Send all correspondence to:
Ted Joyce
National Bureau of Economic Research
365 Fifth Avenue, 5^th Floor
New York, NY 10016
Ted_joyce@baruch.cuny.edu

New York City has the largest concentration of births to foreign-born women of any vital registration area in the country. Over half of all births in the city are to women whose own birth took place outside the continental United States (New York City Department of Health 1999). The health of those infants in no small way predicts the future health of New Yorkers. Infants born sick face diminished chances of survival and increased chances of life-long illness. An infant weighing less than 5.5 pounds at delivery, the clinical threshold of low birth weight, is 25 times more likely to die in the first year of life and is more likely to be mentally retarded, suffer cerebral palsy, blindness, psychomotor problems and to fail in school than infants of normal weight ( Hack, Klein and Taylor 1995).

But the US would seem a poor destination for immigrants trying to maximize the survival of their offspring. The US ranks 25 in the world in infant mortality despite dramatic strides in reducing infant mortality over the last 30 years and remarkable strides in the past 10 (National Center for Health Statistics 1997). From 1965 to 1988 the infant mortality rate fell 3.9 percent per year with an acceleration in the rate of decline to 4.3 percent annually between 1989 to 1996 (US Bureau of the Census 1974,1993,1998). Despite these stellar achievements, the proportion of low birth weight infants has remained virtually unchanged over the same period. In other words, while our ability to save and care for sick infants has increased dramatically, the incidence of sick babies has improved minimally. Consequently, understanding the determinants of low birth weight has been a major area of policy and research over the past few decades.

The health of infants born to immigrants has been a separate vein of research mined for its potential insights into the determinants of low birth weight. It was observed almost three decades ago that deaths among Spanish surname infants in Texas were comparable to non-Spanish surname whites in the US and half as large as rates to infants of African American women (Gee, Lee and Forthofer 1976). What was remarkable was that Spanish surname women received less prenatal care and ranked lower on many common indicators of socio-economic status (SES) than non-Spanish surname whites (Markides and Hazula 1980). The gradient between SES and infant health—so evident among infants of non-Spanish surname women—appeared less relevant to Spanish surname parents. Attempts to explain differences in outcomes by the underreporting of Spanish-surname deaths were tested and finally rejected (Powell-Griner and Streak 1982; Henderson and Daudistel 1982; Eberstein and Pol 1982). So impressive was the health advantage of infants to Spanish-surname parents conditional on SES, that researchers described it an epidemiological paradox (Markides and Coreil 1986).

As vital data improved and it become possible to identify nativity as well as ancestry on birth certificates, a second version of the epidemiological paradox began to emerge. Researchers noticed a divergence in birth outcomes again by nativity but within ethnicity (Ventura and Taffel 1985). Infants of Mexican origin whose mothers were born in the United States, for example, had higher rates of low birth weight than infants of Mexican origin whose mothers were born in Mexico (Williams, Binkin and Clingman 1986; Scribner and Dwyer 1989; Guendelman et al. 1990). The divergence in outcomes within ethnicity among infants of US- and foreign born-women was not limited to Latinos, but was characteristic of women from China, other Asian countries and to a smaller extent Japan (Singh and Stella 1996; Alexander et al. 1996 ) .

Recent efforts to understand differences between US and foreign-born women of the same ethnicity focus on differences in behavior and health insurance, both of which are related to more general differences in acculturation. There is, for instance, significant variation in smoking, and illicit drug use by nativity. Pregnant Latinos in California are between five and 10 times more likely to use illicit drugs if they were born within as compared to outside the US (Vega et al. 1997). A similar pattern has been observed among the general population of Latinos in the US (Amaro et al. 1990; Vega et al. 1998). Moreover, the extent of drug use was directly correlated with the degree of acculturation as measured by English language usage.

Other explanations for the birth outcome advantage enjoyed by foreign-born women include a more supportive family structure. The proportion of births to unmarried women, a possible proxy for family cohesion, is generally higher among US- as compared to foreign-born Latinos ( Ventura and Taffel 1985; and Guendelman and English 1995; Collins and Martin 1998). More generally, researchers have suggested that Latin culture, as proxied by mother’s birthplace or English language usage, is protective of immigrant health. (Scribner and Dwyer 1989; Collins and Shay 1994).

Most studies that compare birth outcomes of US and foreign-born women use vital data and almost invariably make comparisons at a single point in time. The advantage of vital data is the large number of observations that are consistently collected across states. The disadvantage is the limited number of covariates. The potential for omitted variable bias is substantial. What some authors have attributed to immigrant culture, might in fact be explained by better data on income, insurance, stress, smoking, and illicit drug use. Comparisons of US and foreign-born women based on specially collected data have richer sets of covariates, but are often limited by sample size and geographical variation (Guendelman and English 1995; Scribner and Dwyer 1989). Efforts to directly control for acculturation have improved the predictive power of models, but they don’t elucidate the causal factors.

In this study I use data from New York City birth certificates to explore differences in perinatal health and health care utilization between US- and foreign-born women from 1988 to 1998. I have two objectives. The first is to characterize the financing of prenatal care, the utilization of prenatal care and birth outcomes of US- and foreign-born women in New York City during a period in which the federal government first expanded insurance coverage for pregnant women and infants and later restricted access for classes of immigrants. The second objective is use the change in these policies and the dramatic decline in prenatal drug use to explain the longstanding observation that the health of immigrants is often better than the health of the citizens they beget. As I detail below, changes in the financing of prenatal care and prenatal illicit drug use affected US- and foreign-born women differently. If access to prenatal care and illicit drug use are important explanations for why we observe large differences in newborn health between US- and foreign-born women, then I should see a widening or narrowing in birth outcomes by mother’s birthplace in the wake of these events.

The distinguishing characteristic of the analysis is the use of panel data to analyze perinatal differences between US- and foreign-born women. Previous work on the differences between US- and foreign-born women has relied exclusively on cross-sectional differences between the two groups. Panel data provide time-series variation in key indicators of maternal and infant well-being. Thus, I can ask whether differences by nativity have been widening or narrowing over the period, and if so, look for explanations. Another advantage of panel data is that time-invariant differences between US- and foreign-born women can be differenced out by analyzing changes in outcomes pre and post exogenous shifts in policy or the environment. For example, acculturation is an important confounding factor in cross-sectional analyses of perinatal differences between US- and foreign-born women. Even with an index of acculturation, it is often correlated with other determinants of infant health that are hard to measure. However, as long as the effect of acculturation is constant, then it can be netted out by analyzing changes overtime.

Three major and plausibly exogenous events in New York City occurred over the past decade. Each had a potentially significant and differential impact on the perinatal health and health care utilization of US- and foreign-born women and infants. The first was the precipitous decline in the use of crack cocaine beginning around 1988; the second change was the expansion in Medicaid eligibility for pregnant women and the streamlined enrollment procedures initiated in 1990; and the third major event was the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) of 1996, more commonly referred to as welfare reform. The key is that each event appears to have affected groups of foreign- and US-born women differently.

The crack/cocaine epidemic that peaked around 1988 in New York City had a profound impact on the birth outcomes of women in poor communities (Joyce, Racine and Mocan 1992). There is substantial evidence that illicit drug use was far more prevalent among US- as compared to foreign-born pregnant women (McCalla et al. 1992; Vega et al. 1993; Noble et al. 1997; Vega et al 1997). One prediction, therefore, is that we should observe a narrowing of the difference in birth outcomes between US-born and foreign-born women as the epidemic waned. If this finding is obtained, then we would have important evidence that behavioral differences between US- and foreign-born women contribute to the variation in outcomes by nativity.

The second major change was the unprecedented expansion in Medicaid eligibility for pregnant women beginning in 1990. The expansion in eligibility was the New York State’s response to a series of Federal initiatives designed to improve the utilization of prenatal and thus birth outcomes of poor and near-poor women (Currie and Gruber 1996). Data on take-up of Medicaid by eligible women, however, suggests that other aspects of the Medicaid reforms besides the increase in income eligibility thresholds may have been more effective at increasing participation. The legislation allowed states to streamline enrollment procedures by dropping the requirement that applicant’s assets be included in the determination of eligibility and by permitting obstetrical providers to declare a woman presumptively eligible for Medicaid pending completion of a more detailed application. Moreover, providers that participated in the Prenatal Care Assistance Program (PCAP) were allowed and even encouraged to assist applicants in their claim for medical assistance. These changes in the enrollment procedures appear to have had a dramatic effect on the participation of foreign-born women (Joyce 1999). As a result, I should observe greater increases in prenatal care utilization among foreign- as compared to US-born women. If the association between appropriate receipt of prenatal care and favorable birth outcomes is causal, then differences in birth outcomes between US- and foreign-born women should actually widen.

