Part 3. Determinants of Infant Health with Special Emphasis on Public Policies and Programs

Introduction to Part 3



A



fter I completed my 1976 paper dealing with the correlation between

schooling and adult health, I turned to the determinants of infant and

child health. The last three papers in the previous section are examples of my

work in that area. Those three papers, however, focus on the effects of parents’

schooling. The five papers in the current section are much broader. Although

they all deal with infant health, they employ a rich analytical framework in

which my demand for health model is combined with economic models of the

family developed most notably by Gary S. Becker and H. Gregg Lewis (1973),

Robert J. Willis (1973), and Becker (1981). In these models, parents maximize

a utility function that depends on their own consumption, the number of children,

and the quality of each child. It is natural to associate infants’ health with their

quality and thus to “marry” an economic approach to the family with an economic

approach to the demand for health.

Why did I decide to conduct a series of studies of the determinants of infant

mortality after I completed my last paper that employed measures of child health

beyond the first year of life in 1981 (Shakotko, Edwards and Grossman—the

second paper in the previous section)? One reason is that in the United States

and the rest of the developed world, the infant mortality rate is much higher

than child and teenage mortality rates—thirteen times greater than the largest

of these rates in the United States in 2013. Moreover, adult age-specific death



297



|



I N T R O D U C T I O N T O PA R T 3



rates do not exceed the infant mortality rate until the rate at ages fifty-five to

fifty-nine (Xu, Murphy, Kochanek, and Bastian 2016). In addition, relatively

high infant mortality rates in areas within a given country signal poor health in

all segments of the population (Fuchs 1983).

A second set of factors revolves around trends in the U.S. infant mortality

rate between the mid-1950s and the early 1980s and the relative impacts of public policies, programs, and advances in medical technology in explaining these

trends. During the period at issue, the infant mortality rate was characterized by

a decade of relative stability followed by almost two decades of rapid decline.

The rate fell by only 0.6 percent per year compounded annually between 1955

and 1964. By contrast, infant mortality dropped by 4.5 percent per year (compounded annually) between 1964 and 1982.

The period beginning in 1964 witnessed the legalization and diffusion of

abortion, the widespread adoption of oral and intrauterine contraceptive techniques, and dramatic advances in neonatal science. It also witnessed the introduction and rapid growth of programs associated with President Lyndon B.

Johnson’s War on Poverty: Medicaid, federally subsidized maternal and infant

care projects and community health centers (hereafter community health projects), federally subsidized family planning services for low-income women,

and the Special Supplemental Food Program for Women, Infants, and Children

(WIC program).1 Although other researchers had pointed to these developments

in explanations of the acceleration in the downward trend in infant mortality,

the question had not been studied in a multivariate context prior to my work.

The infant mortality rate is defined as deaths of infants within the first year

of life per thousand live births and has two components. Neonatal deaths pertain to deaths within the first twenty-seven days of life, and postneonatal deaths

occur from the twenty-eighth through the 364th day of life. Low birth weight

(weight less than 2,500 grams or less than 5.5 pounds) and prematurity (gestational age of less than thirty-seven weeks) are the two most important proximate

causes of infant and especially of neonatal mortality. These outcomes, particularly neonatal mortality and low birth weight, are featured in this section.

The first two papers in this section are motivated by the importance of

infant mortality as a key health indicator, by its trends between the mid-1950s

and the early 1980s, and by the role of the developments just mentioned in

those trends. My colleagues and I capitalize on variations in key determinants

of neonatal mortality at a moment in time (1971 in the first paper and 1977 in

the second) among counties of the United States to estimate the relative effects

of each one and their contributions to the decline in that rate (Grossman and

Jacobowitz 1981; Corman and Grossman 1985; Corman, Joyce, and Grossman

1987). We focus on neonatal mortality because the neonatal mortality rate was



D E T E R M I N A N T S O F I N F A N T H E A LT H



|



298



twice as large as the postneonatal mortality rate in 1982, and the former rate fell

much faster than the latter rate beginning in 1964. Hence, the decline in neonatal mortality rate accounted for almost 80 percent of the decline in the infant

mortality rate. Both papers apply estimated coefficients to trends in abortion

availability or use, public program availability or use, and neonatal intensive

care availability, to explain the downward trend in neonatal mortality from 1964

through the late 1970s or early 1980s.

