males (less than $1,000).
However, this is primarily because welfare payments are time-limited, intermittent, and often tied to
family composition (e.g. numbers of children).
3.6 Federal Benefits from RFSP
The main focus is on the consequences of the RFSP for taxpayers in New Jersey. However, the RFSP also
influences taxes and spending at the federal level. Indeed, given that the main benefit of education is its
positive association with income, the federal government is the largest beneficiary, as shown in Tables 3
and 4.
However, the federal government also benefits in terms of savings on the criminal justice
system, Medicaid-related health expenditures, and welfare expenditures. These savings are
summarized in Table 10. Not counting the federal income tax effects, the federal government will gain
from the RFSP. Each male [female] participant will yield lifetime savings of $32,000 or $35,000 [$29,000
or $31,000] for the federal government.19 To get the full value for the federal government, these
differences should be added to the tax gains reported in Table 4 above.
18
Programs are excluded where there is no direct evidence on the relationship between enrollment and education
levels.
19
Note that although these numbers are very similar by sex, the explanations are not: male savings derive largely
from lower criminal activity; female savings from a combination of health and welfare savings.
9
4. Aggregate Benefits from RFSP
4.1 Fiscal Benefits from RFSP
The total benefits from RFSP are the sum of the additional taxes paid and the savings via crime, health,
and welfare. These total benefits are expressed in present values at the start of the RFSP, i.e. the dollar
figures can be compared to amounts spent in the first year of the program. Estimates from the
conservative model are summarized in Table 11.
The top panel of Table 11 shows the fiscal consequences for New Jersey taxpayers. For each
male [female] participant, the savings to the taxpayer are $64,000-$70,000 [$47,000-$48,000]
depending on their progression rates into college. Taking a simple average from the conservative
model, state taxpayers benefit by $56,000-$59,000 per RFSP participant. This is a substantial amount
and may be compared directly to the costs of the program to see whether the program passes a cost–
benefit test.
However, the benefits from the program should be considered to be much higher: they should
account for the additional federal gains. The legitimacy of including these federal government values
depends on the extent to which federal spending in New Jersey relates to federal receipts from New
Jersey taxpayers. If there is a perfect correspondence between these two, taxpayers would be
indifferent between federal and state/local government and would value reduced taxes from either
source. Estimating the extent to which the federal government ‘subsidizes’ or ‘taxes’ New Jersey
residents relative to the rest of the U.S. is difficult. However, based on analysis by the Tax Foundation
(2007) of the Census Bureau’s 2005 data on Consolidated Federal Funds, New Jersey has historically
been one of the heaviest subsidizers of the federal government: for every $1 paid in federal taxes the
state receives only $0.6 in federal spending. Therefore, the federal savings should be valued by New
Jersey citizens, but discounted by 40%.
The bottom panel of Table 11 therefore reports the direct benefits to New Jersey taxpayers
(from the top panel) but adds in the discounted benefits indirectly received via federal government
expenditures in New Jersey. These estimates show that the gains per participant are very large. For
males [females], the gains are $152,000 to $203,000 [$109,000 to $110,000]. Taking a simple average
by sex, the total fiscal gain of the RFSP is $130,000 to $157,000 per participant.
4.2 Sensitivity Tests
The range of models used here show the robustness of the finding of significant gains. Importantly, the
figures reported above are based on a model which is best described as ‘conservative’. As noted above,
there are several reasons why this description is appropriate, including: the underestimate of the full
10
earnings gain from education (see Note 5); the use of single filing for tax purposes (Note 6); the
valuation of non-work time (especially for females) at zero; and the exclusion of some crime and health
care costs. As well, the discount rate of 5% is higher than that recommended by Moore et al. (2004),
such that all present values are reduced. Finally, these predictions assume that the gains from
additional education will stay constant over the lifetimes of the RFSP participants. In fact, the impact of
education as a determinant of economic well-being has grown over recent decades; and future
demographic and labor market changes are likely to accelerate this trend (Tienda and Solon, 2007;
Bailey, 2007). Consequently, the gains from additional education may grow as these participants reach
adulthood.