Finally, PRWORA ended the guarantee of cash assistance to eligible women but did not revoke federal entitlement to Medicaid for US citizens. However, PRWORA did curtail eligibility for Medicaid and other federal benefits for undocumented immigrants and those that arrived legally after August 1996. Despite PRWORA, New York State continues to fund prenatal care for all immigrants regardless of status due to a long-standing course case. Nevertheless, reports by providers and advocates suggested that the legislation engendered fear and confusion in immigrant communities and may have discouraged foreign-born women from seeking benefits such as Medicaid or obtaining appropriate prenatal care (Jessop et al. 1998). In other words, among immigrants, welfare reform may have reversed the potential gains in prenatal care utilization achieved by the Medicaid eligibility expansions. If true, welfare reform may add to our understanding of whether differential access to health insurance by mother’s birthplace explains differences in perinatal outcomes between US-and foreign-born women.

1. Data

All data are from New York City birth certificates. New York City is the only municipality in the US that is a separate vital registration area. All other vital registration areas are states. As with birth certificates nationally, the New York City birth certificates contain information on mother’s race, ethnicity, and the country of birth. There is also information on the month in which the woman obtained her first prenatal care visit, the number of visits, and the birth weight and gestational age of the infant at delivery. There is also standard demographic information on the mother’s age, marital status, completed schooling, and previous live births.

Most birth certificates nationally have indicators for whether the mother smoked during pregnancy. In New York City, the smoking field indicates whether the mother smoked at least one-half a pack of cigarettes daily. This field was changed in 1995 to include the number of cigarettes per day. To create a consistent series we use a dichotomous indicator that is one if the mother smoked at least 10 cigarettes per day from 1995 to 1998 and we link this to the indicator of one-half pack or more as reported from 1988 to 1994.

Unlike other birth certificates nationally, the New York City birth certificate has indicators for specific illicit drugs: cocaine, marijuana, heroin and methadone. We will use these indicators to measure the decline in the crack epidemic. The prevalence of illicit drug use as obtained from the birth certificate underestimates true exposure by as much as 60 percent (Joyce et al. 1995). If the misclassification were random, or non-differential, then estimates of the effect of cocaine on birth outcomes would be biased towards zero. Evidence suggests, however, that the measurement error associated with illicit drug use is nonrandom in that only the fetuses of heavy users are classified as exposed. Under this form of misclassification, the estimated effect of exposure on infant health may actually overestimate the true impact of illicit drug use (Kaestner, Joyce and Wehbeh 1996). The upshot is that if the underreporting is relatively consistent over time it may still capture differences in the prevalence of heavy use by nativity.

Another item on the New York City birth certificate that is contained in only few vital registration areas is the method of payor. Thus, I know whether the birth and presumably prenatal care, were paid for by Medicaid, an HMO, another third party or was self pay. I interpret self pay as an indication that the woman was uninsured.

a) Outcomes

I analyze three outcomes: the financing of prenatal care and delivery, the utilization of prenatal care and birth outcomes. For financing, I use two outcomes: the probability that the delivery or prenatal care was covered by Medicaid or the probability the woman was uninsured. The New York City birth certificate does not distinguish between financing at prenatal care and financing at delivery. The two might differ if the woman was not enrolled in Medicaid until delivery. This is more likely to occur before 1990 when the state’s Medicaid program had not yet instituted presumptive eligibility, a procedure by which the provider can bill the Medicaid program for care administered during pregnancy pending completion of the woman’s Medicaid application. The provider is paid regardless of whether the woman is ultimately deemed ineligible. As I show below, the rapid uptake in Medicaid among foreign-born women suggests that the program was enormously successful.

I use two measures of prenatal care. The first relates to the time during pregnancy of the first prenatal visit. Specifically, I use a dichotomous indicator that is one if the woman initiates prenatal care in the first four months of pregnancy and zero otherwise (Kotelchuck 1994). I also use the number of prenatal care visits obtained during pregnancy. The timing measure is a useful proxy of a woman’s access to obstetrical care and her motivation to achieve a health birth outcome. Once enrolled in prenatal care, the number of visits is set by protocol as described by the American College of Obstetrics and Gynecology (ACOG). Many women may not comply with the suggested number of visits and other women need more care. Initiation of prenatal care is determined by the mother, and is arguably a more sensitive indicator of access and motivation.

As to birth outcomes, I focus primarily on the incidence of a low birth weight birth, an infant at delivery that weighs less than 2,500 grams or 5.5 pounds. The primary advantage of low birth weight is as a predictor of newborn survival. In 1998, the infant mortality rate was 53.3 per 1000 live births for newborns less than 2500 grams as compared to 1.8 deaths per 1000 live births for infants greater than 2,500 grams. However, low birth weight is a broad measure that includes infants born prematurely, but of normal weight for their gestational age, as well as infants born at term but whose growth was retarded. The causes of prematurity are not well known (Hauth et al. 1998). In contrast, there is a great deal of evidence that smoking inhibits fetal growth (Sexton and Hebel 1984). There is also strong evidence that cocaine and other illicit substances are related to fetal growth retardation. Another reason to study birth weight is that it is one the most accurately measured items on the birth certificate. Gestational age, by comparison, is measured with greater error because it depends on the accurate recall of a woman’s last menstrual period. Ultrasound, a more accurate measure of gestational age, is now commonly used to estimate gestation and the expected date of delivery. Unfortunately, birth certificates do not indicate how gestational age was determined. Moreover, the use of sonography to date the fetus has grown over time. There is evidence from Canada, for example, that the incidence of prematurity has risen overtime due in part to more accurate assessments (Joseph et al. 1998).

A final reason that recommends the study of low birth weight is illustrated in Figure 1. As is evident, the decline in infant mortality (deaths in the first year of life divided by live births in that year) over the past decade has far exceeded the decline in the incidence of low birth weight. The point is that improvements in newborn survival have come about because of the ability to save infants born prematurely and of extremely low birth weight. The underlying morbidity of infants, as measured by the percent of low birth weight births, has declined relatively little. My interest in this study lies in understanding factors that affect the healthiness of newborns and why underlying morbidity varies by mother’s birth place.

2. Econometric specification and considerations

Although much of the analysis is descriptive, the data are novel and at times startling. Changes in the financing of prenatal care by race/ethnicity and nativity have not previously been presented. I also present data on the proportion of birth exposed to tobacco and illicit drugs, primarily cocaine, also trends that have not been previously displayed. The results suggest that differences in birth outcomes by nativity reflect difference in adverse behaviors. They also suggest that the growth in Medicaid financed deliveries has not led to an improvement in birth outcomes despite substantial increases in early prenatal care.

The third objective of the study, however, is to analyze whether the changes in drug use, Medicaid eligibility and welfare reform have altered differences in prenatal care utilization and birth outcomes between US and foreign born women. To address these questions, I use a "difference-in-difference" methodology (Gruber 1994) in which changes in outcomes pre and post a policy change among foreign-born Latinos are compared to changes pre and post among US-born Latinos. In a regression format a simplified specification for low birth weight (LBW) can be written as follows:

(1)

where X_it represents maternal characteristics such as age, marital status, parity, and schooling. In other specifications I include behaviors such as smoking and illicit drug use; Pol_t is an indicator of the year the policy became effective; For_it is one if the woman was not born in the continental United States and zero otherwise; and (Pol_t x For_it) is the interaction of policy and nativity. I present results that show the changes in each outcome before and after the policy for each ethnic group and nativity cell. I refer to this change as is the first difference. I also show changes in each outcome for foreign- relative to their US-born counterparts. This is termed the difference-in-differences (DD) between foreign- and US-born women. In the above specification, for instance, the change in low birth weight among foreign-born women following the Medicaid expansions is (a₁ + a₃); the change among US-born women is simple a_1. Thus, the differential change among foreign- as compared to US-born women is a₃, or the adjusted difference-in-differences (DD) estimate.

Most of the outcomes are binary. Therefore I use logistic regression to estimate the parameters in equation (1). To facilitate interpretation, I present estimates of the marginal effects, d , or the percentage point change in the outcome associated with each covariate. In terms of equation (1), the difference-in-differencesestimates can be expressed as d = LBW₁- LBW₀.where LBW₁ is the predicted mean rate of low birth weight evaluated at, for instance (Pol x For)=1 and with all other covariates set to their mean values and LBW₀ is the same except (Pol x For)=0. Standard errors are obtained by the delta method (Greene 1997) and are computed with a special ado file obtainable from the Stata Corporation. (College Station, TX. V.6).

The results are organized as follows. I first present a general overview of birthing by race, ethnicity and nativity in New York City between 1988 and 1998. In the subsequent sections I focus exclusively on Latinos. As I explain below, Latinos are the only ethnic groups with sufficient birth to US-born women to use a comparison group. I also compare outcomes of foreign-born Latinos to those of Puerto Ricans. The objective of these analyses is to understand differences in outcomes between US- and foreign-born Latinos. Toward this end I test whether responses to major health initiatives and changes in substance abuse differed by mother’s place of birth.

