Hispanic workers and employer sponsored health insurance.

Paringer, Lynn

I. BACKGROUND

Forty-five million Americans currently lack health insurance. Most of these individuals reside in households in which at least one family member works (Collins et al., 2004). Nearly 70 percent of insured Americans obtain their coverage through their own employer or the employer of another family member. Yet, there is mounting evidence that this system is on the verge of collapse. Between 2000 and 2004 employer-sponsored health insurance fell from 68% to 64% for adults in the United States and from 61% to 58% in California (Dube et al. 2005).

Uninsured workers are heavily concentrated among those with minimum skills and low-wages (Quinn, 2000, Glied et al. 2003). Only 67 percent of low-income workers in large firms are offered employer sponsored health insurance compared to 89 percent of mid and high-income workers. Given their relatively low education and earnings levels, Hispanics are particularly disadvantaged with respect to access to employer sponsored health insurance (Schur and Feldman 2001, Stanton 2004). Although this population group is more likely than other groups to be stably employed, as many as 80 percent of low-income Hispanics were uninsured at some point between 1996 and 1999 (Doty and Holmgren 2004). Fronstin (2005) reports that nearly 60% of employed Hispanics reported that their employer did not offer employer sponsored health insurance compared to 37.5% of whites and 38.4% of Blacks in 2002. The plight of Hispanic workers, particularly those in California, is interwoven with other characteristics such as minimal English skills and lack of a high school education; characteristics which relegate many Hispanic workers to low paying jobs with high turnover and an inability to access employment-based health insurance.

There have been a number of studies to date that attempt to explain firms' decisions to offer health insurance to their employees. Studies by Hadley and Reschovsky (2002) and Feldman et al. (1997) view firms as agents for their employees, offering insurance if their employees' collective reservation price exceeds the price at which the employer can make insurance available. This static approach views firms as making a yes/no decision with respect to the offering of health insurance and ignores many of the ways in which firms might offer insurance to employees but effectively limit employees' access to insurance through restrictions such as requiring a minimum tenure with the firm, requiring a minimum numbers of hour of work per week to receive benefits (and then only hiring certain classes of workers to work for fewer hours than the minimum), outsourcing certain low-wage positions (e.g. janitorial) to temporary help firms that may not offer health insurance to their workers, etc.

Numerous other studies have used the individual worker as the unit of observation. Polsky et al. (2005) examined the relationship between the type of plans offered by employers and employee decisions to enroll in various plans. However, this study treated as exogenous the worker's decision to accept employment at a particular firm. Dushi and Honig (2005) examined the role that race and ethnicity play in the likelihood that a worker is offered employment sponsored health insurance and on the probability, that, once offered employer sponsored health insurance, the worker takes up coverage. Monheit and Vistnes (2000) analyzed changes from the mid-1980s to the mid 1990s in offers, take-up and insurance coverage by race. They found a significant decline in employer sponsored health insurance offer rates being accorded to Hispanic men over this time period. Waidmann, Garrett and Hadley (2004) found that Hispanic families had both lower offer and take-up rates than non-Hispanic whites.

Because of the high correlations between Hispanic ethnicity, education, immigration status, and English language skill, it is often difficult to separate the effect each characteristic plays in affecting access to employer sponsored health insurance. In addition, many studies of employee access to employer sponsored health insurance rely on national data bases making it difficult to control for labor market and public policy characteristics that may vary from state to state (e.g. worker compensation, Medicaid policies, safety net providers, insurance regulation).

This paper provides information on the factors that affect access to employer sponsored health insurance and assesses the role that firm, worker employment, worker demographic and worker health characteristics play in affecting access to insurance among California's Hispanic work force. It also examines the factors that impact take-up of employer sponsored health insurance by workers. In particular, we examine how much of the variation in access between Hispanics and non-Hispanics can be explained by such factors as education, earnings, language skills, health status and immigration status.

II. DATA

The data used in this study is from the 2000 California Work and Health Survey (CWHS). The CWHS is a telephone-based, longitudinal survey of California adults aged 18 and over designed by faculty and staff at the Institute for Health Policy Studies at the University of California, San Francisco. The survey contains considerable information on the insurance status of the survey participants both at the time of the survey and during the year prior to the survey. Information on employer sponsored health insurance as well as the characteristics of the firms at which the individual works, and the individual's employment, earnings, demographic and health characteristics were collected. The 2000 CWHS was administered between May 1 and July 1, 2000 and includes 2,168 California adults. The CWHS data are uniquely suited to address the question of which population groups in California are disadvantaged with respect to employer sponsored health insurance. We limit the analysis to workers between the ages of 18 and 64 since individuals over age 65 are eligible for Medicare (the survey itself is limited to those 18 and over).

III. CONCEPTUAL FRAMEWORK

Firms have a financial incentive to offer health insurance coverage to workers. Preferential tax treatment of employer sponsored insurance and other fringe benefits induce firms to substitute fringe benefits for wages. Firms use pretax dollars to pay for health insurance and workers receive employer based health insurance as nontaxed compensation. Because of the tax deductibility of insurance premiums when insurance is obtained through the workplace and the lower cost to employers of obtaining insurance coverage in the group market, the cost of obtaining insurance coverage through the employer is substantially lower than it would be if the worker had to obtain insurance through the individual market.