Table 12 reports results for two alternative models, compared to the conservative model. One
is classified as the lower bound – it deliberately adopts assumptions about the impacts of RFSP that are
weaker than the conservative model. For this model, which represents probably the lowest value of the
program, there are still gains of $67,000 to $80,000 per participant. The final column of Table 12 shows
a model that might be described as ‘likely’ – it adopts plausible assumptions regarding the benefits from
the RFSP but ones which are higher than for the conservative model. With this model, the RFSP yields
present value gains of $171,000 to $217,000. Overall, therefore, the gains from RFSP are expected to be
economically significant.
5. Conclusions
The potential economic value of the RFSP is likely to be significant. The estimates reported here can be
compared to the costs of the program to see whether the program passes a cost–benefit test.
It is important to emphasize two aspects of this cost–benefit analysis. First, only the taxpayer
effects of the RFSP have been counted. Typically, cost–benefit analyses include all the social
consequences (see for example early education programs in Nores et al., 2006). These social
consequences would include the private earnings gains to the participants, which are conservatively a
few hundred thousand dollars. Second, no distributional weighting is used: each dollar gain or loss is
counted the same regardless of who receives or pays it. Yet, one might believe that resources devoted
to 8th graders who currently face significant obstacles to economic independence should receive a
disproportionate weight. According to this logic, it might be appropriate to invest in educational
programs for disadvantaged children, even if the dollar costs exceed the benefits. For the RFSP, where
strong benefits are predicted already, this weighting would further strengthen the case for investment.
11
References
Allgood, S & A Snow. 1998. The marginal cost of raising tax revenue and redistributing income. Journal
of Political Economy, 106, 1246-1273.
Arcidiacono, P. 2004. Ability sorting and the returns to college major. Journal of Econometrics, 121, 343375.
Backlund, E, Sorlie, PD, & NJ Johnson. 1999. A comparison of the relationships of education and income
with mortality: the National Longitudinal Mortality Study. Social Science Medicine, 49, 1373-1384.
Bailey, T. 2007. Implications of educational inequality in a global economy. In Belfield, CR & HM Levin
(Eds). The Price We Pay: The Economic and Social Costs of Inadequate Education. Brookings
Institution: Washington, DC.
Belfield, CR & HM Levin. 2007. The Price We Pay: The Economic and Social Costs of Inadequate
Education. Brookings Institution: Washington, DC.
Belfield, CR & HM Levin. 2009. The economic burden of juvenile crime: A case study for California.
Working Paper, Center for Benefit–Cost Studies in Education, www.cbcse.org.
Black, D, Kolesnikova, N & L Taylor. 2009. Earnings functions when wages and prices vary by location.
Journal of Labor Economics, 27, 21-47.
Bollinger, CR & BT Hirsch. 2006. Match bias from earnings imputation in the Current Population Survey:
The case of imperfect matching. Journal of Labor Economics, 24, 483-519.
Card, D. 1999. The causal effect of education on earnings. In Ashenfelter O & D Card (Eds.) Handbook of
Labor Economics. North Holland: New York.
Cohen, M, & A Piquero. 2009. New evidence on the monetary value of saving a high risk youth. Journal
of Quantitative Criminology, 25, 1, 25-49.
Cutler, D & A Lleras-Muney. 2009. Understanding differences in health behaviors by education. Journal
of Health Economics, forthcoming.
Dahl, GB. 2002. Mobility and the return to education: Testing a Roy model with multiple markets.
Econometrica, 70, 2367-2420.
Delisi, M & JM Gatling. 2003. Who pays for a life of crime? An empirical assessment of the assorted
victimization costs posed by career criminals. Criminal Justice Studies: Critical Journal of the
Criminal Law Society, 16, 283-293.
Farrington, D., & B Welsh. 2007. Saving Children From a Life of Crime. New York: Oxford University Press.
Fass, SM, & CR Pi. 2002. Getting tough on juvenile crime: An analysis of costs and benefits. Journal Of
Research In Crime And Delinquency, 39, 363-399.
Fullerton, D. 1991. Reconciling recent estimates of the marginal welfare cost of taxation. American
Economic Review, 81, 302-308.
12
Grogger, J. 2004. Welfare transitions in the 1990s: The economy, welfare policy, and the EITC. Journal of
Policy Analysis and Management, 23, 671-695.
