key: cord-0709437-3blserlh
authors: Sohn, Heeju
title: Will you be covered during the next recession? Unequal safety-nets for private health insurance in the United States
date: 2020-05-22
journal: Health policy open
DOI: 10.1016/j.hpopen.2020.100006
sha: fa26b39350d0a816711b84c614e06f2c06474722
doc_id: 709437
cord_uid: 3blserlh

Abstract Health insurance coverage for working-age adults in the United States is still predominantly determined by their jobs or their spouse’s jobs. This article revealed that SES disparities in access to a spouse’s coverage as a safety-net significantly contributed to the inequities in coverage loss during economic instability. Using the longitudinal Survey of Income and Program Participation (1996-2012), this article examines insurance transitions between socioeconomic groups during the two most recent recessions in the United States. The SES disparity in private coverage spiked during periods of high job-loss. Higher SES men’s and women’s coverages were barely affected as they became insured by their spouses even as they lost their own. Wives’ insurance plans played a large role in mitigating declines in higher SES men’s coverages during the 2008 recession.

J o u r n a l P r e -p r o o f likely to be full-time employees in industries that offer employment-based health insurance and their rates of private coverage remained stable since the 1990s (Fronstin, 2015; Holahan & Cook, 2008; Kalleberg, Reskin, & Hudson, 2000; Reschovsky, Strunk, & Ginsburg, 2006) . On the other hand, jobs available to people with less education shifted from industries with high offer-rates of employment-based coverage (i.e., manufacturing) to industries with low offer-rates (i.e., service sector) (Holahan & Cook, 2008) . These jobs were more susceptible to economic downturns, and more low SES people lost their insurance coverage during 2001 and 2008 recessions (DeNavas-Walt et al., 2013; Farber, 2015; Farber & Levy, 2000; Holahan, 2011) . The recessions amplified health coverage inequalities between education groups (Grusky, Western, & Wimer, 2011; Reschovsky et al., 2006) .

The literature is relatively sparse on how a spouse can affect insurance coverage despite the fact that over 79 million people were insured as a dependent in 2014 and more than half of married women were insured through their husbands (Fronstin, 2015; Sohn, 2015) . Being married enabled access to their spouses' employment-based insurance plans, and many rely on their spouses' plan when they lose their own job-based coverage (Abraham & Royalty, 2005; Montez, Angel, & Angel, 2009; Short, 1998; Vistnes & Monheit, 2011) . The strong correlation between a person and their spouse's educational attainment (Schwartz & Mare, 2005) suggest that high SES people are more likely to have high SES spouses with jobs that also offer health insurance. If a one lost their job and insurance during a recession, presumably, they could have maintained coverage by switching over to their spouse's plan. This alternative safety-net may not have been as readily available for people with low SES.

Using nationally representative longitudinal data from the U.S., this article quantifies the degree to which access to a spouse's health insurance plan exacerbated the health coverage disparity between low and high SES people during the two most recent recessions.

The study uses four panels of the Survey of Income and Program Participation (SIPP) to examine sources of insurance by gender and educational attainment among married people. The SIPP is a nationally representative series of longitudinal panels whose survey duration ranges from 2.5 to 4 years. The first SIPP panel was sampled in the early 1980s, and a new panel was re-sampled from the non-institutionalized population in the U.S. every one to four years. The analyses use the SIPP's 1996 SIPP's , 2001 SIPP's , 2004 SIPP's , and 2008 The study also utilizes specific waves from these SIPP panels to compare insurance transitions during periods of high-and low-job loss. Waves 11 and 12 from the 1996 panel In addition to the same core questionnaire that the SIPP administered every wave, the SIPP includes topical modules containing a different theme. These topical modules contain detailed questions on selected topics, but the topics and questions differ across the duration of the panel. I utilized the Employer-Provided Health Benefits topical module to examine the sources of employer-sponsored health insurance and whether they are enrolled in those plans. This module was only administered once in each panel and is only available for years 1997, 2003, J o u r n a l P r e -p r o o f to consider only the first reference month of each wave to evaluate respondents' insurance statuses. This method allows insurance status change to occur only once in a four-month period.