1. Summary Statistics of US- and Foreign-born Women.

Table 1 displays selected characteristics and perinatal outcomes of New York City residents that gave birth in 1988 and 1998 by race and ethnicity. I show births to the three largest groups of foreign-born women: Latinos, non-Latino women from the Caribbean and South America and Asians. For comparative purposes I also show births to US-born non-Latino whites and blacks. The largest number of births are to Latinos, 37,992 in 1998. The number of births to Puerto Ricans, still the largest Latino subgroup, has fallen from 18,137 in 1988 to 12,186, while births to Mexicans have grown more than three-fold over the same decade. The category, Other Latinos, is the second largest group in 1998 and includes women from other Central and South American countries. The three largest groups of births to women from non-Spanish speaking countries of the Caribbean and South America are Jamaica, Guyana and Haiti. China and India are the two most important countries of origin for Asians. Finally, births to US-born non-Latino whites and blacks have fallen by more than 6,000 or between 24 and 32 percent between 1988 and 1998.

The percent of births to US-born Latinos is small. I refer to Puerto Ricans born in the continental US as mainland-born and Puerto Ricans born on the island as island-born. Three-quarters of all births to Puerto Ricans in 1998 were to women born on the mainland, up from 61 percent in 1988. Fourteen percent of births to Dominicans in 1998 and 21 percent of births to other Latinos were to women born in the US. Among other Caribbean and South American women and Asians the percent of births to US-born women is less than 5 percent, which underscores the recency of many immigrants. This fact also explains my focus on Latinos in analyzing births to US and foreign-born women: there were simply insufficient births to US-born women within these two subcategories to use as a comparison group to their foreign-born counterparts.

There are important differences among the women who give birth in the three main immigrant groups. The percentage of births financed by Medicaid is greatest among Latinos. In 1998, 73.1 percent of all births to Latinos were financed by Medicaid as compared to 55.3 among other Caribbean and South American women and 54.7 percent among Asians. More interestingly, the change in Medicaid-financed births is staggering. Although the increase among Latinos is 18.2 percentage points (73.1-54.9) the change among Mexicans and Other Latinos is 53 and 33 percentage points respectively. Similarly among Asians, the change is equally staggering: from 5.8 percent in 1988 to 54.7 percent in 1998. As I will argue below, changes in enrollment procedures associated with the expansions in Medicaid eligibility thresholds in 1990 greatly facilitated uptake among foreign-born women. By contrast the change in Medicaid- finance births among Puerto Ricans between 1988 and 1998 is relatively modest and reflects their greater and more stable participation in Aid to Families with Dependent Children (AFDC).

Teen and out-of-wedlock childbearing are two other social indicators that often distinguish US and foreign-born women. They are also important risk factors for adverse birth outcomes. The proportion of births to teens was 15.5 percent for all Latinos in 1998, 7.5 among non-Latino Caribbeans, and 1.5 percent among Asians. The percent unmarried followed a similar pattern: 63.8, 52.5 and 13.2 respectively. Schooling differences are also pronounced. Again Latinos have the highest percentage of births to women with less than a high school education. Yet, despite the apparent disadvantage of Latinos in these broad indicators, the proportion of low birth weight births was only 6.8 percent in 1998 as compared to 9.4 among non-Latino Caribbean women, 4.6 among Asians, 12.0 among US-born non-Latino blacks, and 4.4 among US-born, non-Latino whites.

2. Times-series of Perinatal Outcomes

One question that motivates this analysis is whether differences in perinatal outcomes among US- and foreign-born women have narrowed over the decade. One mechanism by which this could occur is through the increase in Medicaid-financed births. As shown in Table 1, the percent of births financed by Medicaid rose dramatically among foreign-born women. If the growth in Medicaid was accompanied by a decline in women that were uninsured, and if this generated an increased demand for prenatal care, then differences in birth outcomes between US- and foreign-born women may widen.

Figure 2 shows the percent of Latino births financed by Medicaid between 1988 and 1998 by race, ethnicity and nativity. There are separate series for mainland- and island-born Puerto Ricans (PRUS and PRIS), US- and foreign-born Dominicans (DOUS and DOFO), and US- and foreign-born other Latinos including Mexicans (OLUS and OLFO). For comparative purposes I also show time-series for US-born, non-Latino whites and blacks (WHUS and BLUS). What is immediately apparent is the difference between US whites and all other groups. At its peak in 1996, only 18.3 percent of births to US whites (WHUS) were financed by Medicaid. The other significant observation is that growth in Medicaid- financed births is greatest among other Latinos, both foreign- and US-born. Among foreign-born Latinos, 36.7 percent of births were financed by Medicaid in 1988 as compared to 83.8 percent in 1997, a rise of 47 percentage points! In contrast, both US-born blacks and mainland-born Puerto Ricans began and ended the decade at approximately 60 percent of births financed by Medicaid.

A rise in Medicaid-financed births, however, unaccompanied by a decline in the percent uninsured would suggest that the expansion in government-financed prenatal care simply substituted public for private insurance. If true, then one would expect little change in prenatal care utilization and birth outcomes. The time-series in Figure 3 suggest that there has been a clear decline in the percent of births to women that were uninsured and a remarkable convergence in insurance coverage among US- and foreign-born women. Nevertheless, the percentage-point decline in uninsured women was approximately half the percentage- point rise in Medicaid-financed births, suggesting substantial crowd out. As with Medicaid, other Latinos that were foreign-born experienced the greatest change. In 1988, 23.3 percent of births to foreign-born Latinos other than Dominicans and Puerto Ricans lacked insurance coverage at delivery. The comparative figure for mainland-born Puerto Ricans was 10 percent. By 1998, the percentage of births to uninsured women stood at 4.5 and 2.3 among foreign-born other Latinos and mainland-born Puerto Ricans, respectively.

The next question, therefore, is whether the spread of insurance coverage increased utilization of prenatal care, a major policy objective of federal and state initiatives in the mid to late 1980’s (Currie and Gruber 1996; Dubay et al. forthcoming). Early initiation of prenatal care is a commonly used indicator of access to prenatal care. Early initiation of prenatal care is also associated with higher levels of schooling as well as low rates of smoking and illicit drug use. The simple association between early prenatal care and favorable birth outcomes is also positive although many question whether it is causal (Joyce 1999). The time-series data in Figure 4 show the annual percent of births to women that initiated prenatal care in the first four months of pregnancy. Again, US-born white women stand alone from the other groups in both the level of early initiation—over ninety percent—and the relatively modest change over the decade. All other groups of women are similar in both the level of early initiation and the change over time. Taken together, early initiation of prenatal care increased approximately 10 percentage points between 1988 and 1998.

Increased insurance coverage and earlier initiation of prenatal care would seem important accomplishments of the Medicaid eligibility expansions. However, the most important objective of these initiatives was to improve the health of newborns. Times-series data on the percent of low birth weight births are presented in Figure 5. Several features merit comment. First there appears to be three cluster or racial and ethnic groups. US-born blacks have the greatest rate of low birth weight. In 1987, 16.5 percent of all births to US-born blacks in New York City were low weight, an incidence 2.75 times greater than that of US-born whites in the same year. The second cluster contains mainland- and Island-born Puerto Ricans as well as US-born other Latinos. For all three series the rate of low birth weight peaks around 1987 at approximately 10 percent. The final cluster contains US-born whites as well as foreign-born Dominicans and other Latinos. The incidence of low birth weight is about six percent among these groups.

The second observation is that these series are consistent with what is frequently referred to as the "epidemiological paradox." As displayed in Figure 5, the incidence of low birth weight among foreign-born Latinos is essentially equal to that of US-born whites despite, as I showed in Table 1, higher levels of poverty (as proxied by Medicaid enrollment), out-of-wedlock childbearing and lower levels of completed schooling among Latinos.

The last observation pertains to the decline in the rate of low birth weight after 1987 among US-born blacks, Puerto Ricans and US-born other Latinos. The series in Figure 5 begin in 1985 which shows a steep rise that peaks in 1987. Many believe the rise was due to the crack/cocaine epidemic (Joyce 1990; Joyce, Racine and Mocan 1995). After 1987, there is a considerable decline among US-born blacks from a peak of 16.5 to 12.0 percent in 1998, a fall of 4.5 percentage points or 27 percent. The fall among Puerto Ricans is approximately two percentage points whereas the decline among other US-born Latinos is 3.5 percentage points. There is also a notable decline among US-born whites from a high of 6.0 to 4.4 percent.

The decline in low birth weight is roughly coincident with the fall in the percent of births to uninsured women (Figure 3) and the rise in early initiation of prenatal care (Figure 4). However, the greatest increases in insurance coverage and prenatal care occurred among foreign-born other Latinos, a group that experienced relatively little improvement in low birth weight. US-born blacks, by contrast, showed relatively little gain in insurance coverage and average gains in early prenatal care.