Hamermesh, DS & SG Donald. 2008. The effect of college curriculum on earnings: an affinity identifier
for non-ignorable response bias. Journal of Econometrics, 144, 479-491.
Haveman, RH & BL Wolfe. 1984. Schooling and economic well-being: The role of non-market effects.
Journal of Human Resources, 19, 377-407.
Hjalmarsson, R. 2008. Criminal justice involvement and high school completion. Journal Of Urban
Economics, 63, 2, 613-630.
Holzer, HJ, Offner, P & E Sorensen. 2004. Declining employment among young black less-educated men:
The role of incarceration and child support. National Poverty Center Working Paper Series,
University of Michigan.
Jayakody, RS, Danziger, S & H Pollock. 2000. Welfare reform, substance abuse, and mental health.
Journal of Health Politics, Policy and Law, 25, 223-243.
Levin, HM, Belfield, CR, Muennig, P & C Rouse. 2005. The Costs and Benefits of an Excellent Education
for America's Children. Working paper, www.cbcse.org.
Levin, HM, Belfield, CR, Muennig, P & C Rouse. 2007. The public returns to public educational
investments in African-American males. Economics of Education Review, 26, 699-708.
Levitt, SD & L Lochner. 2001. The determinants of juvenile crime. In J Gruber (Ed.) Risky Behavior among
Youths. An Economic Analysis. University of Chicago Press: Chicago.
Lleras-Muney A. 2005. The relationship between education and adult mortality in the United States.
Review of Economic Studies, 72, 189-221.
Lochner, L & E Moretti. 2004. The effect of education on crime: Evidence from prison inmates, arrests,
and self-reports. American Economic Review, 94, 155-189.
Luallen, J. 2006. School's out... forever: A study of juvenile crime, at-risk youths and teacher strikes.
Journal of Urban Economics, 59, 75-103.
Merlo, A & KI Wolpin. 2009. The transition from school to jail: Youth crime and high school completion
among black males. Working Paper, University of Pennsylvania.
Moore, MA, Boardman, AE, Vining AR, Weimer, DL & DH Greenberg. 2004. Just give me a number!
Practical values for the social discount rate. Journal of Policy Analysis and Management, 23, 789812.
Muennig, P. 2007. Consequences in health status and costs. In CR Belfield & HM Levin (Eds.) The Price
We Pay: The Social and Economic Costs to the Nation of Inadequate Education. Brookings
Institution Press: Washington, DC.
Nores, M, Belfield, CR, Barnett, WS & L Schweinhart. 2006. Updating the economic impacts of the
High/Scope Perry Preschool program. Educational Evaluation and Policy Analysis, 27, 245-261.
13
Rouse, C. 2007. The earnings benefits from education. In CR Belfield & HM Levin (Eds.) The Price We
Pay: The Social and Economic Costs to the Nation of Inadequate Education. Brookings Institution
Press: Washington, DC.
Schmitt, J & D Baker. 2006. The impact of undercounting in the Current Population Survey. Working
paper, Center for Economic Policy and Research, Washington DC.
Sprott, JM, Jenkins JM, & AN Doob. 2005. The importance of school: protecting at-risk youth from early
offending. Youth Violence and Juvenile Justice, 3, 59-77.
Swanson, CB. 2004. Who Graduates? Who Doesn’t? A Statistical Portrait of Public High School
Graduation, Class of 2001. Working Paper, Urban Institute.
Tax Foundation. 2007. Federal tax burdens and spending by state. Special Report 158,
taxfoundation.org.
Tienda, M & S Alon. 2007. Diversity and the demographic dividend: Achieving educational equity in an
aging white society. In Belfield, CR and HM Levin (Eds). The Price We Pay: The Economic and Social
Costs of Inadequate Education. Brookings Institution: Washington, DC.
Waldfogel, J, Garfinkel, I & B Kelly. 2007. Public assistance programs: How much could be saved with
improved education? In CR Belfield & HM Levin (Eds.) The Price We Pay: The Social and Economic
Costs to the Nation of Inadequate Education. Brookings Institution Press: Washington, DC.