Results in the two papers point to the growth in legal abortions as the single most important factor in reductions in white and black neonatal mortality

rates. Increases in hospitals with neonatal intensive care units also have sizable

effects for infants of both races. The extrapolations also point to the relevance of

Medicaid, organized family planning clinics, the WIC program, and community

health projects in accounting for reductions in the black rate. With the exception

of community health projects, all these factors also are relevant for whites.

I continued to work on infant health outcomes because of the availability of

data to estimate both health production functions and health demand functions.

Thus I had the opportunity to make important methodological contributions to

the literature on the determinants of health outcomes as well as to make important empirical contributions. In my economic formulation of the determinants

of health in my 1972 Journal of Political Economy (Grossman 1972a) paper

and in my 1972 National Bureau of Economic Research monograph (Grossman

1972b), I drew a distinction between these two outcome equations. The health

production function relates health to a set of endogenous inputs or choice variables including medical care services, diet, and exercise, as well as exogenous

determinants of the efficiency of the production process such as age and formal schooling completed. The health demand function relates health to income,

wage rates, input prices, and efficiency variables. It is obtained by replacing

the endogenous variables by their exogenous determinants and for that reason

is a reduced form equation.2 On the other hand, the production function is a

structural equation because endogenous variables appear on its right-hand side.

In my NBER monograph and my 1976 paper on the correlation between

health and schooling (Grossman 1976—the first paper in part 2), I focused

on the estimation of health demand functions as opposed to health production functions. I did so because unobserved biological factors, such as an individual’s exogenous health endowment and endowed rate of depreciation, and

hard-to-measure inputs, such as the avoidance of stress, can play major roles in

the determination of health outcomes. If an individual’s behavior is shaped in

part by knowledge of his or her endowments or if the unmeasured endogenous

inputs are correlated with the included inputs, then estimation of the health

technology will be biased and inconsistent.



299



|



I N T R O D U C T I O N T O PA R T 3



In the specific case of infant health outcomes, women who anticipate a

problematic birth outcome based on conditions unknown to the researcher may

seek out more remedial medical care, whereas women with positive expectations seek out less. This adverse selection in input use is emphasized by Mark

Rosenzweig and T. Paul Shultz (1983). It will understate the effect of, for example, the receipt of prenatal medical care early in the pregnancy on birth weight

or survival. Rosenzweig and Schultz use adverse selection to justify the estimation of birth weight production functions by two-stage least squares. In their

specification, delay in the receipt of prenatal care, maternal smoking during

pregnancy, previous number of live births, and mother’s age at birth are treated

as endogenous inputs in a micro level dataset. Hope Corman, Theodore J. Joyce,

and I use the same argument to fit neonatal mortality rate production functions

in the third paper in this section (Corman, Joyce, and Grossman 1987). These

functions differ from the neonatal mortality rate reduced form or demand function estimates in Corman and Grossman (1985). The primary distinction is that

county-specific input availability measures in the 1985 paper are used as instruments for input use measures in the 1987 paper. Results in our 1987 paper and

related results by Joyce (1987) and by Rosenzweig and Schultz (1983) point to

larger input use effects when the endogeneity of use is taken into account by

estimating production functions by two-stage least squares.

Adverse selection in input use, is not, however, the only source of bias

due to selection. The efficacy of prenatal care, for example, may be seriously

overstated if early care is but one form of healthy behavior. Pregnant women

who initiate care promptly may eat more nutritiously, engage in the appropriate

exercise, and use fewer drugs and other potentially harmful substances than

women who begin care later. The omission of these hard-to-measure inputs

tends to overestimate the impact of early prenatal care on infant health—an

example of favorable selection.