Another explanation for the decline in low birth births among primarily US-born women is the fall in prenatal use of cocaine. Figure 6 shows the percent of births exposed to cocaine as recorded on birth certificates. Several points are notable. First, the level of exposure by race and ethnicity is consistent with the incidence of low birth weight. US-born blacks have the highest level of exposure, 7 percent in 1989, and the highest rate of low birth weight, 15.6 percent in 1989. Puerto Ricans, first mainland- and then island-born, have the second highest level of prenatal cocaine use, and the second highest rate of low birth weight. The percent of births exposed to cocaine among US-born Dominicans is highly unstable, due in part to the small annual number of births in this category. The second point to notice is that the decline in cocaine is also consistent with the decline in rate of low birth weight (Figure 5). However, as with prenatal care, the timing of the downturn in cocaine lags the downturn in low birth weight.

The use of antenatal cocaine as recorded on birth certificates clearly underestimates true exposure (Joyce et al. 1995). Moreover there is no indication of when or how many times exposure occurred. In addition, who is tested or asked about drug use may be a function of prejudice based on race or class or the protocol of the facility. Despite these limitations, the level of exposure by ethnicity is remarkably consistent with more rigorous studies of substance use among pregnant women. In the most comprehensive attempt to measure prenatal drug use, Vega et al. (1993) tested the urine of over 29,000 women that delivered in a stratified sample of California hospitals in 1992. Cocaine was found in 7 percent of the urine of non-Latino blacks, as compared to .5 percent among Latinos and .6 percent among non-Latino whites. In a follow-up study Vega et al. (1997) stratified drug use among Latinos by nativity. They found that exposure was between 5 and 10 times greater among those born in the US as compared to those born outside the US. Since there are relatively few Puerto Ricans in California, the most analogous comparable groups in New York City are US- and foreign-born other Latinos. As shown in Figure 6, exposure to cocaine among the US-born other Latinos was about 2.5 percent in 1988, an order of magnitude greater than the exposure among their foreign-born peers.

Another important substance strongly related to low birth weight is smoking. Figure 7 displays data on the proportion of births to women that smoked at least one-half a pack of cigarettes per day during pregnancy. As with cocaine, non-Latino blacks and Puerto Ricans have the greatest incidence of smoking followed by non-Latino whites. The lowest incidence of smoking occurs among Dominicans. The other notable feature in Figure 7 is the marked fall off in prenatal smoking after 1994. New York State raised excise taxes on cigarettes by 27 cents in 1993, after which smoking rates among all residents declined significantly (Evans, Ringel, and Stech 1999). The decline among pregnant women is consistent with this overall fall.

In summary, the time-series offer several insights: 1) there was a large increase in Medicaid-financed births over the decade; 2) growth in Medicaid-financed births was greatest among foreign-born women; 3) early initiation of prenatal care also increased but the rise was relatively uniform among all women and; 4) there were important decreases in the rate of low birth weight but they appeared less connected to the increased use of prenatal care and more consistent with decline in the prenatal use of cocaine and possibly tobacco.

3. Medicaid eligibility expansion in 1990.

In this section I present results from the analysis of the impact of the Medicaid eligibility expansions and enrollment reforms on the financing of prenatal care, the utilization of prenatal care and birth outcomes among US- and foreign-born Latinos in New York City between 1988 and 1992. As data in Figures 2 and 3 showed, the uptake in Medicaid and the fall in percent of births to uninsured women was substantially greater among foreign-born as compared to US-born Latinos. One result, therefore, should be a greater increase in prenatal care utilization and possibly greater improvements in the rate of low birth weight among foreign- relative to US-born women. The question is important for two reasons. First, previous analyses of the Medicaid expansions have had difficulty finding effects of the expansions on prenatal care and birth outcomes due to the often modest increases in Medicaid take-up over the period under study (Piper et al. 1990; Haas et al. 1993). An advantage of analyzing changes among foreign-born women in New York City is that the take-up among the foreign-born was substantial, which increases the power to detect statistically significant effects on prenatal care and birth outcomes. The take-up among foreign-born women was so large because relatively few foreign-born women received cash assistance through AFDC in the pe-expansion period. AFDC was the pre-expansions mechanism by which the vast majority of pregnant women obtained Medicaid. An important and often overlooked component of the expansions was the leeway states were given to simplify enrollment. In New York, obstetrical providers participating in the Prenatal Care Assistance Program (PCAP) were allowed to declare a woman seeking prenatal care presumptively eligible for Medicaid based on a one-page assessment form. The obstetrical provider, most often a hospital outpatient clinic, then helped the woman process her application for Medicaid. Moreover, all pregnant New York residents were eligible for Medicaid with federal sharing even if they were undocumented aliens, a policy unique to New York that has twice been upheld in court in response to suits by the Federal government (Lewis v Grinker).

Additional evidence for the importance of the expansions to foreign-born Latinos comes from a linkage of Medicaid administrative data with New York City birth certificates. The data contain all births to Latinos in the third quarter of 1989 and 1991. The data distinguish women on cash assistance, primarily AFDC, from women who receive Medical assistance only. The breakdown for Latinos stratified by year of birth and nativity is presented in Table 2. The primary insight is that between 63 and 80 percent of US-born Latinos that gave birth in 1989 received Medicaid because they participated in AFDC. Moreover, in the year after the expansions, the proportion on AFDC remained essentially unchanged. The opposite was true for foreign-born women. Fifty-one percent of foreign-born Dominicans and 83 percent of other Latinos were covered by Medicaid through the Medical Assistance Program in 1989. Both figures are higher by 1991, the year after the expansion. The higher participation of US-born Latinos, primarily Puerto Ricans, in AFDC explains their relatively modest increase in Medicaid-financed births between 1988 and 1992 as shown in Figure 2.

The upshot of this rather lengthy discussion is that the large increase in Medicaid take-up among foreign-born Latinos provides an important source of exogenous variation with which to identify the differential effect of the expansions on US- and foreign-born Latinos. A major objective of this analysis is to explain why perinatal outcomes of US- and foreign-born women differ so much. The Medicaid expansions provide a natural experiment with which to assess whether differences in insurance coverage contribute to differences in prenatal care utilization. If so, and if more appropriate prenatal care improves birth outcomes, then the gap in low birth weight between US- and foreign-born women should have increased in the early 1990’s. An ancillary question is why the birth outcomes of Puerto Ricans compare so poorly to those of foreign-born Latinos from other countries of the Caribbean as well as Central and South America. Puerto Ricans as US citizens had much greater access to cash assistance programs such as AFDC and thus, were more likely to be insured during pregnancy than were Latinos from other countries. The Medicaid expansions essentially eliminated this advantage for Puerto Ricans, which raises doubts as to whether access to prenatal care contributes to the difference in birth outcomes between Puerto Ricans and other Latinos born outside the US.

Before presenting estimates associated specifically with Medicaid expansions, I first show what might be termed naïve regressions of prenatal care financing, prenatal care utilization and birth outcomes over the period of the expansions, 1988-1992. I say naïve because this is the type of regression typically used to explain differences between US- and foreign-born women. The estimates in Table 3 are the changes in the probability of each outcome associated with maternal characteristics and behaviors available on birth certificates. The estimates are somewhat more interesting than standard cross-sectional estimates in that they cover a period of substantial change in the financing of prenatal care. With respect to the financing of prenatal care and delivery the probability that a woman was on Medicaid is 7.2 percentage points greater among teen than women 20 to 34 years of age [column (1)]. Unmarried women are 21.5 percentage points more likely than are married women to be on Medicaid. Results for parity and schooling are as expected.

The estimates also indicate that Dominicans are more likely to be on Medicaid than are Puerto Ricans; other Latinos are less likely. The differences are substantial, about 10 percentage points. However, both groups, Dominicans and other Latinos are more likely to be uninsured relative to Puerto Ricans, suggesting that the latter have more insurance through private coverage. The interactions between ethnicity and nativity indicate that US-born Dominicans are more likely to be on Medicaid than their foreign-born counterparts whereas other Latinos born in the US are less likely.

Other results from Table 3 indicate that foreign-born Dominicans and other Latinos are less likely than their US-born counterparts to initiate prenatal care early, but there are no differences in prenatal care visits per pregnancy. However, differences in birth outcomes are substantial. The rate of low birth weight among US-born other Latinos is 2.7 percentage points greater than their foreign-born counterparts. The difference by nativity between Dominicans is 1.3 percentage points. Differences in low birth weight are even larger when I compare foreign-born Dominicans and other Latinos to Puerto Ricans: between 3.1 and 3.5 percentage points if the comparison is with Puerto Ricans born on the island and between 3.9 and 4.3 when compared with Puerto Ricans born on the mainland. All estimates are from specifications that exclude smoking and illicit drugs [column (5)]. Inclusion of the latter covariates narrows these differences by about 20 percent.