Wilson, M, & M Daly. 2006. Are juvenile offenders extreme future discounters? Psychological Science,
17, 989-994.
14
Table 1
Earnings
Lifetime Present Value by Education Level
High school
dropout
High school
graduate (incl.
GED)
Some college
Associate
degree
Bachelor degree
Males:
Model [1]
$806,870
$1,098,243
$1,241,035
$1,430,397
$1,870,123
Model [2]
$496,491
$672,387
$752,732
$847,319
$1,117,213
Model [3]
$259,936
$347,886
$380,214
$414,451
$545,755
Model [1]
$511,548
$778,831
$910,440
$991,059
$1,401,802
Model [2]
$311,049
$483,614
$557,991
$614,921
$873,796
Model [3]
$160,814
$255,751
$286,968
$320,398
$455,985
Females:
Notes: Earnings data from the Current Population Survey from 2008, adjusted to New Jersey average incomes based on BLS
data. Earnings weighted according to racial composition of RFSP (48% Hispanic, 40% African American, 12% other). Present
values at age 14. No adjustment for labor market participation or unemployment differences, GED receipt, or incarceration
rates. Labor market activity begins at age 18 and lasts until age 65. Model [1]: health benefits incidence as per Rouse (2007)
with valuation at 10% of earnings; discount rate 3.5%; productivity growth 1.5%. Model [2]: health benefits incidence/valuation
as per model [1]; discount rate 5%; productivity growth 1%. Model [3]: no adjustment for health benefits incidence and
valuation; discount rate 7.5%; productivity growth 0.5%.
15
Table 2
Earnings
Lifetime Expected Present Value Gain from RFSP
(1)
Expected income
value per non-RFSP
student
(2)
Expected income
value per RFSP
participant
Difference per
participant
(2)-(1)
Progression rates: Current vs. top
SES quintile
Males:
Model [1]
$ 1,128,984
$ 1,627,234
$ 498,250
Model [2]
$ 686,418
$ 972,826
$ 286,408
Model [3]
$ 350,189
$ 478,204
$ 128,015
Model [1]
$ 791,242
$ 1,194,263
$ 403,022
Model [2]
$ 489,162
$ 742,197
$ 253,035
Model [3]
$ 256,006
$ 386,724
$ 130,718
Model [1]
$ 1,128,984
$ 1,650,260
$ 521,275
Model [2]
$ 686,418
$ 982,266
$ 295,848
Model [3]
$ 350,189
$ 480,103
$ 129,914
Model [1]
$ 791,242
$ 1,196,430
$ 405,189
Model [2]
$ 489,162
$ 744,358
$ 255,196
Model [3]
$ 256,006
$ 388,192
$ 132,185
Females:
Progression rates: Current vs. half
Assoc. and half BA degrees
Males:
Females:
Notes: Values are present values at age 14. Earnings gains calculated based on Table 1. See Notes to Table 1 for model
specifications. Current progression rates for black males females taken from Belfield and Levin (2007). Top SES quintile
progression rates taken from Beginning Postsecondary Students Longitudinal Study 1991-96.
16
Table 3
Federal Tax Payments
Lifetime Present Value by Education Level
High school
dropout
High school
graduate (incl.
GED)
Some college
Associate
degree
Bachelor degree
Males:
Model [1]
$131,959
$223,392
$275,988
$347,211
$518,397
Model [2]
$79,700
$133,757
$163,135
$199,687
$305,658
Model [3]
$40,809
$66,940
$79,424
$93,163
$145,656
Model [1]
$38,103
$76,928
$98,646
$116,803
$216,949
Model [2]
$35,460
$69,880
$88,606
$105,215
$193,832
Model [3]
$18,333
$36,331
$44,141
$53,267
$99,280
Females:
Notes: Earnings data from models [1]-[3] in Appendix Table 1. Present values at age 14. Federal tax payments include Social
Security and payroll taxes paid by the individual. Taxes calculated at 2009 tax rates, single filer status, new Jersey resident,
using TAXSIM9. Labor market participation rates applied by education level for males and female separately. The marginal
excess tax burden for collecting federal tax is estimated at 7%.