Worker willingness to search out employment with a firm that offers employer sponsored health insurance depends on their tolerance for risk, their anticipated need for medical care, their assets, the availability of insurance through another family member, the cost of insurance if they were to purchase an individual policy, and the quality and accessibility of safety net providers in the community. Since workers differ in their marginal tax rates, the value of employer sponsored health insurance is higher for workers with higher marginal tax rates than for workers with low marginal tax rates. In 2005, the annual premium for family coverage was $10,880, more than the gross earnings for a fulltime minimum wage worker ($10,712) (California HealthCare Foundation 2005). Thus, employer sponsored health insurance constitutes an extremely large share of the total compensation package for many workers, particularly those at the low end of the wage spectrum.

A worker views his/her compensation as being composed of wage compensation plus the value of fringe benefits minus any cost he/she must incur to participate in the fringe benefits. Consequently, the higher the cost sharing a worker faces, the lower the net value of employer sponsored health insurance. Workers may self-select into those jobs that offer employment sponsored health benefits based on the expected net value of the benefits.

We draw on utility maximization theory in developing our model of demand for employer sponsored health insurance (Phelps 1973). We assume that workers maximize the utility derived from a particular mix of compensation and select jobs accordingly. Compensation is composed of wage compensation and employer sponsored health insurance. A worker's level of risk aversion, asset position, marginal tax rate, health status, and access to health insurance through another family member, may all affect the value that a given worker places on employer sponsored health insurance and will affect the mix of compensation that the worker seeks in an employment relationship.

IV. COMPARING HISPANIC AND NON-HISPANIC WORKERS

Hispanic workers in the state of California differ dramatically from non-Hispanic workers along a variety of dimensions but are also quite similar to their non-Hispanic counterparts in some very important areas (Table 1). To begin with, Hispanic workers are significantly less likely than non-Hispanic workers to be offered employer-sponsored health insurance (66.9% versus 78.7%) They are also less likely to have health insurance from any source (own employer, employer of another family member or other private source). However, Hispanic workers are not significantly different from their non-Hispanic counterparts in terms of being either part-time workers, having relatively short tenure with their firm, or being covered with a union contract. However, Hispanic workers are significantly more likely than non-Hispanic workers to earn less than $20,000 and less likely to earn more $40,000. Over one-third of Hispanic workers did not graduate from high school compared to less than 3% of non-Hispanic California workers. Hispanic workers are younger, less likely to be female, more likely to be married and more likely to be parents than non-Hispanic workers. Slightly over 40% of Hispanic workers in California were born in the United States compared to over 84% of non-Hispanic workers. Just over two in five Hispanic workers in California speak English at home compared to 93.4% of non-Hispanic workers. Hispanic workers are more likely to reside in families with an income below the poverty level and more likely to be in poor health than non-Hispanic workers.

Table 2 shows the offer rates of employer sponsored health insurance by selected worker characteristics. Two out of five part-time workers were offered employer sponsored health insurance compared to over four out of five full-time workers. Workers with less than one year of tenure with the firm were less likely to be offered employer sponsored health insurance than their longer tenured counterparts. Less than half of workers earning less than $20,000 were offered employer sponsored health insurance compared to over 90% of workers earning more than $40,000. Workers in small firms were less likely to receive offers of employment sponsored health insurance. This likely reflects the fact that small firms face higher costs of offering health insurance due to higher administrative costs and their inability to risk pool. Workers covered by union contracts were more likely to receive offers of employment-sponsored health insurance than workers not covered by union contracts (Fronstin 2005). Workers in retail trade/housekeeping were less likely to receive an offer of employment sponsored health insurance than workers in manufacturing.

Immigrants, workers whose primary language was other than English and workers with less than a high school education were less likely to receive offers of employment sponsored health insurance from their employers than non-immigrants, workers whose primary language was English and workers with a high school education. Asians were more likely to receive offers of employment sponsored health insurance than workers of other races. Those in excellent/very good health were more likely to receive an offer of employer sponsored health insurance than those in fair/poor health.

Because Hispanic workers are more likely to have characteristics that are associated with low levels of access to employer sponsored health insurance, we compared the percent of Hispanic and non-Hispanic workers offered employer sponsored health insurance within education, immigration, age, earnings and language categories. (Table 3). Hispanic workers with a high school education, those born in the United States, those earning the same income as non-Hispanic workers and those who speak English at home face virtually the same probability of being offered employer sponsored health insurance as their non-Hispanic counterparts. Hispanic workers born out of the United States have a lower probability of being offered employer sponsored health insurance than non-Hispanics. However, this reflects the fact that the other major employee group of California workers that was born outside of the United States is those born in Asia and they have considerably higher levels of education than Hispanic workers. The same is true for those whose primary language at home is other than English. Thus, it appears that the level of education, earnings and language proficiency may have a significant effect on explaining the low offer rates of employer sponsored health insurance accorded to California workers.

V. MULTIVARIATE ANALYSES

In conducting our analysis we draw on other research in the field that views coverage under employer sponsored health insurance as a three stage process (Waidmann, Garrett and Hadley 2004). First, the individual selects into employment. Second, given the individual has selected into employment, the individual selects a position that offers employer sponsored health insurance. Third, given the worker is offered employer sponsored health insurance, the worker takes up the offer. (1) In the first stage of our analysis, we estimate an employment equation:

[EMP.sub.i] = [[beta].sub.0] + [[beta].sub.1]WORKERDEMO + [[beta].sub.2]WORKERHEALTH + [[epsilon].sub.i]

Where:

EMP = 0,1 a dummy variable which takes on the value of 1 if the individual is employed, and 0 otherwise.

