Stephen H. Long M. Susn Mrquis Low-Wge Workers nd Helth Insurnce Coverge: Cn Policymkers Trget Them through Their Employers? Mny policy inititives to increse helth insurnce coverge would subsidize employers to offer coverge or subsidize employees to prticipte in their employers helth plns. Using dt from the 1997 Robert Wood Johnson Foundtion Employer Helth Insurnce Survey, we contrst low-wge employers with ll other employers. Employees in low-wge es hve significntly worse ccess to employmentbsed insurnce thn other employees do; they re less likely to work for n employer tht offers insurnce, less likely to be eligible if working in tht offers insurnce, nd less likely to be enrolled if eligible. employers contribute lower shres of premiums nd offer less generous benefits thn other employers do. Policies tht would trget subsidies to selected employers to increse insurnce offers to low-wge workers re difficult to design, however, becuse severl commonly mentioned employer chrcteristics (including firm size) re found to be poor indictors of low-wge worker concentrtion. Progrms tht would set minimum stndrds for employer plns to be eligible for buy-ins need to bse these stndrds on the less generous terms offered by low-wge employers in order to effectively rech low-wge workers nd their dependents. Policymkers re ctively serching for wys to reduce the lrge number of uninsured Americns. It is well known tht most of the uninsured re workers or their dependents. In 1997, 84% of the uninsured lived in fmilies heded by workers (Fronstin 1998). Workers in smll es nd those with low ernings re more likely to be uninsured thn other workers. Nerly one-hlf of ll uninsured workers were either self-employed or working in privte-sector firms with fewer thn 25 employees in 1997 (Fronstin 1998). In the sme yer, 31% of workers with nnul individul ernings under $10,000 were uninsured, compred to 5% of those with nnul individul ernings of $40,000 or more (Fronstin 1998). Consequently, discussions of the working uninsured often focus on smll es low rtes of offering insurnce; on restrictive rules for coverge eligibility, especilly for prt-time workers; nd on low-wge employees filure to enroll in the plns tht re offered to them (McLughlin nd Zellers 1992; Zellers, McLughlin, nd Frick 1992; Cooper nd Schone 1997; Thorpe nd Florence 1999). Nerly ll of the current proposls to reduce the rnks of the uninsured re incrementl nd voluntry. 1 They would build on the existing mixed publicprivte system of coverge by offering direct subsidies or tx credits to encourge low-income individuls to enroll in helth plns or encourge employers Stephen H. Long, Ph.D., nd M. Susn Mrquis, Ph.D., re senior economists t RAND. This reserch ws supported by the Robert Wood Johnson Foundtion (grnts 028561, 031565, nd 039498). Address correspondence to the uthors t RAND, 1200 South Hyes St., Arlington, VA 22202. Inquiry 38: 331 337 (Fll 2001). 2001 Blue Cross nd Blue Shield Assocition nd Blue Cross nd Blue Shield of the Rochester Are. 0046-9580/01/3803 0331$1.25 331
Inquiry/Volume 38, Fll 2001 to begin offering them. Employer subsidies generlly would be limited to smll es, nd sometimes to the subset tht hs disproportiontely low-wge employees. An importnt design issue for cost-effective policies is deciding how to trget limited public funds for ssistnce to best rech employers who do not offer insurnce or employees who do not enroll for coverge tht is offered to them, while voiding subsidizing or chnging the behvior of those who currently do. Proposls designed for smll es often re viewed s policies trgeted to low-income workers, becuse wge rtes re bout 30% lower in smll firms thn lrge ones (Gbel, Ginsburg, nd Hunt 1997). But, of course, trgeting smll es would miss some low-income workers nd benefit some high-income workers. Under the Stte Children s Helth Insurnce Progrm (SCHIP), the federl government nd mny sttes re struggling with rules under which the progrm pys the employee shre of premiums to buy low-income children into their prents employer plns (Rosenbum et l. 1998). One issue involves setting minimum shre of the employer s premium contribution. On the one hnd, the minimum should be set high enough to discourge employers from reducing their shres, thereby shifting costs from employers to the government. On the other hnd, the minimum must be set low enough to permit most low-income children with employer-sponsored coverge to qulify for the buy-ins. Another issue is setting minimum cturil equivlency stndrds for employer plns to qulify for the buy-in. Here the trde-off is between ssuring n dequte qulity of insurnce coverge for subsidized workers nd their dependents, yet setting the stndrd low enough to encompss the kind of coverge most low-wge workers re offered by their employers. Finlly, s with ll public helth insurnce