Introduction

The majority of New York State’s Medicaid enrollees have historically come from New York City, despite the city accounting for less than half of the state’s overall population. In 2012, for example, 62 percent of the state’s Medicaid enrollees and 43 percent of its total population lived in the city. City residents are also less likely to be covered by employer-sponsored health insurance (ESI) and more likely to lack any health coverage.

Demographic and labor market differences between New York City and the rest of the state could explain this phenomenon. Compared with individuals in the rest of the state, New York City residents are on average younger, have a lower family income, are more likely to be female, and less likely to be employed (if an adult), or have an employed parent (if a child). All of these factors increase the chances that an individual will be eligible for Medicaid. New York City also has a considerably larger foreign-born and minority population than the rest of the state, groups that tend to be over represented on the Medicaid rolls.

In 2013, IBO released a fiscal brief called Growth in New York’s Medicaid Enrollment and Costs examining enrollment and expenditure trends in New York State’s Medicaid program over the 13-year period from 2000 through 2012. In that report, we provided background information on how Medicaid works in New York State—both before and after the Affordable Care Act (ACA)—and laid out the myriad policy and economic changes behind Medicaid’s rapid enrollment and expenditure growth since 2000. We also parsed the program’s statewide enrollment and expenditure data by region and eligibility category to determine whether certain regions and populations have been driving more of this growth than others.

As Medicaid is a major use of city tax dollars—$6.4 billion, or nearly 12 percent of the city-funded portion of the budget in fiscal year 2015—it is critically important to understand who is enrolling in this program and why. To that end, this report builds upon IBO’s earlier analysis by using census data to delve deeper into the differences between the Medicaid caseload and access to employer provided health insurance in New York City and elsewhere in the state as of 2012.

It is worth noting that there have been significant changes to the health insurance landscape in New York State since these data were collected, most notably the roll out of a new health insurance marketplace for Medicaid, private individual, and small business plans, along with the introduction of federally funded tax subsidies for private insurance and a fairly minor increase in Medicaid eligibility levels. Nonetheless, we believe that this report provides an important picture of health insurance patterns in New York prior to the implementation of the ACA that can act as a baseline for assessing future changes in the Medicaid caseload.

One of the key questions we explore is whether demographic differences alone are responsible for New York City’s higher Medicaid enrollment numbers compared with the rest of the state. Specifically, are regional differences in average income levels, age distribution, and the size of the immigrant population enough to explain New York City’s higher Medicaid enrollment rates? We then consider the relationship between Medicaid enrollment and the availability or accessibility of other types of health insurance, notably employer-sponsored insurance in New York. Historically, most Americans have obtained health insurance at work, but the share of employers offering this type of coverage is on a long-term downward trajectory. In New York State for example, the percentage of working age adults (ages 19–64) with ESI has declined from just under 66 percent in 2002 to 61 percent in 2012^.1 Given very high costs in New York State’s individual health insurance market—at least prior to the implementation of the ACA—most individuals unable to obtain insurance through an employer end up either uninsured or with public coverage such as Medicaid. Therefore, this analysis also probes the relationship between ESI and Medicaid enrollment rates and examines regional differences in various factors that may affect access to ESI, such as employment rates, industry types, job types, and educational attainment levels.

Data and Methodology

Like IBO’s previous report on Medicaid, this fiscal brief will compare New York City with two other regions of the state: the downstate suburbs (Nassau, Putnam, Rockland, Suffolk, and Westchester counties) and the upstate counties (all other, non-New York City counties). The analysis will again focus on the same three broad eligibility categories examined in our prior report: children, adults, and seniors and the disabled.² There are, however, a number of key differences between our two analyses of the Medicaid caseload due to differences in the types of data used.

The analysis in our earlier report was based on administrative data from the New York State Department of Health (DOH), while most of the analysis here is based on data from the Census Bureau’s American Community Survey (ACS). Specifically, this report uses the 2012 Public Use Microdata Sample (PUMS), which consists of records of individuals’ responses to the ACS questionnaire. Before these data are publicly released, all personally identifying information is removed. Place of residence (and place of work) are identified, but only within broad geographic areas that in the case of New York City approximate community planning board districts.

When asking questions about the demographic and employment characteristics of the Medicaid caseload, the PUMS data have several advantages over the administrative data used in our initial report. While the administrative data provided to the public are available only on the county level (or the city level in the case of New York City), PUMS data are available on the individual level. Moreover, PUMS data contain information on a great number of variables not captured in the state’s administrative data for Medicaid, including employment status, family income, race, ethnicity, and other demographic variables. Lastly, PUMS data also include information on enrollment in other types of health insurance besides Medicaid, allowing researchers to compare the Medicaid-enrolled population with those who are uninsured or have other types of insurance.

There are, however, also a number of limitations associated with using the ACS. The first is that health insurance enrollment questions were only added to the ACS in 2008, so it is not possible to do a long-term analysis of changes over time. Second, the health insurance questions that now appear in the ACS do not differentiate between children’s enrollment in Medicaid and the Children’s Health Insurance Program (New York State’s program is Child Health Plus). This differs from current practice in New York State, which is to administer the two as separate programs and to exclude children with Child Health Plus from its Medicaid totals. A related problem is that most data in the ACS are self-reported—though there is some post-survey editing by Census Bureau staff—which may lead to under- or over-reporting on certain variables. The health insurance questions in the ACS also exclude the institutionalized population, notably individuals in nursing homes who tend to be heavily Medicaid dependent. Lastly and most significantly, there is a well-documented Medicaid undercount problem associated with the Census Bureau’s survey data.