WORKERDEMO = a set of dummy variables reflecting the demographic characteristics of the individual including age, sex, education, marital status, ethnicity, race, immigration status, language and parent status. (2)

WORKERHEALTH = a set of dummy variables indicating the self assessed health status of the worker.

[[epsilon].sub.i] = error term

In the second stage of the model, we examine the offer of employer sponsored health insurance. This equation is estimated only for those individuals that are employed.

Because firms can not discriminate by ethnicity in terms of who is offered health insurance, if workers do not engage in job sorting, (i.e. workers select employment in certain jobs depending on the compensation mix they desire) we would not expect worker demographic characteristics such as ethnicity, age and race to be significant in terms of affecting the probability that a worker is offered employer sponsored health insurance. For example, let us assume that Hispanic workers are more likely to work for small firms and small firms are less likely to offer employer sponsored health insurance than large firms. In this case we might observe a significant relationship between the employer sponsored health insurance offer and firm size but we should not observe a significant relationship between a health insurance offer and Hispanic ethnicity. However, if we omit firm size from the regression we might observe a negative relationship between employer sponsored health insurance offers and Hispanic ethnicity. The relationship should disappear once firm size is included in the equation. However, if workers do sort themselves into jobs based on their preferences for employer sponsored health insurance, and if these preferences are related to worker demographic characteristics, we might see a significant relationship between worker demographic characteristics and being employed in a job that offers employer sponsored health insurance. To examine the relationship between firm characteristics, worker employment characteristics, worker demographic characteristics, worker health status and the probability of receiving an offer of employer sponsored health insurance, we estimate the following equation:

[ESHI.sub.i] = [[beta].sub.0] + [[beta].sub.1]FIRMCHAR + [[beta].sub.2]WORKEREMP + [[beta].sub.3]WORKERDEMO + [[beta].sub.4]WORKERHEALTH + [[epsilon].sub.i]

Where:

ESHI = 0,1 a dummy variable which takes on the value of 1 if the worker received an offer of employer sponsored health insurance and 0 otherwise.

FIRMCHAR = a set of dummy variables reflecting the industry and size of the firm.

WORKEREMP = a set of dummy variables indicating whether the worker works part-time, has less than one year of tenure with the firm and earns under $20,000 or between $20,000 and $39,999 per year.

WORKERDEMO = a set of dummy variables reflecting the demographic characteristics of the worker including age, sex, marital status, ethnicity, race, parental status, immigration status, and primary language. WORKERHEALTH = a set of dummy variables indicating the self assessed health status of the worker

[[epsilon].sub.i] = error term.

Whether or not a worker is actually insured through his/her employer depends on whether an offer of employment sponsored insurance is taken up by the worker. This depends on the price that the worker faces for such insurance and on a host of other factors such as the worker's risk tolerance, availability of insurance from other sources, ability to pay, and perceived need for services. The CWHS does not have information on the premium the employee must pay for employer sponsored health insurance or on the characteristics of the actual policies themselves. (3) However, certain characteristics of the work environment such as firm size, union coverage, and work status of the employees have been found to be related to the premium the worker must pay for coverage and we use these as indirect measures of price.

The take-up equation we estimate is:

[TAKEUP.sub.i] = [[beta].sub.0] + [[beta].sub.1]FIRMSIZE + [[beta].sub.2]WORKERDEMO + [[beta].sub.3]WORKEREMP + [[beta].sub.4]WORKERHEALTH + [[epsilon].sub.i]

Where:

TAKEUP = 0,1 a dummy variable which takes on the value of 1 if the worker accepts employer sponsored health insurance and 0 otherwise.

FIRMSIZE = a set of dummy variables reflecting the size of the firm (proxy for employee share of cost).

WORKEREMP = a set of dummy variables indicating whether the worker works part-time, has less than one year of tenure with the firm and earns under $20,000 or between $20,000 and $39,999 per year.

WORKERHEALTH = a set of dummy variables indicating the self assessed health status of the worker.

[[epsilon].sub.i] = error term.

Because the dependent variable in each equation is dichotomous, we use binary logistic analysis to estimate the equations. We estimated models for two different groups. The first set of equations includes all adults age 18-64 (Table 4). The second set of equations is estimated only for unmarrried adults 18-64 (Table 5). We do this because accessing employment opportunities that offer employer sponsored health insurance is likely to be jointly determined in households where the partners are married. (4) If one partner has access to employer sponsored health insurance that can cover the entire family, another worker in the family may be less concerned with whether or not employer sponsored health insurance is included in the compensation package. By restricting analysis to unmarried workers we greatly diminish the likelihood that employer sponsored health insurance can be accessed through another family member.

We anticipate that individuals with higher levels of education are more likely to be employed than less educated workers. Given the results of other literature in the area of labor supply, we expect that married women will be less likely to be employed than either single women or men. Younger adults are more likely to be transitioning between school and work and more likely to be moving from job to job so we expect that workers 18-24 will have a lower probability of employment than older workers. However, given the trend toward early retirement, we expect individuals over 55 to be less likely to be employed that their younger counterparts. Those with minimal English skills are expected to be less likely to find employment than those with greater English competency.