17
Table 4
Federal Tax Contribution
Lifetime Expected Value Gain from RFSP
(1)
Expected tax
contribution per nonRFSP student
(2)
Expected tax
contribution per RFSP
participant
Difference per
participant
(2)-(1)
Model [1]
$240,762
$424,489
$183,727
Model [2]
$143,121
$249,192
$106,071
Model [3]
$70,469
$118,662
$48,193
Model [1]
$84,994
$167,166
$82,172
Model [2]
$76,973
$149,696
$72,723
Model [3]
$39,588
$76,359
$36,771
Model [1]
$240,762
$432,804
$192,042
Model [2]
$143,121
$252,672
$109,552
Model [3]
$70,469
$119,409
$48,941
Model [1]
$84,994
$166,876
$81,882
Model [2]
$76,973
$149,523
$72,551
Model [3]
$39,588
$76,274
$36,686
Progression rates: Current vs. top
SES quintile
Males:
Females:
Progression rates: Current vs. half
Assoc. and half BA degrees
Males:
Females:
Notes: Additional tax contributions calculated based on Table 3. Current progression rates for black males females taken from
Belfield & Levin (2007). Top SES quintile progression rates taken from Beginning Postsecondary Students Longitudinal Study
1991-96.
18
Table 5
State Tax Payments
Lifetime Present Value by Education Level
High school
dropout
High school
graduate (incl.
GED)
Some college
Associate
degree
Bachelor degree
Model [1]
$48,412
$64,773
$73,134
$84,155
$110,030
Model [2]
$29,789
$39,663
$44,371
$49,845
$65,711
Model [3]
$16,435
$21,886
$24,013
$26,225
$34,596
Model [1]
$30,693
$46,730
$54,626
$59,464
$84,108
Model [2]
$18,947
$29,465
$33,997
$37,475
$53,228
Model [3]
$9,649
$15,345
$17,218
$19,224
$27,359
Males:
Females:
Notes: Earnings data from models [1]-[3] in Table 1. Present values at age 14. State tax payments include income tax
(calculated at 2009 tax rates, single filer status, new Jersey resident, using TAXSIM9); sales tax; and property tax.
19
Table 6
State Tax Contributions
Lifetime Expected Value Gain from RFSP
(1)
Expected tax
contribution per nonRFSP student
(2)
Expected tax
contribution per RFSP
participant
Difference per
participant
(2)-(1)
Progression rates: Current vs. top
SES quintile
Males:
Model [1]
$66,785
$95,764
$28,979
Model [2]
$40,608
$57,241
$16,632
Model [3]
$22,102
$30,280
$8,179
Model [1]
$47,474
$71,656
$24,181
Model [2]
$29,801
$45,217
$15,416
Model [3]
$15,360
$23,203
$7,843
Model [1]
$66,785
$97,093
$30,308
Model [2]
$40,608
$57,778
$17,169
Model [3]
$22,102
$30,411
$8,309
Model [1]
$47,474
$71,786
$24,311
Model [2]
$29,801
$45,351
$15,550
Model [3]
$15,360
$23,291
$7,931
Females:
Progression rates: Current vs. half
Assoc. and half BA degrees
Males:
Females:
Notes: Tax contributions calculated based on Table 5. Current progression rates for black males females taken from Belfield &
Levin (2007). Top SES quintile progression rates taken from Beginning Postsecondary Students Longitudinal Study 1991-96.
20
Table 7
State Government Savings from Crime Reduction
Lifetime Expected Value Gain from RFSP
(1)
Expected fiscal
burden per non-RFSP
student
(2)
Expected fiscal
burden per RFSP
participant
Difference per
participant
(2)-(1)
Progression rates: Current vs. top
SES quintile
Males:
Model [a]
$62,110
$32,273
$29,837
Model [b]
$53,732
$27,920
$25,812
Model [c]
$45,354
$23,566
$21,788
Model [a]
$3,458
$2,135
$1,323
Model [b]
$2,991
$1,847
$1,144
Model [c]
$2,529
$1,561
$968
Model [a]
$62,110
$25,992
$36,118
Model [b]
$53,732
$22,486
$31,247
Model [c]
$45,354
$18,980
$26,374
Model [a]
$3,458
$1,606
$1,852
Model [b]
$2,991
$1,358
$1,633
Model [c]
$2,529
$1,857
$672
Females:
Progression rates: Current vs. half
Assoc. and half BA degrees
Males:
Females:
Notes: Present values with 5% discount rate. Fiscal burden includes costs of policing, criminal justice system, incarceration,
parole, and probation, school-site expenditures on crime, and marginal excess tax burden. Estimates derived from Belfield &
Levin (2009), adjusted for New Jersey prices and expenditures on crime. Costs estimated for general offenders and chronic
offenders separately by sex. Models [a]-[c] vary by proportions classified as general and chronic offenders. Only state savings
are reported; federal savings are estimated at 50% of state savings.