progrms, low rtes of prticiption hve been problem in implementing SCHIP. Some sttes re considering trgeting their outrech efforts to low-wge employers in order to get informtion to eligible prents nd their children. The purpose of this pper is to bring recent employer helth insurnce survey dt to ber on these difficult empiricl issues tht hve plgued policymking for incresing ccess to helth insurnce for the working uninsured. We investigte differences mong low-wge es nd other es in helth insurnce offer rtes, in the eligibility nd enrollment of their workers, nd in chrcteristics of the benefit pckges. Methods nd Dt Smple Our dt re from the 1997 Robert Wood Johnson Foundtion (RWJF) Employer Helth Insurnce Survey, which collected informtion from ntionl smple of 21,545 privte estblishments. 2 The smple ws drwn from the Dun s Mrket Identifiers ntionl census of employment estblishments. We excluded from our smple self-employed people who hve no employees. Estblishments in the 60 communities followed by the Community Trcking Study nd in 12 sttes with significnt smll-group insurnce reforms were oversmpled (Kemper et l. 1996; Reserch Tringle Institute 1998). Within geogrphic res, the smples were llocted to strt defined by the number of workers t the estblishment. Dt were collected using computer-ssisted telephone interviewing system (CATI) to contct nd interview the person (or persons) most knowledgeble bout helth benefits nd bout chrcteristics of the estblishment nd ll its workers. The response rte to the survey ws 60%. This flls within the rnge of response rtes (50% to 66%) reported for two other lrge-scle surveys of employer helth insurnce dministered in the sme period (Gbel, Ginsburg, nd Hunt 1997; Brnscome et l. 2000). Response rtes vried slightly by size of estblishment, from 54% for estblishments with four or fewer workers to 61% for estblishments with five to 49 workers. Reponse rtes lso vried cross geogrphic strt, rnging from 44% in the Nssu-Suffolk, New York, Metropolitn Sttisticl Are to 86% in group of rurl counties in North Crolin. Smple design weights were post-strtified using informtion from the Bureu of Lbor Sttistics on the number of privte employers by county nd employer size to djust for differentil response rtes. The survey collected informtion bout the totl employment t ech plce of nd the distribution of key chrcteristics of its workers. It lso collected detiled informtion bout every helth insurnce pln offered to employees, s well s the number of employees enrolled in ech pln nd the overll number of employees eligible for helth insurnce. We used these dt nd counts to develop smple weights to mke estimtes for severl different units of observtion: ll privte employees, employees eligible to enroll in employer plns, nd employees electing to enroll in employer plns. 332
Low-Wge Workers Concepts nd Definitions We define low-wge worker to be n employee erning less thn $7 per hour in 1997, or less thn $14,000 per yer. estblishments re those in which t lest two-thirds of the employees re lowwge workers. We lso considered definition of low-wge estblishments tht required t lest threequrters of employees to be low-wge workers. The results from the two definitions were very similr so we report only those from the less restrictive definition. We mesure employer contribution rtes for single nd fmily coverge. The survey collected informtion bout employer contributions for ech helth insurnce pln offered. 3 Some employers offer coverge for specil services, such s dentl cre or prescription drugs, in plns tht re seprte from the generl medicl pln. We constructed employer contributions for the pckge of benefits held by enrollees by dding together the contributions pid both for the generl medicl pln nd for ny plns covering specil services. For employers tht offer more thn one benefit pckge, we constructed n verge contribution rte by weighting contribution rtes for the different pckges by the number of employees electing the pckge. We lso report the cturil vlue of the benefits, which is mesured s the expected pln benefit pyments reltive to the expected totl helth expenditures for stndrdized popultion. Acturil vlues for multiple plns offered by single employer re verged using enrollment shres s weights. Results If policymkers were to use the workplce to effectively trget subsidies to uninsured low-wge workers, they would wnt to identify es tht employ lrge shre of low-wge employees nd smll number of non-low-wge employees. Among these es, they lso would wnt to identify those tht do not offer insurnce or hve low employee prticiption. Proposls often use esily observed proxy or criterion mesures to identify employers of low-wge workers in order to reduce the dministrtive costs of directly mesuring wges. The objective is to identify es using criterion mesures tht re sensitive (identify