The results of the first (Employed) stage of our analyses are presented in the first two columns of Tables 4 and 5. Table 4 results pertain to all adults aged 18-64 and Table 5 results pertain only to unmarried adults 18-64. With respect to Table 4, we find that married women are less likely to be employed than either men or unmarried women. Less educated adults are also less likely to be gainfully employed than those with higher levels of education. Those in excellent to very good health are more likely to be employed than those in good health and those in fair to poor health are less likely to be employed than those in good health. The youngest age group (18-24) and the oldest age group (55-64) are less likely to be employed than those aged 25-34. Among all adults 18-64, those whose primary language at home is other than English are less likely to be employed than those who speak English in the home environment. Interestingly, we find that Hispanics and immigrants are less likely to be employed than non-Hispanics and non-immigrants. However, as the interaction term indicates, Hispanic immigrants are significantly more likely to be employed than Hispanics born in the U.S. and more likely to be employed than non-Hispanic immigrants. This likely reflects the fact that most Hispanic immigrants enter the U.S. seeking employment. It also indicates the importance of distinguishing between Hispanic immigrants and other immigrants in assessing employment probabilities and labor force commitment.

The employment results for unmarried adults are somewhat different than those for married adults. Unlike their married counterparts, the health status of unmarried adults does not affect employment status. English competency also does not affect employment status. This could be because adults in married households may be in situations where one adult may have a great competency in English and participate while the spouse elects to remain out of employment. Similarly, when one adult in a married couple has health limitations, the other adult may assume more of the financial responsibilities. This option is less likely to exist for unmarried adults.

However, one major consistency between the two samples is that Hispanic immigrants are significantly more likely to be employed than non-Hispanic immigrants and then U.S. born Hispanics. The importance of the significance levels on the ethnicity, immigration and interaction term should not be overlooked. Clearly there are significant differences between Hispanic adults born in the U.S. versus those born outside of the U.S. in terms of labor force attachment.

In the second stage of the model (columns 3 and 4 in Table 4 and Table 5) we estimate the likelihood that an employed worker is able to access an offer of employer sponsored health insurance. Given the higher administrative costs faced by small firms, we expect that the smaller the firm, the lower the probability that a worker will receive an offer of employer sponsored health insurance. Because firms often limit coverage to full-time workers and those with a minimum amount of tenure with the firm, we also predict that part-time workers and those who have been with the firm for less than one year are less likely to have access to employer sponsored health insurance than full-time workers and those with longer tenure. We also expect that workers in the retail trade/services/housekeeping industries which employ a large proportion of low-wage, high turnover workers to be less likely to receive an offer of employer sponsored health insurance than workers in other industries.

We anticipate that workers covered by a union contract are more likely to be offered employer sponsored health insurance since this is often the subject of collective bargaining. Because the cost of covering a worker with employer sponsored health insurance is the same across workers with different earnings, employer sponsored health insurance represents a larger share of the compensation for low-wage workers than for high wage workers. Firms employing a disproportionate share of low-wage workers may thus be less likely to offer employer sponsored insurance than firms that employ proportionately more high-wage workers.

As expected, part-time workers and those who have been with the firm for less than one year are less likely to receive offers than full-time workers and those with longer tenure. This is true regardless of the sample used for analysis.

Workers employed by small firms are significantly less likely than those employed by large firms to receive offers of employer sponsored health insurance. The coefficient on the firm size less than 10 dummy is significant and greater in absolute value than that on firm size 10-49 suggesting that workers in firms with fewer than 10 employees are much less likely than workers in firms with more than 50 employees to receive offers of employer sponsored health insurance and also less likely to receive such offers than workers employed by firms with 10 to 49 employees.

It is interesting to note that the dummy variables for earning less than $20,000 and earning between $20,000 and $39,999 are negative and significant in both samples. The size of the coefficient on the $20,000-$39,999 is also smaller in absolute value than that on the dummy variable for earning less than $20,000. This result underscores the extreme disadvantage that low-wage workers have relative to high-wage workers in obtaining access to employer sponsored health insurance.

Workers with a less than a high school education are also significantly less likely than more highly educated workers to receive offers of employment sponsored health insurance. This relationship holds even after we control for earnings level and is consistent across both samples.

It is interesting to note that Hispanic workers are significantly more likely to be in positions that offer employer sponsored health insurance than non-Hispanic workers. In the case of all adults 18-64, we find that Hispanic workers are more likely than non-Hispanic workers to access employer sponsored health insurance but that Hispanic immigrant workers are less likely than non-Hispanic nonimmigrant workers to access employer sponsored health insurance. The magnitude of the coefficients on the Hispanic (1.283) and Hispanic x immigrant interaction term (-1.356) are nearly equal in absolute value but opposite in sign. This suggests that there are important differences between Hispanic immigrant and Hispanic non-immigrants in the ability to access jobs that offer employer sponsored health insurance. Unfortunately, our immigration variable is based on place of birth. We do not have direct information on the citizenship status of the individual worker so it is not possible to tell if access to employer sponsored insurance varies by citizenship status.

The final stage of analysis is the decision to take-up employer sponsored health insurance. This analysis is conducted on those workers who received an offer of health insurance from their employer. The results of this analysis indicate no difference in take up rates by Hispanics compared to non-Hispanics and between immigrants and non-immigrants when our sample is unmarried adults (Table 5, columns 5 and 6). However, among all California adults, we find that Hispanic immigrants are less likely to take-up coverage than non-Hispanic, non-immigrants. Again, it would be interesting to know if these findings are related to the citizenship status of the individual. Another interesting finding from the take-up equations is that those with limited English skills appear to be less likely to take-up coverage than those who speak English at home. This finding may reflect a lack of understanding among limited English speakers of the health insurance offerings of employers and/or the U.S. medical care system in general.

