Program Report: Progress Report on the Children's Program, 2004

The impact of public policy on the well-being of children continues to be a major area of interest for policymakers. Likewise, the economics of children's issues continue to concern members of the Children's Program at the NBER. This Program brings together a wide variety of researchers from across many different fields, including labor economics, public economics, industrial organization, econometrics, and development economics. These researchers work on a diverse set of issues related to the wellbeing of children, and present their work in annual meetings each spring and in the NBER's Summer Institute. Much of the work that is done by Children's Program members is funded by an Integrated Research Program Grant from the National Institute of Child Health and Human Development.

The first report on the Children's Program was written four years ago; since that time, the Program has continued to thrive because of a rapid growth in its roster of members, including some of the most exciting young economists in the profession. In this report, I focus on the contributions of this group of researchers in five areas: the economics of education; evaluating welfare reform; modeling child health and insurance coverage; modeling and assessing the implications of risky behaviors among youth; and modeling the determinants of and assessing the implications of family structure for youth.

Education

The economics of education remain the main focus for researchers affiliated with the Children's Program. This research is concerned with estimating the labor market returns to additional schooling. The problem facing researchers has been that individuals who obtain higher levels of schooling may be of higher ability, so their higher wages later in life could reflect the returns to their ability and not to the education per se. David Card (WP 7769) reviews a number of clever solutions to this problem that have been suggested by researchers, including comparing those who differ in their access to schools, or those subject to different compulsory schooling laws. He concludes that the returns to additional education are larger than traditional estimates, reflecting the fact that the low-income groups that are affected by supply-side innovations have particularly high returns to schooling. Consistent with this conclusion, Philip Oreopolous (10155) shows that students who are allowed to drop out of school earlier through looser compulsory schooling laws see enormous reductions in their lifetime earnings, wealth, and health. He finds that dropping out one year later increases future income by more than ten times the foregone wages during that extra year of high school. Esther Duflo (7860, 8710) shows that in Indonesia a massive school construction project was associated with substantial improvements in the labor market outcomes of the generation of students that benefited from that program.

In assessing the appropriate government role in education, however, what matters is not only the private returns in terms of higher wages, but also the public returns to society from having a more highly educated populace. Researchers in the Children's Program have made great progress over the past few years in documenting these public returns, relying on the type of solutions used to estimate the wage returns to education. Thomas Dee (9588) and Kevin Milligan, Enrico Moretti and Oreopolous (9584) show that higher education attributable to college availability and compulsory schooling laws leads to higher levels of civic participation, in terms of voting and awareness of public issues. Adriana Lleras-Muney (8986) demonstrates that higher education related to compulsory schooling laws in the early twentieth century led to reduced mortality later in life. Janet Currie and Moretti (9360) show that higher maternal education linked to college openings leads to improved health for the infants of these mothers. Oreopolous, Marianne E. Page, and Ann Huff Stevens (10164) find that stronger compulsory schooling requirements for parents not only increased their own education, but also the education of their children, suggesting important intergenerational effects of education policy. Finally, there is some debate over whether higher levels of education among some workers "spill over" to higher productivity levels among their co-workers, with Moretti (9108,9316) finding evidence for such spillovers, and Daron Acemoglu and Joshua Angrist (7444) finding no evidence for spillovers.

Another focus of research for Children's Program members has been assessing innovative ways to improve school quality and student outcomes. Angrist and Victor Lavy (9389) study an Israeli program that provided a cash bonus for high school matriculation. They find that such bonuses are effective when provided to an entire school, but not to individual students within the school. Brian A. Jacob and Lars Lefgren (8918) find that assigning students to remedial summer education improved the school performance of third graders, but not sixth graders. David N. Figlio and Maurice E. Lucas (7985) find that students benefit from having teachers who have relatively high grading standards. Alan Krueger (8875) reviews the evidence showing that reduced class size improves student performance, and Krueger and Diane Whitmore (7656) provide additional positive evidence for the effects of a Tennessee experiment that reduced class sizes in terms of increasing college entrance exam-taking, particularly for minorities. Angrist and Jonathan Guryan (9545) find that teacher-testing programs in the United States lead to higher teacher wages, but no improvement in teacher quality.