Estimates based on specifications in Table 3 do not fully exploit the natural experiment afforded by the Medicaid eligibility expansions. Thus, they offer limited insight as to the differential impact of the expansions on US- as compared to foreign-born women. I therefore estimated a specification based on equation (1) presented in the Methods Section. I use 1988-89 as the pre-expansion period and 1991-92 as the post-expansion period. The outcomes are the same as in Table 3. Selected results are presented in Table 4. The top panel displays changes in each outcome pre and post the expansion. I label these first differences. Estimates in the lower panel are the difference-in-differences (DD) or the changes in outcomes among the foreign-born less changes among their US-born counterparts. I also present DD estimates between foreign-born women and mainland-born Puerto Ricans.

For other Latinos and Dominicans, changes in the probability that a birth was financed by Medicaid is substantially greater among foreign- as compared to US-born women (Table 4). For example, the probability of a Medicaid-financed birth increased 23.4 percentage points among foreign-born Latinos and only 6.8 percentage points among their US counterparts [top panel, column (1)]. The DD estimate for this group is .166 [.234-.068] as shown in the lower panel of column (1). For Dominicans, Medicaid financed births increased 20 percentage points among the foreign-born and 13.8 percentage points among the US-born, a DD of 6.2 percentage points. Puerto Ricans, regardless of birthplace, experienced increases of 9.5 percentage points in Medicaid financed births.

Changes in the probability that a woman was uninsured at delivery are shown in column (2). The direction of the change but not the magnitude is commensurate with the rise in Medicaid-financed births. The probability that other Latinos born outside the US were uninsured at delivery fell 6.4 percentage points in the two years after the expansions, only a fourth of the rise in Medicaid coverage. The likelihood that a foreign-born Dominican was uninsured at delivery declined 8.3 percentage points, again, a fall substantially less than the rise in Medicaid coverage. The same pattern persists for Puerto Ricans. The DD estimates reflect these more modest changes in the proportion of women that were uninsured. Only among other Latinos does the decline among the foreign-born exceed the decline among US-born Latinos by a significant amount [-.063].

Changes in probability that a woman initiated prenatal care in the first four months of pregnancy are displayed in the top panel of column 3 in Table 4. Changes in the number of prenatal care visits are shown in column 4. There are significant increases in early initiation of prenatal care that range from 5.8 percentage points for foreign-born Dominicans to 1.4 percentage points for island-born Puerto Ricans [top panel, column (3)]. None of the DD estimates, however, is significant. The change in prenatal care visits are similar and although some of the DD estimates are statistically significant, the magnitude of the changes—about 0.2 of a visits—are not large. The results for prenatal care would seem at odds given the large DD changes for Medicaid-financed births, but they are consistent with the more modest DD estimates of births to uninsured women. This point has often been missed in analyses of the Medicaid expansions. Although the changes in Medicaid-financed births are dramatic, only changes from no insurance to some insurance are likely to alter the demand for prenatal care in any significant manner.

Columns 5 and 6 in Table 4 displays results from regressions of low birth weight. The two specifications differ in that the estimates in column 6 are adjusted for smoking and illicit drugs during pregnancy. The only significant change between 1988-89 and 1991-92 occurs among mainland-born Puerto Ricans. The rate of low birth weight falls 0.7 percentage points, a relative decline of about 7 percent when measures of smoking and illicit drug use are excluded. However, there is no significant change in low birth weight when these covariates are included [top panel, column (6)]. This last result is noteworthy. Recall that the change in uninsured women and early prenatal care among mainland-born Puerto Ricans was less than was found among foreign-born Dominicans or other Latinos. Neither of the latter groups, however, experienced any improvement in low birth weight. This suggests that we must look elsewhere for explanations of the improvement in low birth weight among mainland Puerto Ricans. One hypothesis that I pursue below is that the decline in illicit drug use among mainland-born Puerto Ricans over this period accounts for the improvement in the incidence of low birth weight. The fact that the change in low birth weight fell to 0.3 percentage points [column (6)] among mainland-born Puerto Ricans when we included controls of smoking and drugs is consistent with this hypothesis.

In summary, the Medicaid eligibility expansions appear to have had a large effect on the proportion of births financed by Medicaid among Dominicans and other Latinos that differed by mother’s place of birth. The uptake in Medicaid coverage, however, had no differential effect on early prenatal care initiation or the incidence of low birth weight. One explanation is that the fall in uninsured women was relatively modest suggesting foreign-born Latinos were substituting Medicaid for private coverage. I turn therefore, to the second major policy change of the decade, welfare reform.

4. Welfare Reform 1995-1998

The Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) of 1996, more commonly referred to as welfare reform, fundamentally altered federal law regarding immigrant eligibility for Medicaid and other public benefits to immigrants. All legal immigrants entering the country after August 22, 1996 were barred from receiving federal public benefits for at least five years, except for protected categories including refugees and asylees. New York opted not to provide Medicaid to immigrants that entered the U.S. after August of 1996. However, due to a long-standing federal court decision, Medicaid eligibility for all pregnant women regardless of immigration status remained intact. Nevertheless, a recent study by researchers at the Metropolitan Health and Research Association found that use of food coupons from Special Supplemental Program for Women, Infants and Children (WIC) fell substantially in immigrant neighborhoods after welfare reform (Jessop et al. 1998). The authors speculate that immigrants’ uncertainty regarding the law and its impact on their immigration status may explain the decline in WIC use. There were similar reports from California. Researchers from the Urban Institute found that between 1996 and 1998 in Los Angeles County newly approved applications for cash assistance and Medicaid (TANF/Medi-Cal) dropped over 60 percent more in households in which the parents were foreign-born as compared to households with US-born parents (Zimmerman and Fix 1999). Because PRWORA affected US- and foreign-born women differently in the years right after its passage, foreign-born women that were not naturalized citizens may have perceived themselves vulnerable to additional scrutiny regarding their immigration status. One consequence might have been a fall off in Medicaid-finance births and decreased use of prenatal care.

To test this hypothesis I perform a before and after analysis as was presented above for the Medicaid eligibility expansions. I use 1995 as the pre-PRWORA period and 1998 as the post-period. As before, I compare outcomes of US- to foreign-born Latinos using regressions based on equation (1). Results for first-differences and DD estimates are presented in Table 5. Overall, there is little evidence to suggest that welfare reform had any detrimental effect on the financing or utilization of prenatal care. To the contrary, there was an increase of 5 percentage points in the probability that a birth was financed by Medicaid between 1995 and 1998 among foreign-born other Latinos and a decrease in the probability that these same women were uninsured [top panel, columns (1) and (2)]. Indeed, four of the five DD estimates for births to women uninsured are not statistically different from zero. A similar pattern exists for prenatal care. First differences show an increase in both early initiation of prenatal care and prenatal care visits among all six groups of Latinos [top panel, columns (3) and (4)]. Again, this result is contrary to expectations. Given the lack of any change in the financing or utilization of prenatal care, it is unsurprising that we find no notable changes in the incidence of low birth weight, except one. The DD estimate for other Latinos indicates that the rate of low birth weight increased 1.7 percentage points more among the foreign- as compared to the US-born. However, this is due to the fall of 1.4 percentage points among US-born other Latinos and not a rise among the foreign-born. Moreover, this DD estimate goes to zero when we include measures of smoking and illicit drug use. This pattern of results suggests that that the change in low birth weight was unrelated to welfare reform.

5. Declining use of crack/cocaine

The third event that may have had an important impact on newborn health in some communities is the introduction of crack/cocaine in the early to mid 1980’s in New York City. As shown in Figure 6, cocaine use during pregnancy was greatest among US-born blacks, followed by Puerto Ricans and then US-born other Latinos. What is also of note is that prenatal cocaine use peaked around 1988 and 1989. Nineteen eighty-eight was the first year that New York City birth certificates included indications for cocaine as a distinct item. In this section I explore whether the declining use of cocaine among primarily US-born pregnant women can explain differences in low birth weight births between US- and foreign-born Latinos. If, as most studies show, prenatal cocaine was concentrated among poor US-born women, then differences in birth outcomes between US- and foreign-born women should narrow over time.