significnt shre of low-wge workers) nd specific (identify workplces with primrily low-wge workers) in trgeting the low-wge worker popultion. A substntil shre of low-wge workers Tble 1. Distribution of low-wge workers by chrcteristics nd percentge of workers who re low wge in these es, 1997 Percentge of low-wge workers Percentge of workers who re low wge All es 100 21 Type of Low wge Firm size Fewer thn 10 workers 10 49 workers 50 or more workers Industry Retil trde Service Age of 5 yers or fewer More thn 5 yers Sesonl Yes b No 58 42 19 20 61 38 31 31 12 88 12 88 87 10 29 25 19 46 21 13 30 21 27 21 Note: Dt from 1997 Robert Wood Johnson Foundtion Employer Helth Insurnce Survey. At lest two-thirds of workers ern less thn $7 per hour or $14,000 per yer. b At lest 25% of workers re temporry or sesonl. is concentrted in workplces tht employ predomintely other low-wge workers. Fifty-eight percent of low-wge employees work in es tht we hve defined s low wge (Tble 1). These low-wge es comprise 21% of ll es (not shown), nd 87% of their workforce consists of lowwge employees. Therefore, the coverge offered workers in low-wge es describes coverge options for workers who re primrily low wge nd for the mjority of low-wge workers. However, other esily mesured chrcteristics of employers do not provide criteri tht re sensitive nd specific in trgeting low-wge workers (Tble 1). workers re employed both in lrge nd smll es; only 39% of low-wge workers re in es of fewer thn 50 workers. workers lso re found in ll industries, in new nd old es, nd in sesonl nd permnent positions. Moreover, these chrcteristics do not distinguish employers hiring primrily low-wge workers. For exmple, even in firms with fewer thn 50 workers, only 27% of those workers re low wge. 333
Inquiry/Volume 38, Fll 2001 Tble 2. Access to insurnce in low-wge nd other es, 1997 Access mesure (%) (%) Employees enrolled in their own employer s pln 31 66* Employees in es offering insurnce Eligible employees mong those in es offering insurnce Enrolled employees mong those eligible 60 67 77 87* 86* 88* Enrollees electing fmily coverge 46 52** Enrollees offered fmily coverge Enrollees electing fmily coverge, if offered Note: Dt from 1997 Robert Wood Johnson Foundtion Employer Helth Insurnce Survey. At lest two-thirds of workers ern less thn $7 per hour or $14,000 per yer. * Significnt difference, p.05. ** Significnt difference, p.10. 96 47 99 52** Employees in low-wge es re less likely to be covered by employer-sponsored insurnce thn workers in other es (Tble 2). Less thn onethird of workers in low-wge re enrolled in their own employer s helth pln, in contrst to two-thirds of workers in other es. This reflects differences in the offer of insurnce by lowwge employers, in rules bout the eligibility for workers in firms offering coverge, nd in decisions mong eligible workers to enroll. Employees in lowwge es hve lower ccess t ech stge in the process. Only 60% of workers in low-wge es work for n employer offering insurnce, in contrst to 87% of workers in other es. In ddition, workers in low-wge es confront much greter turnover in the offer of insurnce thn workers in other es. For exmple, 9% of workers in low-wge es offering insurnce in 1995 hd the offer dropped within two yers, in contrst to only 3% of workers in other es (not shown). Fewer workers in low-wge es offering insurnce re eligible to prticipte in their employer s pln thn re workers in other es (67% vs. 86%), nd fewer eligible workers in low-wge es thn in other es ctully enroll (77% vs. 88%). Among helth pln enrollees, those in lowwge es re somewht less likely to be offered the opportunity to enroll fmily members nd to select fmily coverge, lthough only the ltter difference is sttisticlly significnt. The lower rte of eligibility mong workers in low-wge es thn mong workers in other es stems in prt from differences in work ptterns. Mny employers require minimum number of hours worked per week to prticipte in employer benefit progrms; typiclly 30 hours per week in our dt. Workers in low-wge es re more likely to work prt time thn workers in other es; s Tble 3 illustrtes, much lrger shre of low-wge workers thn workers in other es re ineligible for helth benefits becuse they work fewer hours thn required by their own employer. 4 There is lso greter turnover in employment mong workers in low-wge es thn mong other workers, which cn ffect eligibility. For exmple, lmost one-qurter of workers in low-wge estblishments held their current position less thn one yer, in contrst to only 16% of workers in other es (Tble 3). Employers nd insurers often impose some witing period fter joining the Tble 3. Fctors relted to insurnce eligibility in es offering insurnce, 1997 (%) (%) Eligibility conditioned on hours worked Workers ineligible 18 6* Eligibility conditioned on tenure Workers joining within the pst yer 24 16* Workers in es with witing period of: 61 dys or more 1 60 dys No wit for coverge 55 25 20 48** 22** 30** Note: Dt from 1997 Robert Wood Johnson Foundtion Employer Helth Insurnce Survey. At lest two-thirds of workers ern less thn $7 per hour or $14,000 per yer. * Significnt difference, p.05. ** Significnt difference in the distribution of witing period, p.10. 334