VI. DISCUSSION

The results of our study suggest that, all else equal, Hispanic workers are significantly more likely to be employed in positions that offer employer sponsored health insurance than non-Hispanic workers. We do find, however, that among all California adults 18-64, immigrant Hispanic workers are less likely than non-immigrant, non-Hispanic workers to be able to access employer sponsored health insurance. In addition, we find that Hispanic workers are just as likely as non-Hispanic workers to take-up employer sponsored health insurance. The exception is that, among all California adults (married and unmarried), immigrant-Hispanic workers are less willing to take-up employer offered health insurance than non-Hispanic, nonimmigrant workers. The extent to which this may reflect the higher rates of non-citizenship status among both Hispanic-immigrants than among other immigrants is not known since we did not have information on the immigration status of the workers not born in the U.S.

Table 6 shows the odds-ratios of employment, offers of employer sponsored health insurance and take-up of employer offered health insurance by ethnicity and immigration status. We find that the Hispanic immigrants have a much higher probability of being employed than their non-immigrant, non-Hispanic counterparts (odds ratios are 3.48 for all California adults and 2.71 for unmarried adults). Hispanic workers are 3.6 times more likely to be in positions that offer employer sponsored health insurance than non-Hispanic workers. However, Hispanic immigrants are only 26% as likely to be in positions that offer employer sponsored health insurance as non-Hispanic, non-immigrant workers. Among unmarried workers, Hispanic workers are more likely to have access to employer sponsored health insurance regardless of their immigration status. Only in the take-up equation for all adults do Hispanic immigrants appear to have lower take-up rates. Whether this reflects a higher perceived risk of deportation among Hispanic immigrants with families can not be ascertained due to lack of information regarding immigration status.

It does appear that the problem of limited access to employer sponsored health insurance among Hispanic workers is rooted in the low levels of education and consequent low-earnings of many Hispanic workers. Policies directed at improving health insurance coverage among this population group will be most effective if they are directed at improving human capital investment through education. Policies that fund neighborhood health facilities for the uninsured may be effective in the short run but they do not deal with the long term problem posed by inadequate human capital development and improved labor market skills. Subsidies to purchase health insurance will have little effect if they are designed to operate through increasing the tax deductibility of privately purchased premiums given the low wage levels of the uninsured. This type of policy is likely to generate a benefit to high wage-workers without having a discernable impact on increasing health insurance coverage among currently uninsured workers.

One major issued raised by this study is that there may are potentially large differences between immigrant and non-immigrant Hispanics in terms of employment, access to employer sponsored health insurance, and take-up of such insurance. How much of this may be reflected in the citizenship status of Hispanic immigrants to the U.S. is not known but is a fertile and important ground for future research.

In addition, there are a number of other future research questions raised in this study. One is on the relationship between the premiums faced for employer sponsored by workers, the quality of coverage offered under the different plans, and the decisions by the workers to take-up coverage. We did not have direct information on these factors and had to rely on variables such as firm size as proxies.

Another limitation of the study is that we did not have direct information on the hourly wage offered to each worker but only on the workers' yearly earnings. We also did not have information on other aspects of the compensation package such as pensions so we could not examine trade-offs between employer sponsored health insurance and other components of compensation. These issues provide important avenues for future research that examines compensation trade-offs as well as establishes reasonable measures of coverage adjusted premiums faced by firms. To some extent, the problem in this analysis is mitigated by confining the sample to California workers. The health insurance market is rather unique in the state in terms of its reliance on HMOs and the breadth of the safety net is also somewhat controlled since factors such as Medicaid eligibility policies are constant across all workers.

REFERENCES

California HealthCare Foundation (2005), California Employer Health Benefits Survey, 2005, www.chcf.org

Collins, Sara R. Michelle M. Doty, Karen Davis, Cathy Schoen, Alyssa L. Holmgren and Alice Ho (2004), The Affordability Crisis in U.S. Health Care: Findings from the Commonwealth Fund Biennial Health Insurance Survey, The Commonwealth Fund, Publication #723, March.

Currie, Janet and Aaron Yelowitz (1999), "Health Insurance and Less Skilled Workers," National Bureau of Economic Research, Working Paper # 7291, August.

Doty, Michelle M. and Alyssa L. Holmgren (2004), Unequal Access: Insurance Instability Among Low-Income Workers and Minorities, The Commonwealth Fund, Issue Brief #729, April.

Dube, Arindrajit, Ken Jacobs, Sarah Muller, Bob Brownstein and Phaedra Ellis-Lamkins (2005), Falling Apart: Declining Job-Based Health Coverage for Working Families in California and the United States, UC Berkeley Labor Center, Health Care Policy Brief

Dushi, Irena and Marjorie Honig (2005), "Offers or Take-up: Explaining Minorities' Lower Health Insurance Coverage," Economics Research Initiative on the Uninsured, Working Paper 40; http://www.umich.edu/eriu/pdf/wp40.pdf

Feldman, Roger., B. Bowd, S. Leitz and L.A. Blewett (1997), The Effect of Premium's on the Small Firm's Decision to Offer Health Insurance," Journal of Human Resources, 32:4 pp. 633-658.