Another exciting new area of research for Children's Program members is the effect of school choice and school accountability efforts on educational production and student outcomes. The results on the impact of existing school choice programs are mixed. As Caroline Hoxby (8873) argues, existing evidence as of 2001 suggests that school choice could greatly improve the productivity of schooling in the United States. But Julie Cullen, Brian A. Jacob, and Steven Levitt (10113) study a Chicago program which used a lottery to assign slots to the most desired public schools, and they find no improvement in outcomes for those students assigned to the best schools. And, Krueger and Pei Zhu (9418) reevaluate a voucher program in New York City and find no evidence of a resulting improvement in student outcomes. So, the impact of school choice on student outcomes remains a lively area of debate.

Another popular policy initiative at both the state and federal level is to make schools more accountable for the performance of their students on standardized exams, for example by tying school funding to school performance. While the incentives tied to accountability may improve educational production, a number of studies by Children's Program researchers have shown that efforts to make schools more accountable for their performance are also having perverse effects on educational incentives. For example, David Figlio and Lawrence S. Getzler (9307) find that making schools accountable for their performance on a Florida exam led the schools to classify low-skills students as disabled (and thus excluded from testing) in order to raise their test scores. Figlio and Joshua Winicki (9319) find that schools even manipulated their menus around testing time, increasing calories to improve student energy levels and test scores. Jacob and Levitt (9413, 9414) use clever statistical methods to identify cheating by teachers on standardized exams (through giving out the right answers to their students), and propose means of identifying and addressing this cheating. Jacob (8968) finds strong evidence that accountability leads schools to "teach to the test": the introduction of such an accountability system tied to a particular test in Chicago led to improved scores on that test, but no improvement on a more general test which did not matter for school accountability. This finding is mirrored in a paper by Paul Glewwe, Nauman Iilas, and Michael Kremer (9671) in the context of a developing country: they find that such teaching-to-the-test occurs in an incentive program in Kenya. And, Thomas J. Kane and Douglas O. Staiger (8156) highlight the unreliability of year-to-year variation in school test scores, particularly for small schools, as a means of assessing school performance.

As teachers in institutions of higher education, economists in the Children's Program continue to be fascinated by the economics of the higher education sector as well. Susan Dynarski (7756, 9400) finds that state "merit aid" programs that provide subsidies to in-state students who meet a modest high school achievement standard raise the probability of college attendance by 5 to 7 percentage points, but that a particular program in Georgia had benefits that were focused solely on higher income groups because of more stringent achievement requirements and poor income targeting. Dennis Epple, Richard Romano, and Holger Seig (9799) use a sophisticated model of college admissions processes to predict that abolishing affirmative action would lead to a large fall in non-white representation at top colleges and universities. Kane (9703) finds that a California grant program for college costs significantly increased college matriculation among applicants, particularly in four-year colleges. Christopher Avery and Hoxby (9482) find that students respond somewhat irrationally to the design of college aid packages, irrationally preferring grants that are called "scholarships" and are front-loaded to provide more benefits in initial years. David M. Linsenmeier, Harvey S. Rosen, and Cecelia E. Rouse (9228) find some evidence that moving from a loan to a grant system at one university led to increased matriculation by low income minority students.

Welfare Reform

One particular concern about welfare reform is that it led individuals to lose the public health insurance coverage they receive through the Medicaid program; in principle, families could retain health insurance coverage when they leave welfare, but in practice families may not have been aware of this benefit. In fact, Kaestner and Neeraj Kaushal (10033) find that welfare reform was associated with a sizeable rise in the loss of insurance among single mothers, and Kaestner and Won Chan Lee (9769) find that there were reductions in use of prenatal care and increases in low birth weight infants.