As noted previously, birth certificates are a flawed measure of prenatal illicit drug use because it is unclear how indications of exposure were collected. The sources of bias are potentially many. Despite these obvious limitations, rankings of prenatal exposure by race and ethnicity based on birth certificates are consistent with rankings from more rigorous studies of prenatal drug use (Vega et al. 1993; McCalla et al. 1992). Nevertheless, underreporting of illicit drug use in the general birthing population makes it difficult to attribute decline in low birth weight to changes in reported drug use. However, a consistent finding in the illicit drug use literature is that drug use is much higher among pregnant women who initiate prenatal care late (Vega et al. 1997). One reason is that late initiation of prenatal care is a marker for high-risk pregnancy, which can often increase the likelihood that the woman is screened for illicit drugs. In this section, therefore, I limit the analysis to women that were uninsured or on Medicaid, a proxy for women in poverty. I further stratify the data by whether women begin prenatal care after the sixth month of pregnancy or before, which I heretofore refer to as late (L) and non-late (NL) prenatal care. As I show below, the prevalence of exposure is much higher among women that initiate care late. I limit the analysis to poor women—the vast majority of whom are on Medicaid—in an effort to minimize unobserved heterogeneity related to income.

Figures 8-16 show 11 years of data for each of three outcomes: smoking during pregnancy, cocaine use during pregnancy and rates of low birth weight. I present time-series for three groups: non-Latino blacks, Puerto Ricans and all other Latinos. The latter combines all non-Puerto Ricans because sample sizes become too small to separate out Dominicans. Moreover, Dominicans and other Latinos have similar rates of low birth weight. For each racial and ethnic group there are four series in each figure: US-born with late prenatal care (LUS), US-born with non-late care (NLUS), foreign-born, late prenatal care (LFO) and foreign-born, non-late care (NLFO). For Puerto Ricans US stands for women born on the mainland and IS for women born in Puerto Rico. The thick lines refer to women that obtain late prenatal care.

Among non-Latino blacks differences in smoking (Figure 8) and cocaine use (Figure 9) between US- and foreign-born women are enormous. Cocaine use peaks in 1989 for US-born blacks that initiate prenatal care late at almost 17 percent and then declined steadily to 7 percent by 1998 (Figure 9). Exposure among foreign-born blacks that begin care late is never greater than 3 percent. The percent of low birth weight births for US-born blacks that begin care late falls from just under 25 percent to 17 percent by 1998. There is little change over the decade in the rate of low birth weight among foreign-born blacks that begin care late.

The same set of figures for Puerto Ricans are displayed in Figures (11)-(13). Unlike with blacks, levels of smoking and cocaine use vary more by when prenatal care began than by mother’s birth place. In other words, cocaine use is approximately three times greater for Puerto Ricans that initiate care late as compared to non-late. Levels of low birth weight follows the same pattern (Figure 13). Cocaine use fell by half for both mainland- and island-born Puerto Ricans that began care late (Figure 12) but the largest declines in low birth weight, over 5 percentage points, were experienced by mainland Puerto Ricans.

The final three figures pertain to all non-Puerto Rican Latinos. Here it is easy to see the importance of mother’s birthplace. Levels of cocaine use are approximately seven times greater among US-born as compared to foreign-born Latinos that begin care late (Figure 14). Both the level of low birth weight and its decline over time among US-born Latinos that begin care late are consistent with both the level and the change in exposure to cocaine over time. Indeed, by 1998 the rate of low birth weight among US- and foreign-born Latinos that initiated care late is essentially equal, eliminating a gap of 10 percentage over the 10 years!

For insight as to how much of the change in low birth weight among US- and foreign-born Latinos is attributable to smoking and illicit drugs, I estimated a series of models based on equation (1). I limited the sample to poor Latinos, those on Medicaid or uninsured, who initiated prenatal care late between 1988 and 1998. Results are shown in Table 6. As before, I show only estimates of the first difference and then estimates of the DD. I limit ethnicity to Puerto Ricans and non-Puerto Ricans Latinos. The first difference in this exercise subtracts levels in 1988 from those in 1998, the two end points. Model A can be viewed as the naïve estimate. The specification includes dummy variables for year, ethnicity, nativity and the relevant interactions. Model B adds demographic characteristics of the mother and Model C adds indicators of substance use: smoking, cocaine, marijuana, heroin, and methadone.

The largest declines in low birth weight occur among US-born Dominicans and other Latinos [Table 6, top panel]. The decline is 5.2 percentage points between 1998 and 1988, just less than a third of its 1988 level. Inclusion of smoking and illicit drugs reduces the decline from 5.3 to 3.9 percentage points, a reduction of 25 percent [top panel, columns (2) and (3)]. The second largest decline occurs among mainland-born Puerto Ricans, about 3.5 percentage points. Again the inclusion of substance use reduces the decline by about 40 percent, from 3.5 to 2.0

The DD estimates in the bottom panel contrast the change in low birth weight among foreign- relative to US-born women. The naïve estimate for Dominicans and other Latinos indicates that the percent of low birth weight births fell 3.7 percentage points more among US as compared to foreign-born Latinos. The DD adjusted for smoking and illicit substances is 2.8 percentage points, about 25 percent less than the DD unadjusted for substances. The DD estimates for foreign-born Latinos and mainland-born Puerto Ricans is 1.7 percentage points, again suggesting a narrowing of birth outcomes. However, this difference falls to less than one percentage point [bottom panel, column (3)] when adjusted for substance use. In summary, both the time-series plots and regression analyses suggest that the decline in both smoking and illicit drugs played a significant role in the relative improvement of birth outcomes between US- and foreign-born Latinos among the subgroup of Latinos that initiate prenatal care late.

This paper had two primary objectives. The first was to characterize the financing of prenatal care, the utilization of prenatal care and the birth outcomes of foreign-born women in New York City and to describe changes in these outcomes over the past decade. The second objective was to analyze whether policies and events that affect perinatal outcomes can help us understand the longstanding observation that the health of immigrants is often better than the health of their offspring.

As to the first objective several observations were striking. Medicaid is now the most important payor of birthing in New York City and the change in ten years has been remarkable. Almost three-fourths of all births to Latinos in the City are financed by Medicaid, up from 55 percent in 1988. Among Asians, 55 percent were financed by Medicaid in 1998, up from only 6 percent in 1988! The growth in Medicaid-financed births has been concentrated among foreign-born women for as I show, the growth in Medicaid-financed births among US-born non-Latino whites and blacks over the same decade has been trivial. The importance of Medicaid-financed births among the foreign-born explains why the withdrawal of federal funds for Medicaid-eligible immigrants under welfare reform was potentially so important. Had New York chosen to discontinue Medicaid-financed prenatal care for immigrants, both legal and undocumented, the significant gains achieved by the Medicaid expansions only six years earlier might have been undone. The fact that the Mayor of the City as well as the Governor of the State, both Republicans, were vocal in their support for immigrants attests to their importance both politically and economically.

The other major change was the decreased use of cocaine among US-born women over the past decade. As the data make clear, illicit drug use is primarily a behavior of citizens. The decline in cocaine use and the improvement in birth outcomes was also coincident with the dramatic fall in crime and served as another indicator that the City has made substantial gains in the quality of life over the past decade.

As to the second objective the data suggest that access to public insurance and increased utilization of prenatal care have not contributed significantly to our understanding of why foreign-born women have such superior birth outcomes. The data presented here point more directly to behavioral differences as the primary explanation for why US-born women of the same SES do so poorly relative to their foreign-born counterparts. The gap in low birth weight between mainland Puerto Ricans and foreign-born Latinos narrowed significantly with the decline in use of cocaine and other illicit drugs. Differences in smoking also appear to have contributed. The graphic that best illustrates this is Figure 16 in which a difference in rates of low birth weight between US- and foreign-born Latinos of almost 10 percentage points in 1988 points was completely eliminated by 1998. The comparison focuses on a relatively narrow group of women, those who are poor and who initiate prenatal care late. Nevertheless, the fact that such significant gains were achieved among such a high-risk group of women offers hope that by discouraging adverse behaviors with taxes and education, we may be eliminate the association between acculturation and adverse health.

Abraido-Lanza, A. F., et. al., 1999. The Latino Mortality Paradox: A test of the "Salmon Bias" and Healthy Migrant Hypotheses. American Journal of Public Health 89, 1543-1548.

Albrecht, S. L., Miller, M. K., 1996. Hispanic Subgroup Differences in Prenatal Care. Social Biology 43, 38-58.

Alexander, G. R., 1992. Maternal Nativity Status and Pregnancy Outcome Among U.S.-born Filipinos. 39, 278-84.

Alexander, G. R., et. al., 1996. Pregnancy Outcomes of US-born and Foreign-born Japanese Americans. American Journal of Public Health 86, 820-824.

Alexander, G. R., et. al., 1999. 1994-1996 U.S. Singleton Birth Weight Percentiles for Gestational Age by Race, Hispanic Origin, and Gender. Maternal and Child Health Journal 3, 225-231.

Amaro, H., et. al., 1990. Acculturation and Marijuana and Cocaine Use: Findings from HHANES 1982-84. American Journal of Public Health 80(Suppl), 54-60.

Balcazar, H., Aoyama C., Cai, X., 1991. Interpretataive Views on Hispanics’ Perinatal Problems of Low Birth Weight and Prenatal Care. Public Health Reports, 106, 420-426.