The analyses are limited to the sample of married adults aged 26 to 64 who are also married to adults in the same age range. Adults aged 65 and over are almost universally eligible for Medicare and health insurance coverage is less dependent on employment and marriage. In 2010, Young Adult Provisions' of the Affordable Care Act (ACA) allowed adults under age 26 to be insured as a dependent on their parents' plans. The study restricts the sample to couples who are not eligible to be insured by a parent or by Medicare without a medical need to limit the sources of private insurance to an individual or his or her spouse. It drops less than 2% of individuals who were married to someone absent from the household (institutionalized or living elsewhere). It also limits the sample to couples who remained married for at least two consecutive waves to calculate insurance transitions within an intact marriage. The final analysis sample comprised 131,999 individuals aged 26-64 in 79,128 marriages. If a household reference person divorced and remarried during the panel, this would be considered two separate marriages. While some employers offer family health coverage to domestic partners, most limit eligibility to legally married spouses. Some states were beginning to recognize same-sex marriages between 1996 and 2012. However, same-sex couples did not appear in the SIPP panels as married spouses. Thus, the analyses here are limited to legally married, heterosexual couples.

Educational Attainment The study divides the population into two groups: persons with a fouryear college degree (B.A. or equivalent) or higher and persons without a college degree.

Educational attainment roughly captures a person's human capital and potential earning. Unlike J o u r n a l P r e -p r o o f income, another common measure of socioeconomic status, educational attainment is less sensitive to idiosyncratic employment decisions and macroeconomic shocks. Also, it generally remains stable after marriage. Using educational attainment as a measure of SES is particularly suited here as it remains constant throughout the major recessions, unlike jobs, income, and earnings. Using an alternative threshold for low-and high-SES groups (high school grad and below versus some college level education and above) did not yield meaningfully different results.

Health Insurance Status I categorize individuals into three discrete insurance categories:

privately insured by their own plan, privately insured as a dependent, and not privately insured.

Private insurance includes both employer-sponsored plans and individually purchased plans. The analyses divide private insurance coverage into two categories-insured by their own plan and insured as a dependent-to observe how husbands and wives insure each other. The third category includes people with public coverage (Medicaid and Medicare) and the uninsured. The core multivariate analysis combines public coverage and uninsured to maintain an adequate number of observations in each group-insurance-period category. A descriptive analysis that separates public insurance from uninsurance shows that public insurance coverage among married men and women remained low throughout both recessions.

The SIPP's monthly core surveys do not ask dependents whether the primary subscribers of their insurance plans are their spouses. The SIPP however, asks who the primary subscriber is in their Employer-Provided Health Benefits topical module administered in 1997, 2002, 2005, and 2010 . At each of these times, the vast majority (around 95%) of married dependents aged 26 to 64 were insured by their current spouses' plans. Therefore, I consider most dependents to be insured by their spouses throughout the entire study period. The Employer-Provided Health J o u r n a l P r e -p r o o f Benefits topical module also asks respondents not only their actual coverage but also whether they are eligible to enroll in their own employer-provided plans. For example, a person may be insured as a dependent on their spouse's plan while having the option to switch to another plan through their own employer. This once-per-panel topical module captures multiple sources of insurance coverage that some couples can draw upon to insure themselves and each other.

Socio-demographic controls When comparing transitions between groups and across time periods, multivariate analyses control for characteristics that prior research has found to be related to insurance status: employment, race, Hispanic origin, age, and logged family income.

Individuals who work full-time are more likely to have private coverage through their employers (Fronstin, 2015) . Racial and ethnic minorities are less likely to have insurance coverage, and private coverage increases with age and family income (Fronstin, 2015) . Employment and income are measured before insurance transitions.

The core analysis compared the insurance characteristics between a college graduate and nongraduate married individuals during periods of high unemployment associated with the 2001 and 2008 recessions accounting for baseline differences and possible confounding characteristics. I conducted separate analyses for men and women.

I examined each group's insurance gains and losses directly to better capture the sudden changes in insurance coverage conditions before and during the two recessions. Cross-sectional distributions react slowly to changes in transition conditions and can mask year-to-year shifts in how likely people are to lose and gain insurance. During the study period, people moved in and out of insurance groups becoming uninsured or switching coverage to their spouses' plans. Some J o u r n a l P r e -p r o o f years, people may be more likely to lose coverage, and in other years, people may be more likely to gain coverage. I used multinomial logistic regressions to estimate probabilities of switching from one insurance category to another while controlling for covariates: age, race, Hispanic origin, employment, and logged family income. Separate regressions estimated transition probabilities for each gender, education, and staring insurance category. I then converted these estimated probabilities into annualized rates and used these rates to derive a distribution that would emerge if the rates remained constant for a long time (also known as a stationary distribution). I compared the changes in the stationary distribution before and during recessions between low and high SES men and women. Table 1 summarizes the education and insurance characteristics of married couples aged 26 to 64 from four SIPP panels. The percentages are weighted to reflect the U.S. population.