Fronstin, Paul (2005), "Employment-Based Health Benefits: Trends in Access and Coverage, Employment Benefits Research Institute Issue Brief #284, August, 1-28.

Fronstin, Paul (2005), "The Impact of Immigration on Health Insurance Coverage in the United States," Employee Benefits Research Institute Notes, June, 26:6, 1-8.

Fronstin, Paul (2005), "Union Status and Employment-Based Health Benefits" Employee Benefits Research Institute Notes, May, 26:5, 1-5.

Glied, S., J. Lambrew, and S. Little (2003), The Growing Share of Uninsured Workers Employed by Large Firms, The Commonwealth Fund, Publication #672, October.

Hadley, Jack and A. Reschovsky, (2002), "Small Firms' Demand for Health Insurance: The Decision to Offer Insurance," Inquiry, 39:2, pp. 118-137.

Monheit, Alan and Jessica Vistnes (2000), "Race/Ethnicity and Health Insurance Status: 1987 and 1996," Medical Care Research and Review, 57: (Supplement 1): 11-35.

Phelps, C. (1976). The Demand for Health Insurance: A Theoretical and Empirical Investigation, RAND Research Paper Series No. R-105-OEO, Santa Monica: 1973

Polsky, Daniel, Rebecca Stein, Sean Nicholson and M. Kate Bundorf (2005), "Employer Health Insurance Offerings and Employee Enrollment Decisions," Health Services Research, 40:5 Supplement Part 1, 1259-1278.

Quinn, K. (2000), Working Without Benefits: The Health Insurance Crisis Confronting Hispanic Americans, The Commonwealth Fund, Publication #370, March.

Schur, Claudia, and Jacob Feldman (2001), "Running in Place: How Job Characteristics, Immigrant Status and Family Structure Keep Hispanics Uninsured," The Commonwealth Fund, Publication #453, May.

Stanton, Mark W. (2004), "Employer-Sponsored Health Insurance: Trends in Cost and Access," Research in Action, Agency for Healthcare Research and Quality, Issue #17.

Stock, James and Mark Watson (2003), Introduction to Econometrics (New York, Addison Wesley)

Waidmann, T.A., B. Garrett and Jack Hadley (2004), "Explaining Differences in Employer Sponsored Insurance Coverage by Race, Ethnicity and Immigration Status, ERIU Working Paper #42

Wooldridge, Jeffrey (2003), Introductory Econometrics: A Modern Approach (Mason, OH, Thomson Southwestern)

NOTES

(1.) The three stage process raises questions of selection bias in the second and third stages of the model since these stages are estimated only on a subset of individuals (in the second stage, only those who have selected to work are included in the analysis and in the third stage only those who have selected to work AND selected a position that offers employer provided health insurance are included in the analysis). Sample selection bias induces a correlation between one or more regressors and the error term and can lead to bias and inconsistency in an estimator (Stock and Watson 2003, Wooldridge 2003). To test for selection bias we first estimate the equation for employment and compute predicted values using the regression coefficients. We can then compute the inverse Mills ratio for each observation (ei) and conduct a t-test to determine if there is a sample selection problem.

(2.) We include parent status in the first stage, employment regressions but exclude it from the offer and take-up equations. Likewise, industry is included in the second stage regression but excluded in the third stage.

(3.) Even if we had data on premiums paid by employees, we would need to adjust them for the specific characteristics of the health plan in terms of service coverage, choice of provider, etc. However, the dominance of Kaiser in the California market coupled with the fact that it is generally the least expensive option for most employers to offer their employees (save for high deductible health plans which were a negligible share of the market in 2000) suggests that there is a general standard of benefits covered in the most frequently offered plan in California.

(4.) In California, many firms have domestic partner provisions which provide for health insurance coverage for an employee's unmarried partner. However, in many cases these provisions apply only to same sex couples and they are not uniform across all employers. The CWHS data did not have any information on domestic partner coverage.

Lynn Paringer, Professor of Economics, College of Business and Economics, California State, University, East Bay, Hayward, California 94542, lynn.paringer@csueastbay.edu. The research presented in this paper was funded by a grant from the California Program on Access to Care.

TABLE 1.
Distribution of Hispanic and Non-Hispanic California Workers by
Selected Characteristics

 Workers Age 18-64

 Non-Hispanic Hispanic
Characteristic Percent Percent

Offer of EMPLOYER SPONSORED HEALTH INSURANCE 78.7# 66.9#

Insured 87.1# 65.9#
 Insured through own employer 61.2# 47.4#
 Insured through employer of other family 14.2# 7.1#
 Insured through other private 10.9# 6.4#

WORKER CHARACTERISTICS
Part-time 16.0 17.3
Tenure of less than one year with firm 17.6 16.0
Earn less than $20,000 21.4# 47.4#
Earn $20,000-$39,999 31.9# 35.3#
Earn GE $40,000 46.7# 17.3#

FIRM CHARACTERISTICS
Firm size LT 10 21.9 23.5
Firm size 10-49 15.7# 22.8#
Firm size 50-99 6.4 9.7
Firm size GE 100
Union Coverage 24.1 22.7
Retail/trade housekeeping 12.5# 21.5#
Manufacturing 4.5# 14.1#