Moreover, the resolution of pregnancy itself may be characterized by

self-selection. With regard to this outcome, selection is favorable if women

whose fetuses have poor health endowments are more likely to obtain an abortion or if women who desire to make large investments in their infants are more

likely to give birth. On the other hand, selection is adverse if women who make

relatively small investments are more likely to give birth.

The use of an instrumental variable approach to correct for self-selection

in input use presupposes that this decision is characterized by adverse selection

and ignores the problem of self-selection in the resolution of pregnancies. In the

fourth paper in this section, Joyce and I (Grossman and Joyce 1990) approach

the problem differently and somewhat more generally. Following James J. Heckman (1979), we treat the estimation of infant health production functions and



D E T E R M I N A N T S O F I N F A N T H E A LT H



|



300



prenatal medical care demand functions as a general problem in self-selection.

Specifically, we test whether women who give birth represent a random draw

from the population of women who become pregnant. The widespread use of

induced abortion since its legalization by the Supreme Court in 1973 has permitted much greater choice in the number and timing of births. Thus the extent to

which a failure to incorporate the choice-based nature of micro vital records into

estimates of infant health production functions may bias the parameters of this

function is potentially large. Joyce and I hypothesize that the unobserved factors

that impact on the decision to give birth not only affect pregnancy outcomes but

also condition the behavior of women who choose to give birth during pregnancy

as well.

Our study is based on a cohort of pregnant women in New York City in

1984. In that year, 45 percent of all pregnancies to New York City residents

ended in induced abortions. We estimate a three-equation model. The first equation is the probability of giving birth, given that a woman is pregnant. With this

as our criterion equation, we test for self-selection in the infant health (measured

by birth weight) production function and in the prenatal medical care demand

function. Empirically, our estimates differ from those obtained by Rosenzweig

and Schultz (1983) because they use micro vital records on live births alone.

Not only does our methodology obviate the need to assert a priori whether

adverse or favorable selection is dominant, but the sign pattern of the residual

covariances indicates which type of selection characterizes both the decision to

give birth and the decision to initiate prenatal care promptly.

Because our framework includes an implicit equation for the probability

of becoming pregnant, we incorporate induced abortion as an alternative to

traditional methods of contraception into economic models of fertility control

(for example, Michael and Willis 1976; Hotz and Miller 1988). These models

emphasize the use of contraception to reduce the uncertainty associated with the

number and timing of births.

Induced abortion eliminates much of this uncertainty at a positive price.

By assuming that the prices of contraception and abortion have unmeasured

components that vary among women, we enrich the theoretical literature on

the optimal number and quality of children (for example, Becker and Lewis;

1973; Willis 1973) and gain a better understanding of the earliest indicator of

child quality—infant health—and the resources allocated to its production. In

particular, we show that the prices of contraception and abortion, as well as the

health endowment of the fetus, simultaneously influence decisions with regard

to pregnancy resolutions and input selection.

Joyce and I find strong evidence of selectivity bias in the birth weight

production function and prenatal demand equation among blacks but no



301



|



I N T R O D U C T I O N T O PA R T 3



evidence of such bias among whites. For the former group, the results suggest that the unobserved factors that raise the probability of giving birth are

positively correlated with the unobserved factors that decrease delay in the

initiation of prenatal care and increase birth weight. The sign patterns among

the residual covariances are consistent with a model that emphasizes the cost

of contraceptives. In particular, black women for whom the shadow price

of contraception is relatively high are more likely to abort a pregnancy than

their counterparts who face a lower shadow price and whose pregnancies

were more likely to have been planned. The latter group should consume

more prenatal care (delay less) and invest in other healthy behaviors that

improve birth weight.

One explanation for the racial differences with respect to selectivity bias is

that the shadow price of contraception is greater for blacks than it is for whites.

Further, the shadow price is apt to vary more among blacks than it does among

whites. Racial differences in contraceptive use and abortion in the 1980s and in

more recent years are consistent with this interpretation (Guttmacher Institute

2015, 2016).