Becerra, J. E., Hogue, C.J., Atrash, H. K., Perez, N., 1991. Infant Mortality Among Hispanics: A Portrait of Heterogeneity. Journal of the American Medical Association 265, 217-221.

Becerra, J. E., et. al., 1993. Low Birthweight and Infant Mortality in Puerto Rico. American Journal of Public Health 83, 1572-1576.

Becerra, J. E., Smith, J. C., 1988. Maternal Smoking and Low Birthweight in the Reproductive History of Women in Puerto Rico, 1982. American Journal of Public Health 78, 268-272.

Cabral, H., et. al., 1990. Foreign-Born and US-born Black Women: Differences in Health Behaviors and Birth Outcomes. American Journal of Public Health 80, 70-72.

Collaborative Group on Preterm Birth Prevention, 1993. Multicenter Randomized, Controlled Trial of Preterm Birth Prevention Program. American Journal of Obstetrics & Gynecology 169, 352-66.

Collins, J. W., Martin, C. R., 1998. Relation of Traditional Risk Factors to Intrauterine Growth Retardation Among United States-born and Foreign-born Mexican Americans in Chicago. Ethnicity Dis. 8, 21-5.

Collins, J. W., Shay, D. K., 1994. Prevalence of Low Birth Weight among Hispanic Infants with United States-born and Foreign-born Mothers: The Effect of Urban Poverty. American Journal of Epidemiology 139, 184-92.

Cramer, J.C., 1995. Racial and Ethnic Differences in Birthweight: The Role of Income and Financial Assistance. Demography 32, 231-247.

Currie, J., Gruber, J., 1996. Saving babies: the efficacy and cost of recent expansions of Medicaid eligibility for pregnant women. Journal of Political Economy 104, 1263-1296.

David, R. J., Collins, J. W., 1997. Differing Birth Weight Among Infants of U.S.-born Blacks, African-born Blacks, and U.S.-born Whites. New England Journal of Medicine 337, 1209-1214.

Doucet, H., et. al., 1992. Risk of Low Birthweight and Prematurity Among Foreign-born Mothers. Canadian Journal of Public Health 83, 192-195.

Dubay, L., et al. Forthcoming. Changes in Prenatal Care Timing and Low Birth Weight by Race and Socioeconomic Status: Implications for the Medicaid Expansions for Pregnant Women." Health Services Research.

Eberstein, I. W., Pol, L. G., 1982. Mexican-American Ethnicity, Socioeconomic Status, and Infant Mortality: A County-level Analysis. Social Science Journal 19, 61-71

Ellis, J. M. Spanish Surname Mortality Differences in San Antonio, Texas. Journal of Health and Human Behavior, 125-127.

Engel, T., et. al., 1995. Pregnancy Outcomes of U.S.-born Puerto Ricans: The Role of Maternal Nativity Status. American Journal of Preventative Medicine 11, 34-39.

Evans, W. N., Ringel, J. S., Stech, D., 1999. Tobacco Taxes and Public Policy to Discourage Smoking. In: Poterba, J.M. (Ed.), Tax Policy and the Economy. Vol. 13. MIT Press, Cambridge, MA.

Fang, J., Madhavan, S., Alderman, M. H., 1999. Low Birth Weight: Race and Maternal Nativity¾ Impact of Community Income Pediatrics 103, e5.

Fuentes-Afflick, E., et. al., 1998. Maternal Birthplace, Ethnicity, and Low Birth Weight in California. Arch Pediatr Adolesc Med. 152, 1105-1112.

Fuentes-Afflick, E., et. al., 1999. Testing the Epidemiologic Paradox of Low Birth Weight in Latinos. Arch Pediatr Adolesc Med. 153, 147-153.

Fuentes-Afflick, E., Hessol, N. A., 1997. Impact of Asian Ethnicity and National Origin on Infant Birth Weight. American Journal of Epidemiology 145, 148-155.

Gee, S. C., Lee, E. S., Forthofer, R. N., 1976. Ethnic Differentials in Neonatal and Postneonatal Mortality: A Birth Cohort Analysis by a Binary Variable Multiple Regression Method. Social Biology 23, 317-325.

Goldenberg, R. L., Rouse , D. J., 1998. Prevention of Premature Birth. The New England Journal of Medicine 339, 313-320.

Guendelman, S., English, P. B., 1995. Effect of United States residence on Birth Outcomes among Mexican Immigrants: An Exploratory Study. American Journal of Epidemiology 142(Supplement), S30-S38.

Guendelman, S., et. al., 1990. Generational Differences in Perinatal Health among the Mexican American Population: Findings from HHANES 1982-84. American Journal of Public Health 80(Supplement), 61-65.

Hedderson, J., Daudistel, H. C., 1982. Infant Mortality of the Spanish Surname Population. Social Science Journal 19, 67-78.

Hauth, J.C., et. al., 1995. Reduced Incidence of Preterm Delivery with Metronidazole and Erythromycin in Women with Bacterial Vaginosis. New England Journal of Medicine 333, 1732-6.

Jessop D, Finkelstein R, Rosenberg T, et al. The Impact of Recent Immigration Policy Changes on the Receipt of WIC Services: An Aggregate Analysis. New York:, NY: Medical and Health Research Association, 1999.

Joseph, K.S., et. al., 1998. Determinants of Preterm Birth Rates in Canada From 1981 Through 1983 and From 1992 Through 1994. New England Journal of Medicine 339, 1434-9.

Joyce, T., 1992. The Consequences and Costs of Maternal Substance Abuse in New York City: A pooled Time-Series, Cross-Section Analysis. Journal of Health Economics 11, 297-314.

Joyce, et. al., 1995. The Impact of Prenatal Exposure to Cocaine on Newborn Costs and Length of Stay. Health Services Research 30, 341-358.

Joyce, et. al., 1990. The Dramatic Increase in the Rate of Low Birthweight in New York City: An Aggregate Time-Series Analysis. American Journal of Public Health 80, 682-684.

Kaestner, R. , Joyce, T. and H. Wehbeh, "The Effect of Maternal Drug Use on Birth Weight: Measurement Error in Binary Variables." Economic Inquiry. 34:617-629. 1996.

Kessner, D. M., et. al., 1973. Infant Death: An Analysis by Maternal Risk and Health Care. Institute of Medicine, National Academy of Sciences, Washington, D.C.

Kleinman, J.C., Fingerhut, L. A., Prager, K., 1991. Differences in Infant Mortality by Race, Nativity Status, and Other Maternal Characteristics. AJDC 145, 194-199.

Ko, M., Schulken, E. D., 1998. Factors Related to Smoking Cessation and Relapse Among Pregnant Smokers. American Journal of Health Behavior 22, 83-89.

Kotelchuck, M, 1994. An evaluation of the Kessner Adequacy of Prenatal Care Index and Proposed Adequacy of Prenatal Care Utilization Index. American Journal of Public Health 84, 1414-1420.

Langer, A., et. al., 1996. The Latin American Trial of Phychosocial Support during Pregnancy: Effects on Mother’s Wellbeing and Satisfaction. Social Science & Medicine 42, 1589-1597.

Markides, K. S., Coreil, J., 1986. The Health of Hispanics in the Southwestern United States: an Epidemiologic Paradox. Public Health Reports 101, 253-265.

Markides, K. S., Hazuda, H. P., 1980. Ethnicity and Infant Mortality in Texas Counties. Social Biology 27, 261-271.

McCalla, S., et. al., 1991. The Biologic and Social Consequences of Perinatal Cocaine Use in an Inner-City Population: Results of An Anonymous Cross-Sectional Study. American Journal of Obstetrics & Gynecology 164, 625-30.

Moore, P., Hepworth, J. T., 1994. Use of Perinatal and Infant Health Services by Mexican-American Medicaid Enrollees. Journal of the American Medical Association 272, 297-304.

Morrow, H.W., et. al., 1994. Infant Mortality and Related Risk Factors among Asian Americans. American Journal of Public Health. 84(9): 1497-1500.

Noble, A., et. al., 1997. Prenatal Substance Abuse in California: Findings from the Perinatal Substance Exposure Study. Journal of Psychoactive Drugs 29, 43-53.

Piper, J.M., Ray, W.A., Griffen, M.R., 1990. Effects of Medicaid eligibility expansion on prenatal care and pregnancy outcome in Tennessee. Journal Of The American Medical Association 264, 2219-2223.

Powell-Griner, E., Streck, D., 1982. A Closer Examination of the Neonatal Mortality Rates Among the Texas Spanish Surname Population. American Journal of Public Health 72, 993-999.

Rogowski J. Cost-Effectiveness of Care for Very Low Birth Weight Infant. Pediatrics 102(1, Pt1): 35-43. 1998.