DEMOGRAPHIC CHARACTERISTICS
Age 18-24 10.2# 17.7#
Age 25-34 24.2# 36.7#
Age 35-44 33.3# 26.3#
Age 45-54 21.7# 14.9#
Age 55-64 10.1# 4.4#
English spoken at home 93.4# 40.5#
Born in the U.S. 84.2# 41.5#
Less than a high school education 2.7# 33.8#
Female 46.1# 34.8#
Married 45.3# 50.0#
Parent 33.2# 57.6#

ECONOMIC CHARACTERISTICS
Family income LT $20,000 7.7# 27.3#
Family income $20,000-$39,999 20.8# 35.3#
Family Income LT Federal Poverty Level 4.0# 27.8#

HEALTH CHARACTERISTICS
Excellent/Very Good 71.3# 55.1#
Good 20.0# 25.0#
Fair/Poor 7.8# 19.9#

BOLD indicates significant at the .10 or better level

Total N is approximately 1275 and varies slightly based on item
non-response.

Note: BOLD indicates significant at the .10 or better level indicated
with #.

TABLE 2.
Percent of Workers 18-64 Offered Employer
Sponsored Health Insurance by Selected
Characteristics

Characteristics % Offered HI

WORKER CHARACTERISTICS
Part-time 40.6
Full-time 82.4
Tenure LT 1 year 68.0
Tenure GE 1 year 77.7
Earn LT $20,000 47.6
Earn $20,000-$39,999 78.3
Earn GE $40,000 92.3
FIRM CHARACTERISTICS
Firm size LT 10 29.7
Firm size 10-49 77.4
Firm size 50-99 78.5
Firm size GE 100 92.2
Union coverage 95.9
Non-union coverage 68.8
Retail/trade/housekeeping 63.9
Manufacturing 83.3
Other industry 77.6

DEMOGRAPHIC CHARACTERISTICS
Male 80.2
Female 69.7
Hispanic 66.9
Non-Hispanic 78.7
Age 18-24 64.1
Age 25-34 79.4
Age 35--44 78.0
Age 45-54 72.9
Age 55-64 77.1
White 76.4
Black 76.3
Asian/Pacific Islander 86.6
Other 66.1
Born in the U.S. 78.3
Born out of the U.S. 68.1
English spoken at home 78.6
Other than English spoken at home 64.0
Married 75.5
Not married 75.7
Parent 72.9
Non-parent 77.5
Less than high school education 43.8
GE high school education 79.3

ECONOMIC CHARACTERISTICS
Poverty household 38.1
Non-poverty household 79.7

HEALTH CHARACTERISTICS
Excellent/Very Good 78.3
Good 76.0
Fair/Poor 58.9

TABLE 3.
Percent of Workers Offered Employer Sponsored Health Insurance by
Ethnicity and Selected Characteristics, Workers 18-64

Characteristics Hispanic Non-Hispanic

Less than high school education 41.8# 50.0#
GE high school education 79.2 79.4

Born in the U.S. 81.7 77.8
Born out of the U.S. 56.2# 83.5#

Earn LT $20,000 48.1 47.5
Earn $20,000-$39,999 79.4 77.9

Age 18-24 54.0# 69.2#
Age GE 25 69.0# 79.8#

English spoken at home 81.6 78.2
Other language spoken at home 56.7# 85.2#

BOLD indicates differences between Hispanic and non-Hispanic workers
are significant at the .1 level or better.

Note: BOLD indicates differences between Hispanic and non-Hispanic
workers are significant at the .1 level or better indicated with #.

TABLE 4.
Determinants of Employment, Offer of Employer Sponsored Health
Insurance and Take-Up of Insurance for California Adults 18-64

 Offered Health
 Insurance by
 Employed Employer

Independent B SE B SE
Variables (1) (2) (3) (4)

Constant 1.014# .190# 3.254# .467#

Hispanic -.426# .242# 1.283# .557#

Immigrant -.440# .038# -.211 .441

Hispanic x Immigrant 1.246# .320# -1.356# .712#

Foreign Language -.370# .219# .375 .499

Female .246 .155 .335 .314

Married .973# .197# .332 .332

Married x Female -1.833# .244# -.584 .462

Parent -.084 .145

Black -.324 .212 -.231 .446
Asian/Pacific Island .257 .229 .512 .501
Other Race .206 .219 -.360 .461
(White)

Ed LT High School -.578# .203# -1.587# .446#
High School Grad -.564# .153# .036 .347
(Ed GT High School Grad)

Age 18-24 -.581# .187# .260 .378
Age 35-44 .327# .166# -.128 .296
Age 45-54 -.051 .179 -.576 .352
Age 55-64 -1.015# .204# -.153 .459
(Age 25-34)

Ex/VG Health .539# .134# .470# .286#
Fair/Poor Health -.512# .181# .116 .421
(Good Health)

Part time -1.683# .297#

LT 1 Year Tenure -.819# .295#

Firm Size LT 10 -3.095# .289#
Firm Size 10-49 -.535# .301#
Firm Size 50-99 -.070 .464

Union Coverage 1.516# .421#

Trade/Services -.250 .292
Manufacturing .327 .440
(Other industry)

Earn LT $20,000 -2.001# .348#
Earn $20,000-$39,999 -.949# .301#
(Earn GE $40,000)

-2 Log Likelihood 1865.9 545.4
Cox & Snell R Square .133 .388
Nagelkerke R Square .188 .589
N 1678 856

 Take- Up
 Insurance
 Offer

Independent B SE
Variables (5) (6)