21
Table 8
State Government Savings from Health Improvement
Lifetime Expected Value Gain from RFSP
(1)
Expected fiscal
burden per non-RFSP
student
(2)
Expected fiscal
burden per RFSP
participant
Difference per
participant
(2)-(1)
Progression rates: Current vs. top
SES quintile
Males:
Model [x]
$29,721
$6,995
$22,726
Model [y]
$31,912
$16,495
$10,136
$5,191
$21,766
$11,304
Model [x]
$41,732
$11,737
$29,995
Model [y]
$42,787
Model [z]
$22,684
$13,584
$6,557
$29,203
$16,127
Model [x]
$29,721
$7,056
$22,665
Model [y]
$31,912
$16,495
$10,477
$5,384
$21,435
Model [x]
$41,732
$11,899
$29,833
Model [y]
$42,787
$13,945
$28,842
Model [z]
$22,684
$6,128
$16,556
Model [z]
Females:
Progression rates: Current vs. half
Assoc. and half BA degrees
Males:
Model [z]
$11,111
Females:
Notes: Present values with a 5% discount rate derived from Muennig (2007), with original data taken from MEPS (2003). Fiscal
burden includes Medicare and Medicaid for those with SSDI only. Model [x] applied percentage enrollment rates by education
level from Muennig (2007, Technical Appendix). Model [y] applies direct cost estimates from Muennig (2007, Technical
Appendix). Model [z] follows Model [y] but applies a 7.5% discount rate. Cost savings include only state/local contributions for
health programs.
22
Table 9
State Government Savings in Welfare Expenditures
Lifetime Expected Value Gain from RFSP
(1)
Expected fiscal
burden per non-RFSP
student
(2)
Expected fiscal
burden per RFSP
participant
Difference per
participant
(2)-(1)
Progression rates: Current vs. top
SES quintile
Males:
Model [i]
$394
$94
$300
Model [ii]
$342
$276
$82
$66
$260
$210
Model [i]
$2,396
$696
$1,700
Model [ii]
$2,084
Model [iii]
$1,681
$605
$488
$1,479
$1,193
Model [i]
$394
$15
$379
Model [ii]
$342
$276
$13
$10
$329
Model [i]
$2,396
$146
$2,250
Model [ii]
$2,084
$127
$1,957
Model [iii]
$1,681
$102
$1,579
Model [iii]
Females:
Progression rates: Current vs. half
Assoc. and half BA degrees
Males:
Model [iii]
$266
Females:
Notes: Present values with 5% discount rate. Fiscal burden includes costs of TANF, housing assistance, food stamps, and state
welfare services. Marginal excess tax burden included. Model [i] follows Model [ii] but includes an administrative cost of 15%
for program delivery. Model [ii] is based on estimates from Levin et al. (2005) and Waldfogel et al. (2007), from welfare receipt
rates reported in Jayakody et al. (2000). Model [iii] follows Model [ii] but applies a discount rate of 7.5%. Sources of welfare
funding are: TANF and food stamps, 10% state; housing assistance 70% state; other spending, 100% state.