The level of education among married people aged 26 to 64 increased between 1996 and 2008.

These trends reflect the rise in educational attainment, particularly among women, as well as the increasingly positive association between education and marriage throughout this period. More than half of all married women were insured as a dependent, and about a third of married women were primary subscribers. Overall, about 46 percent of married individuals were the primary subscribers of their private health plans. Private coverage was lower among the more recent cohorts, and the proportion without private coverage peaked in the 2008 panel likely due to the recession. At the couple-level, more than half of couples had access to only one health insurance plan between the two of them (bottom panel of Table 1 ). More than a quarter of couples have college-graduate women's greater propensity to have her own insurance plan. When job loss rose, however, the difference in spousal insurance expanded the coverage gap between collegegraduate and non-graduate women.

Spousal coverage was also a significant source of coverage inequality among married men (Figure 2 ). Disparities in private insurance coverage between college-graduate and nongraduate men had been mostly driven by their differences in having their own plans. In fact, when job loss was low, and job gain was relatively high in 1999 and 2007, college-graduate men were less likely than their non-graduate counterparts to be insured as a dependent. During periods of high job loss, however, insurance coverage through a spouse's plan became a prominent driver of inequality. Coverage gap hovered around 14 percentage points during prerecession periods. In 2009, a period of peak job loss in the United States, the coverage gap reached 25 percentage points, and 15 percent of that difference was due to spousal coverage.

[ Table 2] The proportion of married non-college women with own insurance plans dropped by 12.1 percentage points from 54 percent to 41.9 percent during the 2001 recession. About half of those who lost their own insurance plan switched to their spouses' plans (Table 2) 

The results collectively show the extent to which spousal insurance matters in expanding coverage inequality between education groups during periods of economic instability.

Individuals with more education were less likely to lose their own coverage and were also more likely to be covered by a spouse during past economic crises. Additional analyses revealed that public health insurance (Medicaid) did not provide adequate safety nets for those losing private coverage during this period. Consequently, uninsurance rates were disproportionately greater among groups with less education. With the insurance coverage gap expanding along existing socioeconomic boundaries, the burden of poor health became disproportionately greater among already disadvantaged families.

This article examines insurance coverage between spouses during specific periods of unusually widespread economic instability. However, its findings can be cautiously generalizable J o u r n a l P r e -p r o o f to other periods and populations. First, spousal insurance acts as a safeguard against gaps in coverage in situations other than job loss during recessions. Individuals may rely on their spouses for insurance as they transition between schools and jobs or when engaging in selfemployment. Spousal insurance allows people to pursue riskier opportunities with greater potential payout. Its role in coverage inequality, however, is reduced during periods of economic stability. In fact, the analyses find that more individuals without a college degree (especially men) relied on their spouses' plans during periods of low job loss and unemployment. When the economy is good, being able to share insurance plans between spouses helps lower SES families.

Still, strong correlations between wages, job stability, and health benefits (Fronstin, 2015;  Kalleberg, 2011) make lower SES families more vulnerable to changes in economic cycles. And while components of the ACA provides some protection through Medicaid eligibility expansions and insurance exchanges, employers are still expected to remain the largest source of health insurance for non-elderly Americans (Congressional Budget Office, 2012).

Second, being able to share an insurance plan among immediate family members can prevent gaps in coverage for adult children and cohabiting partners as well as spouses. The Young Adults' Provision of the ACA allowed adults to be insured by their parents' private plans until age 26. Adult children of higher SES parents are more likely to have uninterrupted private coverage than their lower SES counterparts; they can switch to their parents' plans should they lose their own. The advantages of family-based insurance plans may also apply to cohabiting couples. While fewer employers extend benefits to domestic partners than to married spouses (Employee Benefits Research Institute, 2004), a greater proportion of cohabiting households have two earners (Bureau of Labor Statistics, 2016). Family insurance plans may also be a sizable contributor to coverage inequality among those who are in cohabiting relationships.