For this reason, we have opted to focus on percentages of the population enrolled in Medicaid (or enrolled in employer-sponsored insurance, or uninsured), rather than on actual population estimates for much of our analysis. This does not mitigate the problem completely, especially when comparing Medicaid and employer-sponsored insurance rates within the same region. Given that the undercount is greater in New York City than elsewhere in the state, comparisons between New York City and other New York State regions bias the results towards understating rather than overstating the differences in Medicaid enrollment rates between the city and elsewhere. For more information on the underestimate of Medicaid enrollment and various other data limitations of the PUMS data, and how we have addressed them, please see the appendix at the end of this report.

Regional Health Insurance Enrollment Patterns

As of 2012, New York City had a higher share of its population enrolled in Medicaid and a lower share enrolled in employer-sponsored insurance than either the downstate suburbs or the upstate counties. This was true across all eligibility categories and all income levels. New York City also had higher uninsured rates than the rest of the state, both in terms of its overall population and among its higher-income population. However, the city has the lowest uninsured rates in the state for very low-income children, seniors, and the disabled, primarily due to these groups’ strong participation in Medicaid.

Enrollment by Eligibility Category. When 2012 insurance enrollment rates in all regions and eligibility categories are examined, a clear pattern emerges. Namely, that New York City has the highest rate of Medicaid coverage and the lowest rate of employer-sponsored insurance of the three regions, across every eligibility category: children, adults, and seniors and the disabled. Moreover, because the high Medicaid rates in New York City are not high enough to compensate for the low rates of ESI, the city also has the highest uninsured rate for every eligibility category.

Adults have the highest uninsured rates, the highest ESI rates, and the lowest Medicaid rates of the three groups, and the gap between these rates in New York City compared with other areas of the state can be considerable. For example, ESI rates for adults run from a low of 54.6 percent in New York City to a high of 72.7 percent in the downstate suburbs, a difference of nearly 20 percentage points. (The ESI rate for adults upstate, at 70.8 percent, is close to the suburban rate.³) Similarly, the adult Medicaid enrollment rate in New York City (20.4 percent) is three times the rate in the downstate suburbs, and almost double the rate upstate. Lastly, the adult uninsured rate in New York City (20.1 percent) is more than 5 percentage points higher than the uninsured rate in the downstate suburbs or upstate counties.

Most of these basic trends hold true for children and seniors and the disabled as well: there are fairly pronounced differences between New York City and the downstate suburbs (and to a lesser extent, the upstate counties) in terms of ESI and Medicaid coverage rates for these two groups. In New York City, 51.5 percent of children are covered by Medicaid and 42.0 percent are covered by their parents’ ESI; the comparable figures are 20.4 percent and 71.6 percent in the downstate suburbs and 32.9 percent and 60.7 percent upstate. For seniors and the disabled, enrollment in ESI ranges from 29.1 percent in the city to 44.7 percent upstate and 47.2 percent in the suburbs. Enrollment in Medicaid varies from 40.6 percent in the city to 26.3 percent upstate and 19.5 percent in the suburbs. In contrast, the share of children who are uninsured is quite similar across the state, a low 4.3 percent to 4.4 percent in each region.

Trends for the two other types of health insurance coverage—Medicare and private insurance directly purchased in the individual market—are not quite so clear cut. Medicare is almost exclusively a program for seniors, and direct purchase insurance covers only a fairly small share of the overall population. For both of these reasons, neither of these types of insurance will be discussed in any depth in the analysis that follows.

The reader should be aware, however, of some basic facts. First, seniors and the disabled in New York City have lower rates of enrollment in Medicare and in direct purchase insurance than their counterparts in the rest of the state. Both of these factors can help explain why New York City has the highest uninsured rate for this group.

Second, New York City looks more like the rest of the state in terms of directly purchased insurance coverage for children and adults. Directly purchased insurance is a more common option for both of these groups, but as of 2012 it was still used by only a small fraction of the population—less than 10 percent in all regions. PUMS data show that children and adults in the city have virtually identical rates of enrollment in directly purchased insurance as do their counterparts upstate, while enrollment is higher for adults in the suburbs.

Enrollment by Family Income Level. When 2012 insurance enrollment rates are further parsed by family income level, many of the same trends continue to hold. Notably, New York City has the highest rates of Medicaid coverage and the lowest rates of ESI coverage of any region across all income levels. For adults with family income less than or equal to 100 percent of the federal poverty level (FPL)—approximately $23,500 for a family of four in 2012—Medicaid enrollment rates range from 50.9 percent in New York City, to 43.2 percent upstate, to 34.3 percent in the downstate suburbs.⁴ For this same group of adults, ESI rates range from 13.6 percent in the city, to 20.4 percent in the suburbs, to 24.5 percent upstate.

For adults at the opposite end of the income scale (family income greater than or equal to 501 percent of FPL, or about $117,700 for a family of four in 2012), Medicaid enrollment rates range from 2.9 percent in the city, to 1.4 percent in the downstate suburbs, to 1.2 percent in the upstate counties. (To qualify for Medicaid at these income levels, an individual must have significant health expenditures that they are able to deduct from their income and “spend down” to Medicaid eligibility.) The comparable figures in terms of employer-sponsored insurance for these high-income adults are 85.6 percent in the city, 89.7 percent in the suburbs, and 90.4 percent upstate. As these numbers indicate, Medicaid enrollment rates go down and ESI enrollment rates go up as family income increases in every region.