One of the most important changes in programs affecting children over the past several decades was the major reform of cash welfare programs targeted to single mothers in the United States. The 1996 reforms gave states much more freedom to design their cash welfare programs and imposed time limits on the receipt of cash welfare. A large body of work by researchers in the Children's Program over the past few years has assessed the impacts of these reforms. Robert Moffitt (8749) and Rebecca Blank (8983) review this work, and other related work on this topic. An excellent overview of the impacts of all of the redistributive transfer programs in the United States is provided in Moffitt's recent volume, Means Tested Transfer Programs in the U.S. The introduction to that volume is Moffitt (8730).

A major focus of work in this area has been the impact of welfare reform on the labor supply of single mothers and the wellbeing of their families. Robert Scheoni and Blank (7627) find that welfare reform led to reduced welfare participation and increased family earnings, so that total family income rose and poverty declined as a result. Similarly, Bruce D. Meyer and James X. Sullivan (8928) find that the consumption of families headed by single mothers did not decline as a result of welfare reform. Melissa Kearney (9093) and Theodore Joyce, and Robert Kaestner and Sanders Korenman (9406) find that welfare reform had no effect on non-marital fertility in the United States, although Marianne P. Bitler, Jonah B. Gelbach, and Hilary W. Hoynes (8784) do find important effects on living arrangements of women and children. Jeffrey Grogger (7709) finds that the time limits imposed through welfare reform caused significant declines in welfare participation among those families with younger children, possibly concerned that they not "use up" their limited time on welfare.

Child Health and Insurance Coverage

A major topic of longstanding interest to members of the Children's Program has been the health and health insurance coverage of children. Public health insurance is provided for children under the Medicaid program, as I review in WP 7829. The literature as of that year (2000) suggested that expanding Medicaid increases public insurance coverage, decreases private insurance coverage, and reduces use of welfare (by allowing individuals to maintain insurance when they leave welfare). More recent work by Children's Program members has provided more mixed evidence with respect to these conclusions. Card and Lara D. Shore-Sheppard (9058) find that Medicaid expansions to non-poverty groups of children had little impact on either Medicaid or private health insurance coverage, while Currie and Grogger (7667) find that expansions did increase insurance coverage among children, but that this was offset to some extent by welfare contractions. Meyer and Dan T. Rosenbaum (7491) and John C. Ham and Shore-Sheppard (9803) find that Medicaid expansions did not have significant effects on welfare participation.

An interesting series of more recent papers have delved into the important issue of how socioeconomic status affects the health of children. Anne Case, Darren Lubotsky, and Christina Paxson (8344) find that child health rises with family income, and that this relationship strengthens as children age; a large part of this relationship appears to be caused by the superior ability of high-income families to deal with chronic health problems in their children. Currie and Mark Stabile (9098) explore this relationship further using Canadian data and find that the major problem is the higher incidence of negative health shocks for low-income families. Case, Angela Fertig, and Paxson (9788) find that childhood health has a very persistent effect and that individuals in worse health as children have poorer adult health, lower educational attainment, and lower adult earnings. Case and Paxson (7691) find that children living with stepmothers (as opposed to biological mothers) are less likely to have routine medical visits, although this effect is mitigated if the children have regular contact with their biological mothers.

Risky Behaviors

Youths engage in a variety of risky behaviors on a frequent basis: drinking, smoking, consuming illegal drugs, driving recklessly, having unprotected sex, dropping out of school, and others. This motivated an earlier study through the Children's Program that I edited: Risky Behavior Among Youth. In the past few years, a large number of papers have continued this research agenda.

A primary focus of this research has been on smoking. Botond Koszegi and I (7507, 8777) have developed a theory of smoking which highlights the self-control problems faced by smokers who want to quit. This theory suggests much more interventionist government policies than are recommended by traditional economic models.⁽²⁾ Sendhil Mullainathan and I (8872) develop evidence that supports this alternative model, finding that higher cigarette taxes raise the self-reported wellbeing of smokers. John A. Tauras, Patrick M. O'Malley, and Lloyd D. Johnston (8331) find that higher cigarette prices deter teens from starting to smoke. Matthew C. Farrelly, Terry F. Pechacek, and Frank J. Chaloupka (8691) find that increased state funding for tobacco control programs has reduced tobacco use in recent years. Greg Coleman, Michael Grossman, and Ted Joyce (9245) find that higher cigarette prices have a strong effect in deterring women from smoking when they are pregnant, and in preventing relapse to smoking after childbirth. Donna B. Gilleskie and Koleman S. Strumpf (7838) find that price responsiveness is much stronger among teens who do not yet smoke than among teens who are already smokers.