The results for blacks indicate that women who aborted would have given

birth to lighter infants if they had selected the birth option and if they had had

the same mean values of the observed variables in the birth weight equation as

women who actually gave birth. One way to gauge the magnitude of the effect

is to compare it to that of an observed risk factor for birth outcomes. Among

blacks, complications due to maternal cigarette smoking reduce birth weight

by 187 grams or by 5.8 percent relative to a mean of 3,184 for pregnancies not

complicated by smoking. On the other hand, potential mean birth weight in the

abortion sample falls short of birth weight in the birth sample by 116 grams

due to unobserved inputs alone. This amounts to a differential of 3.7 percent

relative to the observed mean of 3,173 for all black women in the birth sample.

Thus the impact of unobserved healthy behaviors is almost two-thirds as large

in absolute value as the effect of smoking. Finally, if we allow for differences in

both observed and unobserved characteristics, the potential mean birth weight

of women who aborted would have 140 grams less than the observed mean birth

weight. This makes the impacts of unobserved healthy behaviors almost threefourths as large as that of smoking.

As I pointed out in the introduction to this book, Joseph P. Newhouse has

characterized the literature in health economics as consisting of two largely

nonoverlapping streams: one dealing with the determinants of the health of

the population and the other dealing with markets for health insurance and

medical services. By focusing on the impact of the introduction of National

Health Insurance (NHI) in Taiwan in 1995 on birth outcomes, Shin-Yi Chou,



D E T E R M I N A N T S O F I N F A N T H E A LT H



|



302



Jin-Tan Liu, and I contribute to both streams in the last paper in this section

(Chou, Grossman, and Liu 2014). There is enormous interest in the impacts

of NHI on health outcomes, but the very nature of this intervention, whereby

entire nations are covered universally, makes it difficult to estimate the health

impacts of the change. The experience of Taiwan, however, provides a potential laboratory for overcoming these limitations. Prior to NHI, government

workers possessed health insurance policies that covered prenatal medical

care, newborn deliveries, neonatal care, and medical care services received

by their children beyond the first month of life. Private sector industrial workers and farmers lacked this coverage. All households received coverage for

the services just mentioned as of March 1995. Therefore, the introduction

of NHI constitutes a natural experiment with treatment and control groups

that form the basis of our empirical design. The former group consists of

nongovernment-employed households, and the latter group consists of government-employed households. We expect that increases in infant health after

the introduction of NHI in the treatment group will exceed corresponding

increases in the control group.

Unlike in the United States, the postneonatal mortality rate in Taiwan is

higher than the neonatal mortality rate. Moreover, stringent requirements for

reporting births introduced in 1994 produced artificial upward trends in early

infant deaths. For those reasons, we limit our analysis to postneonatal mortality.

The introduction of NHI led to reductions in this rate for infants born in

farm households but not for infants born in private sector households. For the

former group, the rate fell by between 0.3 and 0.6 deaths per thousand survivors

or by between 8 and 16 percent. A large decline of between 3.4 and 6.8 deaths

occurred for preterm infants—a drop of between 20 and 41 percent. In the preNHI period, the postneonatal mortality rate of farm infants was approximately

23 percent higher than the corresponding rate of private sector infants. Hence,

our findings are consistent with the notion that the provision of health insurance to previously uninsured infants has larger effects on those born in poor

health than on others. Our result that the effects of NHI rise in absolute value

as the availability of medical care resources in the infant’s county of residence

rises is evidence that increases in medical care services received by infants

made eligible for insurance coverage by NHI may account for at least part

of the improvements in health outcomes that we observe. Farm families have

lower levels of health, education, and income than private sector families and

premature and low-weight infants are in worse health than other infants. Thus,

taken as a set, our findings suggest that health insurance improves infant health

outcomes of population subgroups characterized by low levels of education,

income, and health.



303



|



I N T R O D U C T I O N T O PA R T 3



NOTES

1. Federally subsidized maternal and infant care projects were authorized by the Social

Security Act of 1935 but were greatly expanded by the War on Poverty. Community

health centers, originally termed neighborhood health centers, were started by the Office

of Economic Opportunity as part of the War on Poverty.