Across All Income Levels, Medicaid Enrollment is Highest in New York City

SOURCE: American Community Survey Public Use Microdata Sample 2012

 New York City Independent Budget Office

Across All Income Levels, Employer-Sponsored Insurance Enrollment Rates Are Lowest in New York City

SOURCE: American Community Survey Public Use Microdata Sample 2012

 New York City Independent Budget Office

The same general pattern across income categories holds for children and seniors and the disabled. One key difference, however, is that the Medicaid enrollment rate is much higher for both of these groups than for adults at each income level. For example, Medicaid rates for children range from 85.4 percent at the low end of the income distribution to 5.4 percent at the high end in New York City. The comparable ranges for the downstate suburbs and the upstate counties are 68.7 percent to 2.4 percent and 81.3 percent to 3.2 percent, respectively. Medicaid enrollment rates are not quite as high for seniors and the disabled, but are still markedly higher than for the adult population in every region.

Similarly, and as would be expected, employer-sponsored insurance rates are lower among the senior and disabled population than among the adult population. In contrast, children actually have higher rates of coverage through their parents’ employer-sponsored insurance than do adults in many cases. For those residing in the downstate suburbs, this is the case across all income levels. For those residing in New York City or the upstate counties, however, children have higher rates of ESI coverage than adults only within the higher-income populations. Within the poverty-level population, children’s enrollment in ESI trails that of adults by at least 4 percentage points across both regions.

The reason for this may lie in family size differences between adults who are enrolled in ESI and those who are not. Statewide, adults with health coverage through an employer and the lowest income levels (200 percent of FPL or less) have smaller families on average than those with similar income levels but without comparable health insurance. In contrast, adults with ESI and family income at or above 201 percent of federal poverty level have larger families on average than those with similar income levels who lack employer-sponsored health insurance. Looking at the highest income employed adults for example (family income of 501 percent of FPL or higher), those with employer health coverage live in households containing 0.51 children on average, while those without ESI live in households containing 0.33 children on average.

The relationship between income and lack of health insurance in 2012 is somewhat more nuanced than was the case for Medicaid and ESI. It is only at higher income levels that the share of New York City residents who are uninsured exceeds the share without insurance in the upstate counties or downstate suburbs. Among the poverty-level population, the downstate suburbs consistently have the highest uninsured rates of any region. Conversely, New York City actually has the smallest share of residents lacking health insurance of the three regions for poverty-level children and seniors and the disabled. The share of the city’s poverty-level children without insurance is 4.8 percent versus 7.2 percent upstate and 8.9 percent downstate. Similarly, 5.6 percent of the city’s poverty-level seniors and disabled lack insurance versus 5.9 percent upstate and 6.4 percent downstate. Given that New York City has the lowest rates of ESI coverage for poverty-level individuals of any age, its comparatively low uninsured rates for these groups are the result of the city having higher Medicaid enrollment rates than elsewhere in the state.

The last trend of note is that the share of the population without health insurance does not consistently decline as incomes rise. For example, the shares of adults in New York City and the upstate counties who are uninsured are actually slightly higher for those just above the poverty level than for those below it. In the city, 32.0 percent of adults with income 101 percent to 200 percent of FPL are uninsured, versus 29.6 percent of adults with income at or below 100 percent of FPL. Upstate, the gap between the uninsured rate for those above and below the poverty line is only about half a percentage point in size and is not statistically significant, but runs in the same direction. This higher uninsured rate just above the poverty line is likely due to the threshold for Medicaid eligibility, which in 2012 was open to childless adults only if their family income fell at or below the poverty line.⁵

Insurance Enrollment and Employment

Could regional differences in terms of employment rates and industry types explain the regional variation observed in terms of employer-sponsored insurance and Medicaid enrollment? Our analysis suggests that relatively little of the variation in 2012 is explained by such differences. Even within the same industry, workers who reside in New York City have lower ESI rates than their counterparts elsewhere in the state. However, New York City does have a larger share of the lowest income part-time workers than the rest of the state, and these workers have very low participation in ESI. Educational attainment levels, which are strongly correlated with ESI enrollment rates, also suggest a less educated workforce in New York City than in the suburbs or upstate counties on average.

Enrollment by Employment Status. According to the PUMS data, only 71.0 percent of New York City adults age 19 to 64 were employed in 2012, versus 75.0 percent of these individuals upstate and 76.4 percent in the downstate suburbs. This can at least partially explain the low rates of employer-sponsored insurance in New York City compared with the other regions. Looking only at employed adults (rather than all adults), rates of enrollment in ESI rise and rates of enrollment in Medicaid fall. Similarly, the population of children with at least one employed parent has a higher rate of enrollment in ESI and a lower rate of enrollment in Medicaid than the population of all children. This pattern holds in each region, although the gaps between enrollment rates in New York City and other regions narrow slightly. Nevertheless, New York City still has the lowest rates of ESI and the highest rates of Medicaid coverage for employed adults (and for the children of employed adults) of any region at every income level.

The trends are not quite so clear cut in terms of the share uninsured, however. For middle- and high-income adults—those with a family income of 201 percent of FPL or greater—uninsured rates decrease when the sample is restricted to include only employed adults. This is true across all regions and within each of these income bands. Within the two lowest income bands in New York City and the suburbs, however, uninsured rates are slightly higher for employed adults than for all adults. For example, uninsured rates in New York City are 30.7 percent for employed adults at or below 100 percent of FPL, compared with 29.6 percent for all adults at this income level. These findings—significant at the 5 percent level in this case—suggest that the increase in ESI coverage associated with employment is not large enough to fully offset the corresponding decrease in Medicaid enrollment for low-income adults with jobs.