Other research in this area has focused on costs for youths of using illicit drugs or consuming alcohol. Rosalie L. Pacula and Beau Kilmer (10046) find that consumption of marijuana is associated with increased property crimes and increased odds of getting caught for violent crimes. Naci Mocan and Erdal Tekin (9824) find that use of illicit drugs by one twin, when the other does not use the drugs, is associated with a much higher incidence of criminal behavior. Pacula, Karen E. Ross, and Jeanne Ringel (9963) find that marijuana use is associated with a 15 percent decline in performance on standardized exams. Grossman and Sara Markowitz (9244) find that alcohol use does not increase the likelihood of having sex or of having multiple sexual partners, but it does lower the odds of using birth control. Michael Kremer and Dan Levy (9876) use randomized assignment of roommates in college to demonstrate that individuals assigned to a roommate who is a drinker in the year before college have a grade point average which is one-quarter point lower than if they were assigned a non-drinking roommate, and that this effect is much larger than effects of the roommate's high school grades or family background. Finally, Jenny Williams, Pacula, Chaloupka, and Henry Wechsler (8401) find that alcohol use and marijuana use among college students are complements, so that reducing use of one substance leads to reduced use of the other.

Family Structure

The final major area of research by Children's Program members is the determinants and consequences of family structure. One exciting topic of research has been the implications of abortion availability. Philip B. Levine and Douglas Staiger (8813) argue that abortion availability both provides a form of "insurance" against unwanted pregnancies and reduces the incentive to avoid pregnancies through birth control or abstinence. They find evidence consistent with these contentions in Eastern Europe, where legalizing abortion led to a large drop in births, but reducing the costs of abortion once available led to a rise in pregnancies that was offset by a rise in abortions. John Donohue and Levitt (8004) argue that increased abortion availability in the early 1970s led to reduced crime in the early 1990s (roughly 18 years later), as the children not born because of abortion availability would have been particularly likely to commit crimes. Joyce (8319) disagrees with those findings, and Donohue and Levitt (9532) respond, leading to a lively debate on this important topic. The opposite fertility incentive was provided in Quebec from 1988 to 1997, which offered a very large bonus to families for having additional children, up to $8000 for the third child. Milligan (8845) finds that this policy caused a sizeable increase in births in Quebec, but that the response was concentrated among higher income families (which is consistent with the lack of birth response to welfare reforms among lower income groups).

Another potentially important determinant of child wellbeing is marital dissolution. I compare children who grew up in states where divorce was easier to obtain to those who grew up in states where divorce was more difficult to obtain, and find that growing up where divorce was easier leads to more parental divorce and much worse outcomes later in life in terms of education, income, and marital stability (7968). Page and Stevens (8786) find that divorce is associated with a dramatic reduction in family resources, with income falling by 40-45 percent and consumption falling by 17 percent six or more years after the divorce.

Conclusions

The past four years have seen continued growth in the areas of interest and depth of research done by members of the Childrens' Program. The results are an important set of findings that dramatically advance our understanding of how education affects children, what the impacts of welfare reform on family well-being were, what determines the health and health insurance coverage of children, how children decide to engage in risky behaviors, and the implications of alternative family structures for child outcomes. The findings summarized here are only a small part of the total research done by the Children's Program, which includes work on youth employment, childcare, nutrition, and other topics. This policy-relevant work will continue to promote our understanding of children's wellbeing in the United States and around the world.

The numbers in parentheses throughout this article refer to NBER Working Papers.

2. J. Gruber, "The New Economics of Smoking," NBER Reporter, Summer 2003.