Low-Wge Workers Tble 4. Employers contributions to helth insurnce premiums nd employees prticiption, 1997 Employer contribution rte Single coverge Less thn 70% of premium 70 99% of premium 100% of premium Fmily coverge Less thn 60% of premium 60 79% of premium 80% of premium or more Distribution of employees 40 36 24 36 44 19 Prticiption by employees (% of eligible workers) 17* 47* 72 80 36* 80 (% of pln enrollees) 28* 33* 39* 41 52 49 79* 89* 91* Note: Dt from 1997 Robert Wood Johnson Foundtion Employer Helth Insurnce Survey. At lest two-thirds of workers ern less thn $7 per hour or $14,000 per yer. * Significnt difference in distribution for low-wge vs. other, p.05. 43 56 55 Tble 5. Acturil vlues of helth insurnce plns covering workers in low-wge nd other es, 1997 Expected shre of costs covered Less thn 75% 75 79% 80% or more Fewer thn 50 employees 32 23 46 50 or more employees (% of enrollees) 23* 22 21* 12 56* 66 12 16 72 Note: Dt from 1997 Robert Wood Johnson Foundtion Employer Helth Insurnce Survey. At lest two-thirds of workers ern less thn $7 per hour or $14,000 per yer. * Significnt difference in distribution, low-wge versus other of sme size, p.05. before helth insurnce coverge is vilble nd the length of these witing periods vries. We find tht workers in low-wge es re more likely to fce witing periods, nd these periods re of longer durtion, thn workers in other es. Greter turnover in employment nd witing periods combine to reduce the eligibility for workers in low-wge es. Finncil incentives prtilly ccount for the lower prticiption rtes mong workers in low-wge es. As recently reported from nother employer survey (Brnscome et l. 2000), employers in lowwge es contribute smller shre towrd the totl premium for eligible workers thn other employers; they lso contribute smller shre of fmily coverge premiums for enrollees who elect to enroll their dependents (Tble 4). Not surprisingly, employer contribution policies ffect employees behvior. In both low-wge nd other es, the shre of eligible workers who enroll in the employer s pln increses s the employer contribution increses. In ddition, the shre of enrollees in both low-wge nd other es tht elects dependent coverge increses s the employer contribution to fmily coverge increses. Employer contributions do not fully ccount for lower prticiption by workers in low-wge firms, however. Even controlling for the mount of the employer contribution rte, enrollments re lower mong workers in low-wge thn mong workers in other. For exmple, 80% of workers in low-wge who re eligible to prticipte do so when employers contribute 70% to 99% of the cost; in contrst, 89% of workers in other es prticipte when fcing these sme employer contribution rtes. This suggests tht other fctors, such s fmily income, my be t work. Helth pln enrollees in low-wge es tend to hve less generous coverge thn enrollees in other es (Tble 5). The expected shre of costs pid by the helth pln is less thn 80% for more thn hlf (55%) of enrollees in plns offered by lowwge, smll es. In contrst, only 44% of enrollees in plns offered by other smll es hve benefits below this level. While expected benefits of plns offered by lrger es re more generous thn those of smll es, differences in generosity of coverge for those enrolled in plns of low-wge vs. other estblishments re still seen, lthough they do not ttin conventionl levels of sttisticl significnce. Less generous offers by smll employers my reflect the nture of coverge tht is supplied to smll rther thn the decision of smll to offer less generous plns. It lso my reflect the choice of low-wge workers to hve some, but less costly insurnce thn tht offered by employers of higher-wge workers. Or it my reflect employers 335
Inquiry/Volume 38, Fll 2001 decisions bout wht they cn fford to offer. But our dt do not let us determine the fctors tht produce this result. Discussion We explored the fesibility of trgeting ssistnce to low-wge workers by subsidizing their employers to offer helth insurnce. The good news is tht lowwge workers re quite highly clustered in workplces. However, our look t esily observed employer chrcteristics does not suggest proxies tht identify low-wge employers nd could be used to determine eligibility for subsidies. Notbly, focusing on smll es, s often proposed, will not prove fruitful. In ddition, some sttes re using employers selected by size or industry s the focus of outrech progrms to increse prents prticiption in