In terms of uninsured rates, New York City has the highest uninsured rates of any region for the same group of middle- and high-income employed adults described above. However, the difference in uninsured rates between the city and its suburbs is not statistically significant at incomes between 201 percent and 400 percent of FPL. For those employed adults with family income at or below 200 percent of FPL, uninsured rates are actually highest in the downstate suburbs (and in this case the difference is statistically significant).

Our analysis thus far suggests that low enrollment in employer-sponsored insurance is a key factor in New York City’s high Medicaid enrollment rates. None of the data we have examined up to this point, however, have shed any light on whether ESI enrollment rates are comparatively low in New York City because fewer employers offer this type of insurance (an issue of availability) or because fewer employees opt to purchase it. As respondents are not asked about insurance offerings, the PUMS data cannot definitively answer this question. Nor can the PUMS data provide evidence as to whether those employees who opt out of ESI do so because their contribution costs are prohibitive (an issue of access) or because they prefer the more comprehensive coverage offered by Medicaid (an issue of crowd out). It is likely that all three factors—availability, access, and crowd out—play a part, but it is difficult to tease out the independent impact of each without more detailed data. However, in the sections that follow we will attempt to get at some of these questions indirectly, by examining industry types, job types, education levels, and immigration status—all factors that impact an employer’s likelihood of offering health insurance. Later, we will also examine Medicaid take-up rates and private insurance costs, which effect access and crowd out as well as availability.

Enrollment by Industry Type. Since some industries are more likely to offer employer-sponsored insurance than others, differences in the types of industries operating in different parts of the state may explain some of the regional differences found thus far. However, the PUMS data do not clearly bear this out; 78.8 percent of employed adults residing in New York City in 2012 worked in industries with a high or medium ESI rate, roughly 2 percentage points below the rate in the downstate suburbs and 1 percentage point below the rate upstate. Using the broad industry categories available in this data set, the five industries that employ the most New York City residents are: (1) educational services, health care, and social assistance; (2) professional and business services; (3) arts, entertainment, recreation, accommodation, and food services; (4) retail trade; and (5) finance and real estate. Of these, just one can be categorized as a low-ESI industry: arts, entertainment, recreation, accommodation, and food services, in which only 49 percent of employed adults are enrolled in employer-sponsored insurance. These same five industries also account for the majority of employees in the downstate suburbs, although manufacturing replaces finance and real estate within the list of the top five industries in upstate New York.

Delving deeper into employment patterns within the three regions, New York City has a higher percentage of workers than upstate or downstate in two low-ESI industries: arts, entertainment, recreation, accommodation, and food services; and other services. Conversely, the city also has a higher percentage of workers than the rest of the state within two high-ESI industries: finance and real estate; and information.

A larger issue than the distribution of the city’s workforce appears to be that even within the same industry, workers in New York City fare worse in terms of ESI coverage than those elsewhere. Employed adults residing in New York City have lower rates of ESI, higher rates of Medicaid, and higher uninsured rates than their counterparts in the rest of the state in nearly every industry. For example, ESI coverage rates for adults employed within the education, health care, and social services industry—the biggest employer in every region—range from 74.4 percent in New York City, to 85.5 percent upstate, and 85.8 percent in the downstate suburbs. The share of workers who are uninsured in this industry ranges from 9.9 percent of workers in New York City, to 6.3 percent in the downstate suburbs, to 5.7 percent upstate. Lastly, Medicaid enrollment rates for this same group of employed adults are 12.9 percent in the city, 4.3 percent in the suburbs, and 5.9 percent upstate. The only major industry that differs from this pattern is professional and business services, and in this case many of the differences in enrollment rates between New York City and the rest of that state are not statistically significant. (Differences are also not statistically significant in the case of the public administration and information industries.)

Regional Differences in Job Types and Education Levels. The industry categories discussed thus far are fairly broad and may be masking regional differences in job types and required education levels that could better explain insurance enrollment patterns. For example, part-time workers all across the state are less likely than full-time workers to be enrolled in employer-sponsored insurance and more likely to be uninsured or on Medicaid.⁶ The same is true for self-employed workers versus those who work for a business, government, nonprofit organization, or other individual. Notably, New York City has a higher share of self-employed workers than other parts of the state: self-employment accounts for 6.4 percent of the adult labor force residing in the city versus 5.4 percent in the suburbs and 5.1 percent upstate. However, as these numbers indicate, self-employment is a small minority of the labor market in all three regions and as such is unlikely to exert much influence on overall insurance patterns.

Part-time employment comprises a more significant share of New York State’s labor market than does self-employment, and as such could be exerting a larger impact on insurance patterns. PUMS data from 2012 show that New York City actually has a slightly lower percentage of part-time workers than the rest of the state: 12.3 percent of employed adults in the city work less than 30 hours a week, compared with 13.2 percent of employed adults in the suburbs and 13.8 percent upstate. Yet, the family income distribution of part-time workers in New York City is different than elsewhere in the state. Nearly half of part-time workers in the city (48.8 percent) have a family income that is 200 percent of FPL or less and only a quarter have a family income that is more than 400 percent of FPL. The income distribution upstate looks decidedly less skewed: 37.7 percent of part-time workers have a family income that is 200 percent or less of FPL and 33.3 percent have incomes of 400 percent or more of FPL. The corresponding percentages in the downstate suburbs are 24.8 percent and 50.6 percent, respectively. These figures likely understate the extent of the income differences between part-time workers in New York City and those elsewhere in the state, as the federal poverty level is defined uniformly throughout the state (and country) and does not adjust for regional differences in the cost of living.

SOURCE: American Community Survey Public Use Microdata Sample 2012
NOTES: Percentages do not sum to 100. Direct purchase insurance and Medicare are excluded, and individuals may have both employer-sponsored health insurance and Medicaid.