Unfortunately, as SIPP does not recognize cohabiting relationships, the analyses here are limited to married couples.

The results underestimate the full extent of spousal insurance's effect on coverage inequality. A person's access to private insurance plans can affect insurance coverage for their children as well. Children with parents with a college degree are more likely to have uninterrupted private coverage than their lower SES counterparts for the same reasons; when one parent lose access to coverage, everyone in the family will be more likely to be able to switch to the other parent's plan. While children have greater access to public insurance options such as Medicaid or SCHIP, one in twenty children under 18 are without insurance coverage (Kaiser Family Foundation, 2017) , and transitions between private and public coverage can be cumbersome to navigate (Summer & Mann, 2006) . The greater precariousness of marriages among couples with less education (Martin, 2006; Schwartz & Han, 2014) means that spousal insurance is even less reliable among lower SES individuals, The analyses exclusively examines insurance transitions of couples in intact marriages and excludes coverages of people who lost their spouse along with its privileges.

Along similar lines, spousal coverage exacerbates coverage differences between married and single individuals. Single adults over 25 generally do not have another source of private coverage other than their own labor force participation. Positive selection into marriage means that the single population is more likely to have fewer resources and poorer health (Goldman, 1993; Lillard & Panis, 1996; Waite, 1995) . Without access to a secondary source of coverage, single people, especially those with less education, are already at high risk for experiencing the negative consequences of living without health insurance. The coverage gap between married J o u r n a l P r e -p r o o f people with a college degree and single people with less education is likely to be even greater during economic crises.

As the U.S. edged closer to the precipice of another recession due to the COVID-19 pandemic, extricating marriage from health insurance coverage gained new urgency (Pollitz & Claxton, 2020) . In April 2020, more than 5 million people claimed unemployment claimed each week (U.S. Department of Labor, 2020). Loss of employer-sponsored insurance coverage following job loss is likely to be uneven across socioeconomic groups. Providing a viable way to continue insurance coverage for all newly unemployed workers is critical in curbing the longstanding injustices (Elo, 2009) . Throughout the pandemic, people with fewer resources were far more likely to be infected and were more likely to die from the novel coronavirus (Butler & Barrientos, 2020; Dietz & Santos-Burgoa, 2020) .

Since 2014, 40 U.S. states expanded Medicaid eligibility to cover more low-income adults regardless of pregnancy or marital status. Marketplace premium tax credits gave another means for health insurance coverage for people with moderate income-100 to 400 percent Federal Poverty Level (FPL)-who were not covered through an employer-sponsored plan.

Expanding the coverage safety-net led to reductions in the SES disparity (Courtemanche, Marton, Ukert, Yelowitz, & Zapata, 2016; Sommers, Maylone, Blendon, Orav, & Epstein, 2017) in participating states, and may have lessened the impact of the COVID-19 pandemic. Conversely, states that did not expand Medicaid likely experienced surges in coverage inequity; many lowincome people in these non-expansion states were left ineligible for Medicaid and Marketplace subsidies (Garfield, Orgera, & Damico, 2020) .

The current patchwork of programs that differs state by state and by eligibility criteria is J o u r n a l P r e -p r o o f 1996, 2001, 2004, and 2008 J o u r n a l P r e -p r o o f 

Does Having Two Earners in the Household Matter for Understanding How Well Employer-Based Health Insurance Works?

Unemployment Rate from the Current Population Survey

Employment Characteristics of Families-2015

The impact of nutrition on COVID-19 susceptibility and long-term consequences. Brain, Behavior, and Immunity

Financial Burden of Medical Care: A Family Perspective

The Effects of the Affordable Care Act on Employment-Based Health Insurance

Impacts of the Affordable Care Act on Health Insurance Coverage in Medicaid Expansion and Non

Determinants of Health Insurance Status among Young Adults

Trends in Marital Dissolution by Women's Education in the United States

Employment, Marriage, and Inequality in Health Insurance for Mexican-Origin Women

Health Insurance and Vulnerable Populations

Health Policy and the Uninsured

Changes in Income and Health Coverage Eligibility After Job Loss Due to COVID-19

Why employer-sponsored insurance coverage changed

The Reversal of the Gender Gap in Education and Trends in Marital Dissolution