SCHIP. This prctice suffers from the sme trgeting problem. Our results lso hve implictions for policies tht would either prtilly subsidize or fully py the employee shre of premiums for low-wge or low-income workers to buy themselves or their dependents into their employer plns. They show tht low-wge employers hve considerbly lower employer contributions nd less generous helth pln benefits thn other employers. Therefore, setting minimum stndrds for such progrms will require creful look t the prctices of low-wge employers in prticulr. If stndrds re set too high, mny low-wge workers my not qulify for the new progrm nd, hence, they will not benefit. Trgeting subsidies to the uninsured through employers is lso difficult becuse not ll low-wge workers without coverge through their own employers re low-income uninsured people. Some re covered s dependents on other workers policies; Cooper nd Schone (1997) estimte tht, in 1996, 13% of low-wge workers ($7 per hour or less) hd ccess to coverge through working fmily member. Additionlly, some low-wge workers re members of non-low-income fmilies. Unfortuntely, our employer dt do not permit us to observe workers fmily circumstnces. We close with n dmonition. It would be esy to tke our results s evidence tht low-wge es do not tret their workers s well s other es. However, es choose whether to offer benefits nd how generous these benefits should be bsed on the preferences of the kinds of workers they wish to ttrct nd retin. The lower offer rtes, the lower tke-up rtes, nd the lower cturil vlues we find my reflect the preferences of low-wge workers in trding off csh wges for benefits. To expnd helth insurnce coverge for the low-income uninsured, policymkers need to find effective progrms to trget these individuls, to help them fford coverge with minimum of disruption to current prctices, nd to muster the public s willingness to py for it. Notes Any views expressed herein re solely those of the uthors, nd no endorsement by the RWJF or RAND is intended or should be inferred. The uthors thnk Lind Andrews nd Rold Euller for their efforts in prepring the survey dt files on which this pper is bsed. 1 Mny current proposls were fetured t Jnury 2000 conference, Helth Coverge 2000, sponsored by the Robert Wood Johnson Foundtion. See Brents Group LLC (2000) for summries of the individul proposls. For further discussion of these issues, see Khn nd Pollck (2001). 2 The study lso collected dt from federl, stte, nd locl governments. However, they hve been excluded from this nlysis. 3 Some employers reported their contribution s dollr mount, others s percentge of the totl premium. The survey lso collected informtion on the totl premium, nd for results presented here, we hve converted ll contributions to shre of the totl premium. 4 Some of these my voluntrily choose prt-time work wheres some my be restricted to prt-time positions by employers seeking to limit their obligtion to provide insurnce. Our dt do not permit us to distinguish between these groups. References Brents Group LLC. 2000. Helth Coverge 2000: Meeting the Chllenge of the Uninsured; A Summriztion of Alterntive Proposls. Wshington, D.C.: Brents Group LLC. Brnscome, J.M., P.F. Cooper, J. Sommers, nd J.P. Vistnes. 2000. Privte Employer-Sponsored Helth Insurnce: New Estimtes by Stte. Helth Affirs 19:139 147. Cooper, P.F., nd B.S Schone. 1997. More Offers, Fewer Tkers for Employment-Bsed Helth Insurnce: 1987 nd 1996. Helth Affirs 16:142 149. Fronstin, P. 1998. Sources of Helth Insurnce Coverge nd Chrcteristics of the Uninsured: Anlysis of the Mrch 1998 Current Popultion Survey. EBRI Issue Brief no. 204. Wshington, D.C.: Employee Benefit Reserch Institute. Gbel, J.R., P.B. Ginsburg, nd K.A. Hunt. 1997. Smll 336
Low-Wge Workers Employers nd Their Helth Benefits, 1988 1996: An Awkwrd Adolescence. Helth Affirs 16:103 110. Khn, C.N. III, nd R.F. Pollck. 2001. Building Consensus for Expnding Helth Coverge. Helth Affirs 20: 40 48. Kemper, P., D. Blumenthl, J.M. Corrign, et l. 1996. The Design of the Community Trcking Study: A Longitudinl Study of Helth System Chnge nd Its Effects on People. Inquiry 33:195 206. McLughlin, C.G., nd W.K. Zellers. 1992. The Shortcomings of Voluntrism in the Smll-Group Insurnce Mrket. Helth Affirs 11:28 40. Reserch Tringle Institute. 1998. 1997 Employer Helth Insurnce Survey: Finl Methodology Report. Reserch Tringle Prk, N.C.: Reserch Tringle Institute. Rosenbum, S., K. Johnson, C. Sonosky, A. Mrkus, nd C. DeGrw. 1998. The Children s Hour: The Stte Children s Helth Insurnce Progrm. Helth Affirs 17:75 89. Thorpe, K.E., nd C.S. Florence. 1999. Why Are Workers Uninsured? Employer-Sponsored Helth Insurnce in 1997. Helth Affirs 18:213 218. Zellers, W.K., C.G. McLughlin, nd K.D. Frick. 1992. Smll Business Helth Insurnce: Only the Helthy Need Apply. Helth Affirs 10:174 180. 337