New York City Independent Budget Office

Although the city has a slightly smaller share of part-time workers than elsewhere in the state, more than half of the state’s lowest income part-time workers (51.3 percent) resided in New York City in 2012. This population is fairly Medicaid-dependent in general, with enrollment rates above 30 percent throughout the state. Thus, having a larger proportion of these workers in the city than elsewhere in the state may partially explain high Medicaid enrollment rates here. Once again, however, this is not the full story. Not only does New York City have more low-income, part-time workers than other regions, but the ones who reside here are also more reliant on Medicaid than those who reside elsewhere. Specifically, Medicaid enrollment rates for these workers range from 45.4 percent in the city to 32.4 percent upstate and 30.4 percent in the suburbs. A greater reliance on Medicaid may be linked to a lower ESI enrollment rate for this population in the city—19.7 percent versus 22.5 percent in the downstate suburbs and 35.4 percent upstate (though the difference in enrollment rates between New York City and the suburbs is not statistically significant at any reasonable confidence level in this case). In addition, low-income part-time workers without employer coverage appear more likely to enroll in Medicaid coverage if they live in the city rather than the downstate suburbs, as judged by uninsured rates that are lower here (29.9 percent) than in the suburbs (38.1 percent).

Regional disparities in skill levels needed for local jobs could also explain some of the observed differences in employer-sponsored insurance, and one way to examine this is by looking at workers’ educational attainment levels. In 11 of the 12 industry categories tracked here, New York City had the state’s highest share of workers without a high school degree or equivalency diploma in 2012. For example, 24.0 percent of adults employed in arts, entertainment, recreation, accommodation, and food services in New York City lack this credential. The comparable percentages upstate and in the downstate suburbs are 9.6 percent and 15.5 percent, respectively.

This pattern is also evident for the industry with the state’s largest share of employees: education, health care, and social assistance. In this industry, 8.4 percent of New York City employees lack a high school degree or equivalency, compared with 2.5 percent upstate and 3.4 percent in the suburbs. The one exception to this general rule occurs within professional and business services. However, the slightly higher share of workers in this industry without a high school degree in the downstate suburbs than in the city is not statistically significant. (Differences in educational attainment between New York City and the rest of the state within public administration and information are also not statistically significant, in part due to the extremely small numbers of workers without high school degrees in these fields.)

Low educational attainment is clearly linked to a low rate of enrollment in employer-sponsored insurance. Statewide, the ESI enrollment rate for employed adults without a high school degree or equivalency diploma is just 33.1 percent, compared with 76.5 percent for those with a high school degree or higher and 84.1 percent for those with an associate’s degree or higher. Thus, the relatively large share of the city’s workforce with low educational attainment may partially explain the lower prevalence of ESI (and related higher incidence of Medicaid) in the city. Even among adults with comparable education levels, however, New York City fares worse than the rest of the state in terms of ESI enrollment. Only 27.8 percent of employed adults without a high school degree have ESI in New York City versus 39.7 percent in the suburbs and 43.5 percent upstate.

Other Factor Affecting Health Insurance Enrollment

Other factors that may affect a region’s ESI and Medicaid enrollment rates are the size of its immigrant population, outreach efforts to encourage enrollment in Medicaid, and the cost of private insurance. IBO’s analysis suggests that the large share of New York City’s labor force that is foreign born does partially explain low ESI enrollment rates here. Medicaid enrollment efforts also appear to be more vigorous in the city than elsewhere, judging from Medicaid take-up rates. Finally, employers in the city and suburbs may be less likely to offer health insurance than those upstate due to higher overall premium costs for private insurance downstate.

Immigration. A larger immigrant population in New York City than elsewhere in the state—37.8 percent foreign born in 2012 versus 20.3 percent in the suburbs and 6.4 percent upstate—provides one potential explanation for city residents’ relatively low rates of employer-sponsored insurance and relatively high rates of Medicaid. While immigrants make up a disproportionate share of the local labor force—47.5 percent of New York City’s employed adults are foreign born—research by the National Immigration Law Center and others has shown that immigrants are less likely than the native born to have employment that includes health insurance benefits.⁷

Evidence for this is also provided by PUMS data for New York State showing that on average, immigrants’ rate of enrollment in ESI is lower than that of the native born population: 41.6 percent versus 61.5 percent. Looking specifically at low-income immigrants (those with family incomes at or below 200 percent of FPL), their rate of enrollment in ESI is 16.0 percent statewide, compared with a 26.1 percent enrollment rate for the low-income native born population. ESI enrollment rates are higher when the sample is limited to employed adults within this income band, but there is an even more sizable gap between the foreign born and native populations: 22.2 percent and 43.6 percent enrolled in ESI, respectively. The share of the population without insurance is also much higher for low-income working immigrants than for their native born counterparts (43.3 percent versus 22.2 percent). Though the ACS does not track individuals’ specific immigration status—beyond whether or not they are citizens and were born outside the U.S.—one possible explanation for this discrepancy is that many of these low income immigrants are undocumented and are thus more likely to work in the underground economy.

New York City’s large immigrant population, especially within the active labor force, may partially explain the city’s low ESI enrollment rates, but not necessarily its high Medicaid rates. As a result of the Aliessa v. Novello New York State court decision in 2001, New York’s rules regarding immigrants’ eligibility for Medicaid coverage are looser than those in federal regulations. New York State provides full Medicaid coverage without a waiting period to otherwise eligible legal residents and to nonlegal residents deemed to be “permanently residing under color of law.”⁸ Thus, over a quarter of immigrants statewide (25.2 percent) were enrolled in Medicaid in 2012. This is statistically significantly higher than the native born population’s Medicaid enrollment rate of 22.1 percent; however, in practical terms the difference between the two is negligible at just 3 percentage points. In contrast, the gap between immigrants and the native born population is much larger—15 percentage points to 20 percentage points—in terms of ESI and uninsured rates.

