Substantial Churn In Health Insurance Offerings By Small Employers, 2014–15

Estimates from the Medical Expenditure Panel Survey–Insurance Component (MEPS-IC), a repeated cross-sectional survey of US establishments, show that from 2014 to 2015 the percentage of small, private-sector employers (those with fewer than fifty employees) offering health insurance declined by 2.8 percentage points. ^1,2 This estimate shows the net effect of the year-to-year inflow and outflow of establishments offering insurance, but it does not provide information on continuity of coverage.

To evaluate the dynamics of employers’ health insurance decisions, which have generally been studied using longitudinal household-level ^3–6 rather than employer-level data, ⁷ we accessed a new data source, the 2014–15 MEPS-IC longitudinal survey. We found that 14.6 percent of private-sector employers with fifty or fewer workers that had offered insurance in 2014 did not offer it in 2015 ( Exhibit 1 ). ⁸ Offsetting some of these losses were gains in offers among small-firm establishments that did not offer insurance in 2014 (5.5 percent). The smallest employers (fewer than ten employees) had the highest rate of dropping coverage (18.8 percent). We did not find a significant net change from 2014 to 2015 in offer rates for establishments with fifty or fewer workers, using the longitudinal survey. ⁹

SOURCE Authors’ analysis of data for 2014–15 from the Agency for Healthcare Research and Quality, Medical Expenditure Panel Survey—Insurance Component (MEPS-IC) longitudinal survey. NOTES “Coverage” refers to offering at least one plan that included hospital and medical coverage. “Firm size” refers to the number of employees in the firm of which the establishment is a part. ^a Relative standard error is greater than 30 percent.

In addition to examining changes in offer rates, we examined within-establishment changes for employers that offered coverage in both years, including changes in the percentages of workers eligible for and enrolled in coverage.

Study Data And Methods

While cross-sectional data provide information on net changes in employers’ offer, eligibility, take-up, and coverage rates across establishments, it is also important to understand the dynamics of employer behavior. One approach to collecting such data is to ask employers retrospective questions. Beginning in 2014 the annual MEPS-IC survey included a question regarding employers’ offers of coverage in the previous year, which was used in a recent analysis of employer offers of coverage. ⁷

An alternative approach is to follow employers over time. The Agency for Healthcare Research and Quality and the Census Bureau, with sponsorship from the Office of the Assistant Secretary for Policy and Evaluation in the Department of Health and Human Services, fielded a series of longitudinal MEPS-IC surveys that selected a sample of private-sector employers from the annual MEPS-IC and reinterviewed them the following year. The 2013–14 longitudinal sample included employers with fifty or fewer employees, the 2014–15 sample included employers with 100 or fewer employees, and the 2015–16 survey included employers of all sizes.

In this study we used data from the 2014–15 MEPS-IC longitudinal survey, the most recent available, which included 3,763 establishments. We classified establishments into categories based on their firm size in the first year of the survey. ^10,11 We applied longitudinal weights to produce nationally representative estimates for private-sector establishments. We also estimated linear regression models of the likelihood that employers dropped coverage. Unless otherwise indicated, all differences discussed in the text are significant ( $p<0.05$ , or better).

A limitation of this study was that because we observed employers’ behavior only in the presence of the Affordable Care Act (ACA), we could not investigate whether the ACA affected these coverage transitions. Future MEPS-IC longitudinal survey data for employers with 51–100 employees will make it possible to compare the employers before and after implementation of the ACA’s employer shared-responsibility provision, which was implemented for these employers in 2016.

Study Results

Offer Rates

As noted above, 14.6 percent of small-firm establishments (with fifty or fewer employees) that offered insurance in 2014 dropped coverage in 2015 ( Exhibit 1 ). These losses were partially offset by gains in insurance among employers that did not offer coverage in 2014. The likelihood of dropping and adding coverage differed by firm size, with 18.8 percent of offering establishments in the smallest firms (fewer than ten workers) dropping coverage in 2015, compared to 8.0 percent and 5.4 percent of establishments in firms with 10–24 and 25–50 employees, respectively. In contrast, in medium-size firms (51–100 workers), 27.1 percent of nonoffering establishments added coverage, while 3.0 percent of offering establishments dropped coverage.