2. Of course, the papers in part 2 and the references in that part question the endogeneity of

schooling.



REFERENCES

Becker, Gary S. 1981. A Treatise on the Family. Cambridge, MA: Harvard University Press.

Becker, Gary S., and H. Gregg Lewis. 1973. “On the Interaction Between the Quantity and

Quality of Children.” Journal of Political Economy 81(2, pt. 2): S279–S288.

Chou, Shin-Yi, Michael Grossman, and Jin-Tan Liu. 2014. “The Impact of National Health

Insurance on Birth Outcomes: A Natural Experiment in Taiwan.” Journal of Development

Economics 111(November): 75–91.

Corman, Hope, and Michael Grossman. 1985. “Determinants of Neonatal Mortality Rates in

the U.S.: A Reduced Form Model.” Journal of Health Economics 4(3): 213–236.

Corman, Hope, Theodore J. Joyce, and Michael Grossman. 1987. “Birth Outcome Production

Functions in the U.S.” Journal of Human Resources 22(3): 339–360.

Fuchs, Victor R. 1983. How We Live: An Economic Perspective on Americans from Birth to

Death. Cambridge, MA: Harvard University Press.

Grossman, Michael. 1972a. “On the Concept of Health Capital and the Demand for Health.”

Journal of Political Economy 80(2): 223–255.

——. 1972b. The Demand for Health: A Theoretical and Empirical Investigation. New York:

Columbia University Press for the National Bureau of Economic Research.

——. 1976. “The Correlation Between Health and Schooling.” In Household Production and

Consumption, ed. Nestor E. Terleckyj. Studies in Income and Wealth, Volume 40, by the

Conference on Research in Income and Wealth. New York: Columbia University Press for

the National Bureau of Economic Research, 147–211.

Grossman, Michael, and Steven Jacobowitz. 1981. “Variations in Infant Mortality Rates

among Counties of the United States: The Roles of Public Policies and Programs.”

Demography 18(4): 695–713.

Grossman, Michael, and Theodore J. Joyce. 1990. “Unobservables, Pregnancy Resolutions,

and Birth Weight Production Functions in New York City.” Journal of Political Economy

98 (5, pt. 1): 983–1007.

Guttmacher Institute. 2015. Contraceptive Use in the United States http://www.guttmacher

.org/pubs/fb_contr_use.pdf.

Guttmacher Institute. 2016. Facts on Induced Abortion in the United States. http://www

.guttmacher.org/pubs/fb_induced_abortion.html.Heckman, James J. 1979. “Sample

Selection Bias as a Specification Error.” Econometrica 47(1): 153–161.



D E T E R M I N A N T S O F I N F A N T H E A LT H



|



304



Hotz, V. Joseph, and Robert A. Miller. 1988. “An Empirical Analysis of Life Cycle Fertility

and Female Labor Supply.” Econometrica 56(1): 91–118.

Joyce, Theodore J. 1987. “The Impact of Induced Abortion on Black and White Birth Outcomes.” Demography 24(2): 229–244.

Michael, Robert T., and Robert J. Willis. 1976. “Contraception and Fertility: Household Production Under Uncertainty.” In Household Production and Consumption, ed. Nestor E.

Terleckyj. Studies in Income and Wealth, Volume 40, by the Conference on Research in

Income and Wealth. New York: Columbia University Press for the National Bureau of

Economic Research, 27–93.

Rosenzweig, Mark, and T. Paul Shultz. 1983. “Estimating a Household Production Function:

Heterogeneity, the Demand for Health Inputs, and Their Effects on Birth Weight.” Journal

of Political Economy 91(5): 723–746.

Shakotko, Robert A., Linda N. Edwards, and Michael Grossman. 1981. “An Exploration of

the Dynamic Relationship Between Health and Cognitive Development in Adolescence.”