It is also worth noting that Medicaid enrollment rates are higher in New York City than elsewhere in the state even holding immigration status constant. Twenty-nine and a half percent of the city’s foreign born population is enrolled in Medicaid compared with 12.7 percent in the suburbs and 18.3 percent upstate, though these figures do not account for income differences. Looking just at low-income, employed adult immigrants, New York City still outpaces the rest of the state with a Medicaid enrollment rate of 35.4 percent. The Medicaid enrollment rate for the equivalent group of immigrants is 26.9 percent upstate and 16.7 percent in the suburbs.

Medicaid Take-Up Rates Vary by Region. Our analysis suggests that low enrollment in employer-sponsored insurance—possibly due to low offer rates—contributes to New York City’s high Medicaid enrollment rates. However, ESI enrollment is not the full story. When the PUMS sample is restricted to roughly approximate the population that was eligible for Medicaid in 2012, New York City also has the highest Medicaid take-up rates.⁹ This is true for all three eligibility categories, although the very slight difference between Medicaid take-up rates for seniors and the disabled in New York City versus upstate is not statistically significant.

In contrast, the difference between New York City and the upstate counties is starkest in the case of children. About 87.1 percent of eligible children were enrolled in Medicaid as of 2012 in New York City, compared with 78.3 percent of eligible children upstate, roughly 10 percent less. The differences between New York City and the downstate suburbs are even larger: a Medicaid take-up rate that is 19 percent lower in the case of children (70.7 percent versus 87.1 percent), and 30 percent lower in the case of adults (40.4 percent versus 57.6 percent). New York City also has the state’s lowest uninsured rate among Medicaid eligible children.

These findings suggest that those responsible for Medicaid enrollment in New York City—as of 2012, the Human Resources Administration and facilitated enrollers—are doing a more effective job at signing up eligible individuals than their counterparts elsewhere in the state. Community-based organizations and Medicaid providers not officially designated as facilitated enrollers may also be playing a stronger role in driving up enrollment in the city as compared to other parts of the state. Lastly, there may be less stigma associated with Medicaid enrollment among city residents than elsewhere in the state, which would also play a role in boosting the take-up rate here.

SOURCE: American Community Survey Public Use Microdata Sample 2012

New York City Independent Budget Office

Private Insurance Costs by Region. Another factor that may impact Medicaid take-up rates is the cost of private insurance in the individual and group markets. High costs in the large and small group markets could lead both fewer employers to offer ESI and fewer employees to take-up offered insurance, assuming a large portion of the costs are passed on to workers. Moreover, high costs in the individual market for health insurance could render individuals without access to employer-sponsored insurance less likely to directly purchase their own coverage and more likely to either sign up for Medicaid (if eligible) or remain uninsured.

In fact, pricing data released by New York State prior to the launch of its health care exchange show that up until recently high annual premium costs placed direct purchase insurance out of the reach of many people in all parts of the state. Specifically, the average annual cost for an individual HMO plan in 2013 was $17,800 in New York City, $18,800 in the downstate suburbs, and $15,400 upstate. The average annual cost for a family HMO plan was $54,600 in the city, $57,800 in the suburbs, and $42,400 upstate.¹⁰ Thus, while direct purchase plan costs were lower in the city than in the suburbs, and lower upstate than in either downstate region, they were by no means affordable anywhere in the state.

Employer-sponsored insurance, particularly for large employers, typically costs significantly less than insurance purchased in the direct market. Yet, it may still be considered unaffordable by some employers and employees. Since the state does not publicly track prices in the large group market, ESI premium cost data broken out by county is not available. But a comparison of these costs upstate and in the downstate region as a whole (New York City plus the five suburban counties) is possible using survey data from the U.S. Agency for Healthcare Research and Quality.¹¹ These data show that as of 2012 the average annual premium cost for employer-sponsored insurance was higher downstate than upstate for both individual and family plans. Moreover, the average employer contribution, the portion of the premium paid by the employer, was higher downstate than upstate for both types of coverage. In 2012, employers paid on average $5,200 a year (80.9 percent of the premium costs) for individual coverage downstate and $4,000 (75.7 percent) for individual coverage upstate. For family coverage, the average employer paid $13,700 a year downstate (74.8 percent) compared with $11,100 upstate (74.4 percent). These price differences may make downstate employers less likely to offer health insurance than those upstate.

SOURCE: Agency for Healthcare Research and Quality, 2012 Medical Ependiture Panel Survey-Insurance Component

New York City Independent Budget Office

The average employee contribution towards ESI costs is more consistent across regions. For individual coverage in 2012, the average employee contribution was actually slightly lower downstate than upstate: $1,200 a year versus $1,300. Employee costs for family coverage were on average higher downstate than upstate, averaging $4,600 downstate compared with $3,800 upstate. Judging from these cost data alone, one would expect slightly higher take-up rates for individual ESI coverage in the city and suburbs than in the upstate counties and lower take-up rates for family ESI coverage. PUMS data show that ESI enrollment rates for both employed adults and the children of employed adults are lower in New York City than in the rest of the state. While not conclusive, these findings suggest that employees’ premium costs are probably not the driving factor behind lower ESI enrollment rates (and concomitant higher Medicaid rates) in the city. In contrast, a link between higher employer costs and lower ESI offer rates downstate seems more likely.