The likelihoods described above were calculated based on subsamples of employers that either did or did not offer coverage in 2014. The longitudinal data allowed us to translate these likelihoods into year-to-year average net changes. When we applied the relatively large loss of offers among small employers that offered coverage in 2014 (14.6 percent) to the relatively small share of such employers that offered coverage (34.8 percent), the percentage of all small employers that changed from offering to not offering coverage equaled 5.1 percent (14.6 percent of 34.8 percent) ( Exhibit 2 ). This change was offset by the 3.6 percent increase (5.5 percent of 65.2 percent) of all small employers that added offers of coverage. The net effect was an insignificant decline in the offer rate of 1.5 percentage points.

SOURCE Authors’ analysis of data for 2014–15 from the Agency for Healthcare Research and Quality, Medical Expenditure Panel Survey–Insurance Component (MEPS-IC) longitudinal survey. NOTE “Coverage” refers to offering at least one plan that includes hospital and medical coverage.

Multivariate Analysis

To investigate the factors that might have affected the likelihood of dropping coverage, we estimated linear regression models using pooled data for small-firm establishments from the first two MEPS-IC longitudinal surveys. Details and additional results are available in the online Appendix. ¹² We found that an increase in the concentration of low-wage workers was associated with an increased likelihood that an employer would drop coverage, but this relationship depended on the presence of high-wage workers. Small-firm establishments with a majority of low-wage workers and no high-wage workers were 12.2 percentage points more likely to drop coverage, compared to employers with no low-wage workers ( Exhibit 3 ). However, majority low-wage establishments with some high-wage workers were not significantly more likely to drop coverage. Establishments in firms with fewer than 10 workers were more likely to drop coverage than those with 25–50 employees, and nonprofit employers were less likely than for-profit ones to drop coverage (data not shown).

SOURCE Authors’ analysis of data for 2013–14 and 2014–15 from the Agency for Healthcare Research and Quality, Medical Expenditure Panel Survey–Insurance Component (MEPS-IC) longitudinal survey. NOTES The omitted category is establishments with no low-wage workers; our findings represent comparisons with each wage category and no low-wage workers. “High-wage workers” are those who earn more than $27.00 per hour (in the 2013–14 survey) or more than $27.50 per hour (in the 2014–15 survey). “Low-wage workers” are those who earn less than $11.50 per hour in either survey year. Details about the multivariate analysis are in the Appendix (see Note  12 in text). “Small establishments” are those with 50 or fewer employees. ** $p<0.05$

Eligibility Rates

Cross-sectional MEPS-IC eligibility rates for small-firm employees have been remarkably stable since 1999, hovering between roughly 78 and 79 percent. ² However, the MEPS-IC longitudinal survey data reveal that underlying this stability is a substantial amount of churn. While roughly half of small-firm establishments had unchanged eligibility rates (number eligible divided by number employed), the other half were evenly split between those with increasing and decreasing rates ( Exhibit 4 ). In contrast, more than 40 percent of establishments in firms with 51–100 workers had increasing eligibility rates, while only 25 percent had decreasing rates. A substantial share of establishments experienced year-to-year changes in their take-up rates (number enrolled divided by number eligible) and coverage rates (number enrolled divided by number employed).

One possible explanation for the churn in these measures is measurement error. To investigate this possibility, we estimated a series of regressions for the likelihood that eligibility rates increased, decreased, or remained the same (details on the model specifications are provided in the Appendix). ¹² Our models provide some evidence to support the conclusion that the observed changes in eligibility rates are not due to measurement error, as changes in eligibility were systematically related to the year-to-year changes in the percentages of employees who worked part time and in the minimum number of hours employees were required to work to be eligible for insurance (for more details on the results, see the Appendix). ¹²

Discussion

We used data for 2014–15 from the MEPS-IC longitudinal survey to examine within-establishment transitions in offers of employer-sponsored insurance. We found an insignificant net decline of 1.5 percent in offers by small employers (those with fifty or fewer employees), as 5.1 percent of all small employers dropped offers and 3.6 percent added offers between 2014 and 2015. However, this result masks the more dynamic nature of offers within individual establishments. For example, we found that 14.6 percent of all small employers that offered coverage in 2014 dropped coverage in 2015, while 18.8 percent of the smallest employers (those with fewer than ten employees) did so.

In a similar analysis, using retrospective MEPS-IC data, Jean Abraham and coauthors found that 3.5 percent of all establishments dropped offers and 1.1 percent added offers. ⁷ Because we had access to both retrospective and longitudinal information in MEPS-IC longitudinal survey data, we compared estimates using the two types of information and found evidence that Abraham and coauthors’ calculations might have somewhat understated the amount of churn among all employers, relative to rates from longitudinal information—but not to a degree that would have qualitatively altered their conclusion that offers were “largely stable in 2014.” ⁷ The primary reasons that we reached a different conclusion about the degree of churn in offers were that we focused on smaller employers and that we examined the likelihood that an offering employer would drop, and a nonoffering employer would add, an offer of insurance.