Constant 2.334# .402#

Hispanic .190 .519

Immigrant .506 .450

Hispanic x Immigrant -1.146# .684#

Foreign Language -.892# .464#

Female .185 .313

Married .096 .307

Married x Female -.895# .444#

Parent

Black -.446 .420
Asian/Pacific Island -.011 .443
Other Race .233 .477
(White)

Ed LT High School 1.394# .049#
High School Grad -.067 .320
(Ed GT High School Grad)

Age 18-24 -.242 .390
Age 35-44 -.194 .289
Age 45-54 -.062 .349
Age 55-64 -.678 .403
(Age 25-34)

Ex/VG Health .028 .269
Fair/Poor Health -.422 .438
(Good Health)

Part time -1.373# .345#

LT 1 Year Tenure -1.206# .288#

Firm Size LT 10 -1.317# .338#
Firm Size 10-49 -.812# .266#
Firm Size 50-99 -.159 .422

Union Coverage .795# .281#

Trade/Services
Manufacturing
(Other industry)

Earn LT $20,000 -.493 .350
Earn $20,000-$39,999 .133 .265
(Earn GE $40,000)

-2 Log Likelihood 577.6
Cox & Snell R Square .153
Nagelkerke R Square .248
N 663

BOLD indicates coefficient is significant at the .10 or better level

Note: BOLD indicates coefficient is significant at the .10 or better
level indicated with #.

TABLE 5.
Determinants of Employment, Offer of Employer Sponsored Insurance and
Take-Up of Insurance for Unmarried California Adults Aged 18-64

 Offer of Health
 Employment Insurance

Independent B SE B SE
Variables (1) (2) (3) (4)

Constant 1.215# .245# 4.273# .706#

Hispanic -.583# .324# 1.759# .860#

Immigrant -.569# .268# -.286 .678

Hispanic x Immigrant 1.994# .420# -.123 1.067

Foreign Language -.191 .280 -.189 .735

Female .257 .159 .549 .353

Parent .030 .219

Black -.512# .254# -.618 .539
Asian/Pacific Island .088 .294 .796 .727
Other Race .372 .307 -.521 .771
(White)

Ed LT High School -.941# .253# -2.588# .657#
High School Grad -.803# .203# .084 .493
(Ed GT High School Grad)

Age 18-24 -.677# .223# .989# .499#
Age 35-44 .193 .246 .755 .490
Age 45-54 -.411# .248# -.253 .552
Age55-64 -.963# .289# .075 .735
(Age 25-34)

Ex/VG Health .624 .181 .045 .435
Fair/Poor Health -.442 .234 -.277 .589
(Good Health)

Part time -1.945# .453#

LT 1 Year Tenure -.963# .422#

Firm Size LT 10 -3.900# .494#
Firm Size 10-49 -.809# .440#
Firm Size 50-99 .010 .674
(Firm Size GE 100)

Union Coverage .526 .522

Trade/Services -.854# .436#
Manufacturing -.494 .649
(Other industry)

Earn LT $20,000 -2.887# .568#
Earn $20,000-$39,999 -1.418# .488#
(Earn GE $40,000)

-2 Log Likelihood 1035.0 262.3
Cox & Snell R Square .123 .419
Nagelkerke R Square .174 .637
N 839 411

 Take-up
 Insurance
 Offer

Independent B SE
Variables (5) (6)

Constant 2.449# .563#

Hispanic .209 .802

Immigrant .032 .698

Hispanic x Immigrant -.115 1.014

Foreign Language -2.17# .690#

Female .138 .352

Parent

Black -.433 .560
Asian/Pacific Island .803 .739
Other Race .584 .744
(White)

Ed LT High School .310 1.152
High School Grad -.057 .503
(Ed GT High School Grad)

Age 18-24 .198 .512
Age 35-44 .615 .487
Age 45-54 .476 .572
Age55-64 -.401 .669
(Age 25-34)

Ex/VG Health -.368 .434
Fair/Poor Health -1.639# .658#
(Good Health)

Part time -1.985# .543#

LT 1 Year Tenure -1.154# .444#

Firm Size LT 10 -1.097# .531#
Firm Size 10-49 -1.090# .404#
Firm Size 50-99 2.052# 1.051#
(Firm Size GE 100)

Union Coverage 1.309# .461#

Trade/Services
Manufacturing
(Other industry)

Earn LT $20,000 -.496 .529
Earn $20,000-$39,999 .024 .416
(Earn GE $40,000)

-2 Log Likelihood 265.5
Cox & Snell R Square .227
Nagelkerke R Square .372
N 319

BOLD Indicates coefficient is significant at the .10 or better level

Note: BOLD Indicates coefficient is significant at the .10 or better
level indicated with #.

TABLE 6.
Odds-Ratios of Employment, Offers of Employer Sponsored Health
Insurance and Take-Up of Insurance Offers

 California Adults 18-64 Years

 Offer of Take-up of
 Health Insurance
Variable Employment Insurance Offer

Hispanic .65# 3.61# 1.21
Immigrant .64# .63 1.66
Hispanic x Immigrant 3.48# .26# .32

 California Unmarried Adults 18-64 Years

Hispanic .56# 5.81# 1.23
Immigrant .57# .75 1.03
Hispanic x Immigrant 2.71# .88 .89

BOLD indicates variable is significant at the .10 level or better

Note: BOLD indicates variable is significant at the .10 level or
better indicated with #.