In Contributions to Economic Analysis: Health, Economics, and Health Economics,

ed. Jacques van der Gaag and Mark Perlman. Amsterdam: North-Holland Publishing,

305–328.

Willis, Robert J. 1973. “A New Approach to the Economics of Fertility Behavior.” Journal of

Political Economy 81(2, pt. 2): S14–S64.

Xu, Jiaquan, Sherry L. Murphy, Kenneth D. Kochanek, and Brigham A. Bastian. 2016.

“Deaths: Final Data for 2013.” National Vital Statistics Report 64(2). Hyattsville, MD:

National Center for Health Statistics.



Variations in Infant Mortality Rates



SEVEN



among Counties of the United States

The Roles of Public Policies

and Programs

Michael Grossman and Steven Jacobowitz



ABSTRACT



The purpose of this paper is to shed light on the causes of the rapid decline in the

infant mortality rate in the United States in the period after 1963. The roles of

four public policies are considered: Medicaid, subsidized family planning services

for low-income women, maternal and infant care projects, and the legalization

of abortion. The most striking finding is that the increase in the legal abortion

rate is the single most important factor in reductions in both white and nonwhite

neonatal mortality rates. Not only does the growth in abortion dominate the other

public policies, but it also dominates schooling and poverty.

From 1964 to 1977, the infant mortality rate in the United States declined at an

annually compounded rate of 4.4 percent per year. This was an extremely rapid

rate of decline compared to the figure of 0.6 percent per year from 1955 to 1964.

The reduction in mortality proceeded at an even faster pace in the 1970s than in

the late 1960s (5.2 percent per year from 1971 to 1977 versus 3.8 percent per

year from 1964 to 1971).1

The period from 1964 to 1977 witnessed the introduction of Medicaid,

maternal and infant care projects, federally subsidized family planning services for low-income women, the legalization of abortion, and the widespread



D E T E R M I N A N T S O F I N F A N T H E A LT H



|



306



adoption of oral and intrauterine contraceptive techniques. These developments have been pointed to in discussions of the cause of the acceleration in

the downward trend in infant mortality (for example, Eisner et al. 1978; Lee

et al. 1980), but the question has not been studied in a multivariate context.

Moreover, the relative contribution of each factor has not been quantified. The

purpose of this paper is to estimate the impacts of public policies and programs

on infant mortality.



1. ANALYTICAL FRAMEWORK



Economic models of the family and household production developed by

Becker and Lewis (1973) and Willis (1973) provide a fruitful theoretical

framework to generate multivariate health outcome functions and to assess the

roles of social programs and policies in these functions. Ben-Porath (1973),

Ben-Porath and Welch (1976), Williams (1976), and Lewit (1977) have utilized the economic model of the family to study theoretically and empirically

the determinants of birth outcomes. Following these authors, we assume that

the parents’ utility function depends on their own consumption, the number of births, and the survival probability. Both the number of births and

the survival probability are endogenous variables. In particular, the survival

probability production function depends upon endogenous inputs of medical

care, nutrition, and the own time of the mother. In addition, the production

function is affected by the reproductive efficiency of the mother and by other

aspects of her efficiency in household production. Given the considerable

body of evidence that education raises market and nonmarket productivity,

one would expect more educated mothers to be more efficient producers of

surviving infants.

The above model calls attention to the important determinants of the survival probability and its complement, the infant mortality rate. In general, this

set of determinants is similar to that used in multivariate studies of infant mortality with different and fewer theoretical points of departure (for example,

Fuchs 1974; Williams 1974; Brooks 1978; Gortmaker 1979). Moreover, the

model provides a ready structure within which to interpret the effects of public

programs and policies on infant mortality.2 Thus, Medicaid and maternal and

infant care projects lower the direct and indirect costs3 of obtaining prenatal

and obstetrical care, which should increase the likelihood of a favorable birth

outcome and lower infant mortality. Federal subsidization of family planning

services, abortion reform, and the diffusion of oral and intrauterine contraceptive techniques (the pill and the IUD) reduce the costs of birth control and