In our regression models, we found that an increase in the concentration of low-wage workers was associated with an increased likelihood of dropping coverage, but we also found that the presence of higher-wage workers had a protective effect. If a majority low-wage establishment also employed high-wage workers, it was no more likely to drop its offer of insurance than was an employer with no low-wage workers. It is likely that the overall association between low-wage employees and the likelihood of dropping offers has been ongoing for many years: Estimates from the annual MEPS-IC indicate that offer rates at small, majority low-wage employers fell from 31.4 percent in 2000 to 13.4 percent in 2015. ²

Among employers offering coverage in both 2014 and 2015, we found substantial turbulence in eligibility and coverage rates. While it is not possible to determine from our data the degree to which these changes occurred because of factors related to employers’ decisions, our regression results indicated that at least some of the variation in eligibility rates was due to changes in the minimum number of work hours that employers required before employees were eligible for health insurance—a factor that is under the employer’s control.

Conclusion

Discontinuities in insurance coverage generate challenges for employers, individuals, and policy makers. Our results demonstrate that the extent of discontinuities in coverage offered by small and medium-size employers is greater than would be suggested by examining cross-sectional trends.

ACKNOWLEDGMENTS

The research presented in this article was conducted while the authors were also research associates at the Census Bureau’s Center for Economic Studies. The views expressed in this article are those of the authors, and no official endorsement by the Agency for Healthcare Research and Quality or the Department of Health and Human Services is intended or should be inferred, nor do they necessarily indicate concurrence by the Census Bureau. The results presented in this article have been screened to ensure that no confidential information has been revealed.

NOTES

• 1 An establishment is defined as a business unit in a single location; a firm might consist of one or more establishments. In this article we use the terms employer and establishment interchangeably. Note that MEPS-IC is an establishment survey.
• 2 Agency for Healthcare Research and Quality . MEPS Expenditure Panel Survey: summary data tables [Internet]. Rockville (MD) : AHRQ ; [cited 2017 Jul 24 ]. Available from: https://meps.ahrq.gov/mepsweb/data_stats/quick_tables.jsp Google Scholar
• 3 Carman KG , Eibner C , Paddock SM . Trends in health insurance enrollment, 2013–15 . Health Aff (Millwood) . 2015 ; 34 ( 6 ): 1044 – 8 . Go to the article, Google Scholar
• 4 Graves JA , Mishra P . The evolving dynamics of employer-sponsored health insurance: implications for workers, employers, and the Affordable Care Act . Milbank Q . 2016 ; 94 ( 4 ): 736 – 67 . Crossref, Medline, Google Scholar
• 5 Graves JA , Nikpay SS . The changing dynamics of US health insurance and implications for the future of the Affordable Care Act . Health Aff (Millwood) . 2017 ; 36 ( 2 ): 297 – 305 . Go to the article, Google Scholar
• 6 Abramovitz J , O’Hara B . New estimates of offer and take-up of employer-sponsored insurance . Med Care Res Rev . 2016 Jun 18 . [Epub ahead of print]. Google Scholar
• 7 Abraham J , Royalty AB , Drake C . Employer-sponsored insurance offers: largely stable in 2014 following ACA implementation . Health Aff (Millwood) . 2016 ; 35 ( 11 ): 2133 – 7 . Go to the article, Google Scholar
• 8 In the MEPS-IC longitudinal survey, unlike MEPS-IC, establishments in firms with fifty employees were categorized as small firms. This is because of the sample design of the MEPS-IC longitudinal survey, which used the cutoff in firm size in the Small Business Health Options Program to stratify the sample.
• 9 The difference between the MEPS-IC estimate (a decline of 2.8 percentage points) and the MEPS-IC longitudinal survey estimated decline of 1.5 percentage points was not significant. Note that the MEPS-IC longitudinal survey has a smaller sample than MEPS-IC does and includes only establishments that remained in business for the two-year period.
• 10 Since MEPS-IC does not contain information on full-time-equivalent employees, all of the discussion of firm size refers to the total number of employees.
• 11 Almost all establishments (99.4 percent) in firms with 1–50 workers in the first year of the survey remained in the same size category in the second year. The rate was 77.9 percent for the category of 51–100 workers.
• 12 To access the Appendix, click on the Appendix link in the box to the right of the article online.