Implications for the Future

A larger proportion of New York City’s population is enrolled in Medicaid than is the case in either the downstate suburbs or upstate counties. Demographic differences alone cannot explain this discrepancy, as it persists across all eligibility categories and income levels. More aggressive Medicaid enrollment activities in the city do appear to play some role, but the primary driving factor appears to be the availability of employer-sponsored insurance. A number of factors likely contributed to low ESI enrollment rates in New York City in 2012 including: a smaller share of the population that is employed; a larger share of jobs that do not require college degrees, and a higher incidence of low-wage part-time employment among employed adults; a disproportionately large share of the employed population that is foreign-born who have much less access to ESI, even when controlling for income; and comparatively higher employer costs for private insurance.

Census data also show that while Medicaid enrollment rates in New York City are generally higher than elsewhere in the state, as of 2012 they were not high enough to compensate for low levels of ESI enrollment, leaving the city with the highest share of uninsured residents in the state for many population groups. The one notable exception is very low-income households across eligibility categories (children, adults, seniors, and the disabled). It would appear that New York City’s relatively low uninsured rates for these groups are due to comparatively high Medicaid enrollment rates, a hypothesis supported by the finding that uninsured rates for adults are lower for those at or below the poverty line than for those just above it.

Looking to the future, and more specifically to the potential impact of the Affordable Care Act as it continues to roll out, a number of factors suggest that insurance enrollment patterns in the rest of the state will begin to look more like those in New York City. First, ESI coverage rates have been eroding nationwide for some time and this trend seems unlikely to abate absent significantly higher employer penalties in ACA. Second, as noted in our previous fiscal brief, Medicaid enrollment growth has been stronger upstate and downstate than in New York City since the Great Recession. Lastly, the state is in the process of taking over Medicaid enrollment responsibilities from the counties and has a strong incentive to ramp up its outreach efforts so as to enroll as many eligible individuals as possible. This is because of provisions in the Affordable Care Act that both permanently increase the state’s federal matching rates for Medicaid and decrease national funding levels for uncompensated care. The ACA’s individual insurance mandate as well as a simplified, online enrollment process for Medicaid—introduced for most nonelderly, nondisabled populations through New York State’s health care exchange—should also increase Medicaid take-up rates in the downstate suburbs and upstate counties, making the pattern of Medicaid coverage more uniform across the state.

Report prepared by Christina Fiorentini

Appendix: Additional Comments on the Data

The Medicaid Undercount in the American Community Survey. When Medicaid enrollment totals produced by PUMS data are compared with administrative data from the state Department of Health, it is clear that the ACS undercounts the number of Medicaid enrollees in New York City by a fairly substantial margin. This problem was most notable in 2008, the first year that insurance enrollment questions were added to the American Community Survey questionnaire. Starting in 2009, the Census Bureau began to employ an eligibility edit through which they assigned Medicaid coverage to individuals based on program eligibility rules and their answers to other survey questions. It appears that the quality of New York City’s health insurance data in the ACS improved after that, as the size of the undercount decreased in 2009 and then again in 2010. Nevertheless, the size of the undercount in New York City remained at over 570,000 individuals, or 18.2 percent of total enrollees, in 2012.

The Medicaid estimates produced by the ACS also differ from actual enrollment totals in both the downstate suburbs and the upstate counties, although to date the extent of these differences have been smaller than the differences in New York City. Moreover, the ACS does not consistently underestimate the size of the Medicaid population within these regions as it does in New York City. In the upstate counties, ACS data has overstated Medicaid enrollment in every year since 2009, by as much as 6.7 percent of actual enrollees. Within the 2012 PUMS data used for this report, the size of the overcount was a more manageable 0.7 percent of enrollees. In the downstate suburbs, however, there was an overestimate in just one year (6.9 percent in 2009), and underestimates in all the others. At 13.5 percent of actual enrollees, the undercount in 2012 was larger than those in many previous years, but still smaller than the 18.2 percent undercount in New York City. As a result, any comparisons between Medicaid enrollment rates in New York City and those elsewhere in the state will tend to understate the size of these differences.

As the health insurance questions are still a relatively new addition to the survey, there has not yet been much research exploring why the ACS may be underestimating Medicaid coverage in New York City and elsewhere. However, a Medicaid undercount has been a long and well-established problem with other census data, notably the Current Population Survey (CPS). Research on the Medicaid undercount within the CPS suggests that it is primarily due to two factors.¹² First, the definition of Medicaid coverage and universe of enrollees is not necessarily the same in administrative data as in survey data. People with only partial Medicaid coverage—such as emergency Medicaid or family planning Medicaid—will be counted as Medicaid enrollees in administrative data, but may not report this coverage in a survey. The institutionalized population is also included in administrative data but is not subject to the health insurance questions in the ACS or the CPS.

A second and related factor in the Medicaid undercount is response errors. Specifically, researchers have found that either due to stigma about Medicaid enrollment or confusion about their enrollment status, people underreport their participation. As Medicaid eligibility status can change from month to month, people may not be certain whether they are enrolled at a given time, or may also be confused about whether add on programs such as Family Health Plus are technically part of Medicaid. Suggesting that stigma also plays a role, response errors tend to be correlated with age and income. That is, children are more likely than adults to be reported as having Medicaid coverage, and very low-income enrollees are more likely than relatively higher-income enrollees to report coverage for themselves.