Rosenwaike, I., Hempstead, K., 1989. Differential Mortality by Ethnicity: Foreign-born Irish, Italians, and Jews in New York City, 1979-81. Social Science & Medicine 29, 885-889.

Rumbaut, R.G., Weeks, J.R. Unraveling A Public Health Enigma: Why do Immigrants Experience Superior Perinatal Health Outcomes? Research in the Sociology of Health Care 13B, 337-391.

Scribner, R., Dwyer, J. H., 1989. Acculturation and Low Birthweight among Latinos in the Hispanic HANES. American Journal of Public Health 79, 1263-1267.

Selby, M. L., et. al., 1984. Validity of the Spanish Surname Infant Mortality Rate as a Health Status Indicator for the Mexican American Population. American Journal of Public Health 74, 998-1002.

Sexton, M., Hebel, R., 1984. A Clinical Trial of Change in Maternal Smoking and its Effect on Birth Weight. Journal Of The American Medical Association 251, 911-915.

Singh, G. K., Yu, S. M., 1996. Adverse Pregnancy Outcomes: Differences Between US- and Foreign-Born Women in Major US Racial and Ethnic Groups. American Journal of Public Health 86, 837-843.

Stroup-Benham, C. A., Trevino, F. M., 1991. Reproductive Characteristics of Mexican-American, Mainland Puerto Rican, and Cuban-American Women. Journal of the American Medical Association 265, 222-226.

Trevino, F. M., 1982. Vital and Health Statistics for the US Hispanic Population. American Journal of Public Health 72, 979-981.

Valanis, B. M., Rush, D. A Partial Explanation of Superior Birth Weights Among Foreign-born Women. Social Biology 26, 198-210.

Vega, W, Kolody, B, Hwang, J, and Noble, A. 1993. Prevalence and magnitude of perinatal substance exposures in California. New England Journal of Medicine. 329, 850-854.

Vega, W. A., et. al., 1998. Illicit drug use among Mexicans and Mexican Americans in California: the effects of gender and acculturation. Addiction 93, 1839-1850.

Vega, W. A., et. al., 1997. Perinatal Drug Use among Immigrant and Native-Born Latinas. Substance Use & Misuse 32, 43-62.

Vega, W. A., et. al., 1993. Prevalence and Magnitude of Perinatal Substance Exposures in California. New England Journal of Medicine 329, 850-4.

Ventura, S. J., Taffel, S. M., 1985. Childbearing Characteristics of U.S.- and Foreign-born Hispanic Mothers. Public Health Reports 100, 647-652.

Weeks, J. R., Rumbaut, R. G., 1991 Infant Mortality Among Ethnic Immigrant Groups. 327-334. Social Science & Medicine 33, 327-34.

Williams, R. L, et. al., 1986. Pregnancy Outcomes among Spanish-surname Women in California. American Journal of Public Health 76, 387-391.

Winkleby, M. A., et. al., 1995. Hispanic Versus White Smoking Patterns by Sex and Level of Education. American Journal of Epidemiology 142, 410-18.

Zambrana, R.E., et. al., 1996. Prenatal Care and medical in Low-Income Primiparous, Mexican-Origin and African American Women. Families, Systems & Health 14, 349-359.

Zimmerman W and M Fix. Declining Immigrant Applications for Medi-Cal and Welfare Benefits in Los Angeles County. Washington, D.C.:Urban Institute, 1998.

Selected Characteristics and Outcomes of New York City Residents By Race/Ethnic, 1988 and 1998
	Year	N	US-Born %	Medicaid %	Teen %	< HS Ed %	Unmarried %	Late care %	LBW %
Latin America
Puerto Rican	1998	12,186	75.5	64.1	21.1	42.3	71.7	6.5	9.1
	1988	18,137	61.1	60.8	21.4	46.0	61.8	14.9	10.1
Dominican	1998	9,112	14.1	79.4	12.3	35.9	61.6	6.6	6.0
	1988	9,541	4.5	62.5	8.0	41.0	44.2	15.2	5.6
Mexican	1998	5,390	3.1	89.4	15.8	65.8	69.6	10.1	5.3
	1988	1,491	3.5	36.9	19.9	68.1	44.7	22.9	6.4
Other Latinos	1998	11,304	20.7	70.0	12.1	31.8	54.2	7.0	5.8
	1988	8,386	12.2	36.8	8.3	25.7	40.5	13.3	5.4
Latin America, Latin	1998	37,992	34.2	73.1	15.5	41.0	63.8	7.2	6.8
	1988	37,555	33.5	54.9	15.0	41.1	51.9	14.9	7.8
Other Caribbean/South America
Jamaicans	1998	3,061	2.9	56.8	8.9	24.0	66.6	6.4	8.0
	1988	2,511	1.4	38.2	9.2	28.4	59.1	11.5	8.5
Guyanese	1998	2,414	1.6	51.4	7.0	23.5	42.8	6.0	11.9
	1988	1,658	1.0	32.1	8.3	19.6	35.2	11.0	11.6
Haitian	1998	1,990	5.7	56.4	4.9	17.6	42.1	9.0	8.4
	1988	3,313	1.4	35.7	3.1	24.6	40.6	13.4	6.5
Trinidadian	1998	1,398	3.0	56.8	9.4	13.4	53.1	8.1	9.8
	1988	940	0.6	37.7	6.6	20.5	51.6	15.6	10.3
Latin America, AFR	1998	11,138	3.2	55.3	7.5	20.8	52.5	7.1	9.4
	1988	12,921	1.2	36.0	6.4	24.3	46.3	12.6	8.5
Asia
Chinese	1998	4,345	5.0	58.9	1.3	28.5	13.7	3.8	3.5
	1988	3,494	4.8	3.3	1.5	19.1	5.3	6.5	3.8
Indian	1998	1,644	2.6	43.3	1.9	10.8	12.0	5.7	7.6
	1988	1,355	2.4	12.3	3.2	8.5	12.0	14.2	7.7
Total Asian	1998	13,239	4.3	54.7	1.5	23.6	13.2	4.3	4.6
	1988	10,395	4.2	5.8	2.0	16.1	7.2	8.7	4.9
US-Born Non-Latino Whites	1998	19,907	100	15.9	3.4	7.2	14.5	2.6	4.4
	1988	25,931	100	11.5	4.6	7.5	14.2	5.1	5.1
US-Born Non-Latino Blacks	1998	17,855	100	59.1	18.9	29.7	79.5	9.1	12.0
	1988	25,678	100	57.0	20.7	34.3	75.4	17.7	16.0

Notes: US-born Puerto Ricans refers to those born in the continental United States.
 

Table 2
Number and Percent Distribution of Births to Latinos by Year of Birth, Mother’s Nativity and Whether She Received Medical Assistance or Cash Assistance, New York City 1989 and 1991
Ethnicity	1989			1991
	N	% Medical Assistance	% Cash Assistance	N	% Medical Assistance	% Cash Assistance
Puerto Ricans
Mainland US	1491	20.6	79.4	1728	24.8	75.2
Puerto Rico	881	16.4	83.7	857	18.3	81.7
Dominicans
US born	94	36.2	63.8	137	34.1	65.7
Foreign born	1220	51.3	48.7	1555	56.7	43.3
Other Latinos
US born	131	29.0	71.0	157	38.9	61.2
Foreign born	940	82.8	17.2	1610	89.0	11.0

^{Notes: Cash Assistance refers to women who were eligible for Medicaid because they receive cash assistance through either AFDC, SSI, or Home Relief; Medical Assistance refers to women who were not receiving cash assistance but who were enrolled in Medicaid through the Medical Assistance Program.}
 
 

	Method of Payment		Prenatal Care		Low Birth Weight
	Medicaid	Uninsured	Early Care	Visits	Excludes substances	Includes substances
	(1)	(2)	(3)	(4)	(5)	(6)
Mother’s age
< 20	.072*	.016*	-.073*	-.323*	-.007*	-.001
20-34	--	--	--	--	--	--
35 +	-.085*	-.011*	.070*	.509*	.024*	.024*
Unmarried	.215*	.013*	-.067*	-.653*	.027*	.021*
Parity
0	-.049*	.002	.030*	.305*	.009*	.011*
1-4	---	---	---	---	---	---
5+	.078*	.016*	-.096*	-.942*	.021*	.014*
Education
< high school	.169*	-.002	-.001	-.192*	.006*	.004*
High school	---	---	---	---	---	---
Some college	-.082*	---	.060*	.634*	-.007*	-.004+
College +	-.234*	---	.074*	.754*	-.014*	-.010*
Ethnicity
Dominican	.099*	.037*	-.029*	.448*	-.031*	-.025*
Other Latino	-.100*	.077*	-.034*	.268*	-.035*	-.029*
Puerto Rican	---	---	---	---	---	---
Born in US	-.078*	.0003	.001	-.006	.008*	.005*
US x Ethnicity
Dom x US	-.001	-.041*	.031*	.127