23
Table 10
Federal Savings: Crime, Health and Welfare
Lifetime Expected Value Gain from RFSP
(1)
Expected tax
contribution per nonRFSP student
(2)
Expected tax
contribution per RFSP
participant
Difference per
participant
(2)-(1)
Progression rates: Current vs. top
SES quintile
Males:
Model [1]
$57,002
$22,218
$34,784
Model [2]
$54,437
$22,594
$31,843
Model [3]
$37,285
$16,283
$21,002
Model [1]
$43,843
$12,881
$30,962
Model [2]
$43,481
$13,983
$29,498
Model [3]
$25,366
$6,842
$18,524
Model [1]
$57,002
$18,855
$38,147
Model [2]
$54,437
$19,922
$34,515
Model [3]
$37,285
$13,959
$23,326
Model [1]
$43,843
$11,096
$32,747
Model [2]
$43,481
$12,614
$30,867
Model [3]
$25,366
$6,334
$19,032
Females:
Progression rates: Current vs. half
Assoc. and half BA degrees
Males:
Females:
Notes: Federal savings from crime estimated at 50% of state savings (Belfield & Levin, 2009). Federal savings from health
estimated from proportion of Medicaid and SSDI spending by federal government from Muennig (2007). Federal savings in
welfare based on Table 9, with federal sources of welfare funding at: TANF and food stamps, 90%; housing assistance 30%;
other spending, 0%.
24
Table 11
Total Expected Present Value of Rutgers Future Scholars Program
(1) Additional tax
payments
(2) Savings in crime
expenditures
Per RFSP participant
(over non-participant)
(3) Savings in health
expenditures
(4) Savings in welfare
expenditures
(1)+(2)+(3)+(4)
Total present value
NEW JERSEY STATE ONLY:
Progression rates: Current vs. top SES
quintile
Males
$16,632
$25,812
$21,776
$260
$64,480
Females
$15,416
$1,144
$29,203
$1,479
$47,242
Average
$16,024
$13,478
$25,490
$870
$55,861
Males
$17,169
$31,246
$21,435
$329
$70,179
Females
$15,550
$1,633
$28,842
$1,957
$47,982
Average
$16,360
$16,440
$25,139
$1,143
$59,081
Males
$80,275
$37,910
$32,560
$838
$151,583
Females
$59,050
$1,488
$43,665
$4,372
$108,574
Average
$69,662
$19,699
$38,113
$2,605
$130,079
$126,192
$25,726
$32,050
$32
$203,417
Females
$59,081
$2,049
$43,126
$5,778
$110,033
Average
$92,636
$23,087
$37,588
$3,413
$156,725
Progression rates: Current vs. half
Assoc. and half BA degrees
NEW JERSEY STATE PLUS ADJUSTED
FEDERAL SAVINGS*:
Progression rates: Current vs. top SES
quintile
Progression rates: Current vs. half
Assoc. and half BA degrees
Males
Notes: * Federal spending is adjusted downward by 40% to account for the net transfer New Jersey makes to the rest of the U.S. through the federal government. Figures from
th
Tables 4-10 above using conservative model ((2), b, y and ii). Present value at 8 grade with discount rate of 5%.
25
Table 12
Total Expected Value of Rutgers Future Scholars Program
Range of Estimates
Total Value per RFSP participant
(over non-participant)
Lower Bound
Conservative
Likely
Males
$41,481
$64,480
$81,842
Females
$26,131
$47,242
$57,199
Average
$33,806
$55,861
$69,521
Males
$46,060
$70,179
$89,470
Females
$26,738
$47,982
$58,246
$36,399
$59,081
$73,858
Males
$74,291
$151,583
$217,302
Females
$59,308
$108,574
$125,079
Average
$66,799
$130,079
$171,191
$99,761
$203,417
$307,118
$60,169
$110,033
$127,024
$79,965
$156,725
$217,071
NEW JERSEY STATE ONLY:
Progression rates: Current vs. top SES
quintile
Progression rates: Current vs. half
Assoc. and half BA degrees
Average
NEW JERSEY STATE PLUS ADJUSTED
FEDERAL SAVINGS*:
Progression rates: Current vs. top SES
quintile
Progression rates: Current vs. half
Assoc. and half BA degrees
Males
Females
Average
Notes: * Federal spending is adjusted downward by 40% to account for the net transfer New Jersey makes to the rest of the U.S. through the federal government. Conservative
model is from Table 11 above. Lower Bound model is from models [3], [c], [z] and (iii) in Tables 4-10 above. Likely model is from models [1], [a], [x] and (i) in Tables 4-10 above.
26