Assessing the ACS as a Representative Sample. Despite these issues, the CPS and increasingly the ACS are widely used by health researchers to study insurance enrollment patterns and trends. In order to assess the appropriateness of the ACS data for our own analysis, we first parsed the population of Medicaid enrollees within the 2012 PUMS by eligibility category and then compared this with the actual composition of the Medicaid population in 2012 using DOH data. As the table on this page shows, the percentages of children, adults, and seniors and the disabled (as a group) are fairly similar in both ACS and DOH data across all three regions. This analysis led us to conclude that the ACS likely includes a fairly representative sample of the Medicaid population in New York State, at least across these eligibility groups and regions. Given issues with under- and overcounting, however, it is more appropriate to look at and compare percentages rather than to use enrollee counts from the ACS.

There are also a number of caveats to our conclusion that the ACS contains a representative sample of the Medicaid population across groups. Foremost, as the publically available administrative data contain no information beyond county of enrollment and eligibility category, there is no way to test whether the ACS contains a representative sample of the Medicaid population in terms of the various employment, income, and demographic variables. The PUMS data are also not perfectly representative even in terms of eligibility categories. While the percentage that is children is almost identical in both data sets across all three regions, adults appear to be under represented and seniors and the disabled over represented in the PUMS Medicaid population. Moreover, the eligibility category breakdowns in ACS and DOH data match up better for New York City than they do for the upstate counties and the downstate suburbs. The break downs for the downstate suburbs seem the most off of the three.

This discrepancy is likely at least partially due to PUMS’s comparatively small sample size in the downstate suburbs: 39,100 individuals versus 70,100 in New York City and 87,100 upstate. The fact, however, that the split between adults and seniors and the disabled is also off upstate suggests that Medicaid stigma and/or the over reporting of disabilities in the ACS may also play a role. The disability variable in the ACS is particularly problematic, as it is self-reported and does not necessarily align with Medicaid’s definition of disability. When the disabled are examined independently from the senior population in PUMS, the demographic breakdown for these two groups varies considerably from what administrative data show. As a result, our analysis examines seniors and the disabled as one and not two distinct subgroups within the Medicaid population.

An Additional Note on Industry Classifications. The industry definitions used in this paper were formulated using the INDP, or industry recode, variable within the ACS. The list of all INDP codes included in each of the broader industry categories is as follows:

• 0170–0490: Agriculture, Forestry, Fishing, and Hunting; Mining (not included in this analysis)

• 0770: Construction

• 1070–3990: Manufacturing

• 4070–4590: Wholesale Trade

• 4670–5790: Retail Trade

• 0570–0690 or 6070–6390: Transportation and Warehousing; Utilities

• 6470–6780: Information

• 6870–7190: Finance and Real Estate

• 7270–7790: Professional and Business Services

• 7860–8470: Educational Services; Health Care and Social Assistance

• 8560–8690: Arts, Entertainment, and Recreation; Accommodation and Food Services

• 8770 –9290: Other Services (except Public Administration)

• 9370–9590: Public Administration9670–9870: Military (not included in this analysis)

Endnotes

¹U.S. Census Bureau, Current Population Survey, Annual Social and Economic Supplements, 2003 and 2013.
²The Medicaid eligibility categories used throughout this report are defined as follows. Children: individuals age 18 or younger without a reported disability (based on the recoded disability variable in the ACS). Adults: individuals ages 19 to 64 without a reported disability. Seniors and the disabled: individuals age 65 or older and/or those who report a disability. Unless otherwise indicated, all references to “adults” or “children” refer specifically to nondisabled adults or children.
³Unless otherwise indicated, all differences in health insurance enrollment (and uninsured) rates between New York City and other regions in the state are statistically significant at p ≤ 0.01. This is also true where the sample has been further parsed by income level, employment status, and other variables.
⁴In all cases, income categories are determined by looking at family income in relation to the Federal Poverty Level, or more specifically, by using the POVPIP variable in the ACS. Family income was used rather than household income because Medicaid eligibility is based upon family income.
⁵At that time, slightly higher income thresholds applied to parents and pregnant women than for childless adults. Effective January 1, 2014 and as part of the ACA, the income threshold for Medicaid eligibility has been raised to 138 percent of FPL for all adults.
⁶Part-time workers are defined here as individuals whose usual hours worked per week over the past year were at least 1 but less than 30 hours. Under the ACA, firms with 50 or more full-time workers will eventually be required to provide affordable insurance coverage to all employees who work an average of 30 hours per week or more.
⁷National Immigration Law Center, “Analysis of the Massachusetts Health Care System as a Model,” May 2009; Buchmueller et al, “Immigrants and Employer-Sponsored Health Insurance,” Health Services Research vol 42, Feb 2007.
⁸A person “permanently residing under color of law” is living in the country with the knowledge of immigration officials and without threat of deportation. He or she is considered to have the same rights as legal residents for welfare eligibility purposes in New York State. Undocumented immigrants without PRUCOL status are eligible only for emergency Medicaid.
⁹Medicaid eligibility is defined here as a lack of ESI and a family income of ≤ 400% FPL for children, or ≤ 100% FPL for adults, seniors, and the disabled.
¹⁰New York State Department of Financial Services, Premium Rates for Standard Individual Health Plans, July 2013. Note that average premium prices for 2014 decreased, in some cases dramatically, as a result of the ACA.
¹¹U.S. Agency for Healthcare Research and Quality, Center for Financing, Access, and Cost Trends; 2012 Medical Expenditure Panel Survey–Insurance Component.
¹²Research Project to Understand the Medicaid Undercount: The University of Minnesota’s State Health Access Center et al, Phase II Research Results: Examining Discrepancies between the National Medicaid Statistical Information System (MSIS) and the Current Population Survey (CPS) Annual Social and Economic Supplement (ASEC), March 2008.