Tax Subsidies for Expanding Health Insurance Coverage: Selected Policy Issues for the 108th Congress
CRS Report for Congress
Received through t he CRS W e b
Tax Subsidies f or Expanding Health
Insurance Coverage: Selected Policy
th
Issues for t he 108 Congress
Updated August 5 , 2003
GaryGuenther
Analys t in Business Ta xa tion and Finance
Go ve rnment and Finance Division
Congressional Research Service ˜ The Library of Congress
Tax S ubsidies for Expanding Health Insurance
Coverage: Selected Policy Issues
for the 108th Congress
Summary
An issue o f cont i n u i n g c o n cern t o C ongress is the number o f Americans who
lack health insurance coverage. In 2001, an estimated 41.2 million Americans were
uninsured for t he entire year, 99% of whom were under age 65. The number o f non-
elderly uninsured rose each year from 1987 to 1998 and t hen d eclined in both 1999
and 2000, before increasing again in 2001.
In response t o t he increase i n t he uninsured population, policym akers at t he state
and federal l evel s of governm ent have been searchi n g for effect i v e, affordabl e, and
politically feasible ways to e x pand access to h ealth insurance coverage. W hile a
variety o f p roposed solutions have been ex amined and d ebate d , recent C ongressesth
have shown a growing i nterest i n t ax -based approaches. In t he 107 Congress, the
Trade Act of 2002 established a refundabl e and advanceab l e t a x c r e dit for the
purchase o f qual i fi ed h eal t h i n surance b y i ndi vi dual s who l ose t hei r j obs because of
foreign t rade. A number o f p roposals t o create t ax subsidies f o r t h e ex p ansion ofth
heal t h i n surance coverage have surfaced i n t h e 108 Congress, including measures
to create t wo new t ax -preferred h ealth savings accounts (H.R. 2596 as incorporated
i n t o H.R . 1) and t ax credi t s f o r s m al l em pl oyers t hat p rovide health insurance
coverage for eligible employees (H.R. 450, S. 10, S. 53, S. 86, and S . 414).
This report s u m m a r i z e s what i s known abut the factors s haping the cost-
effect i v eness o f t ax subsi d i es for ex pandi ng health insurance coverage. In doing so,
i t revi ews t he pri n ci pal fi ndi ngs o f recent s t udi es assessi ng t h e cost -effect i v eness o f
a v ariety of proposed subsidies, many of which h ave b een cons i d e r e d i n recent
Congresses. The report will be updated o r revised to reflect important legi slative
activity, o r t o i ncorporate s ignificant n ew research findings o n t he use o f t ax policy
to improve health insurance coverage.
Tax policy can influence t he demand for health insurance by altering its after-tax
cost and t erm s of coverage. If t he pri n ci pal ai m of pol i cym akers i s t o ex p and h eal t h
insurance c o v e r age at a politically acceptable cost t hrough t he use o f n ew tax
subsidies, then certain fact ors would be critical in designing s uch s ubsidies. One i s
the t yp e o f s ubsidy b eing offered. Tax d eductions are m ore v aluable t o i ndividuals
i n hi gh er t ax b racket s t han t hose i n l ower t ax b racket s , b u t t h e v ast m aj ori t y of
uninsured households fall in the l ower brackets. Moreover, non-refun d a b l e t a x
credits for t he purchase o f h ealth insurance m ay have little impact because nearly half
of uninsured households have no federal i ncome t ax liability. A nother factor to
consider is who would b e e l i g i b le for t he tax s ubsidy. The cost p er newly i nsured
appears t o d epend critically on how narrowly a subsidy i s t ar ge t e d. Li kely targets
incl ude low-wage firms, low-inco me workers whose employers do not offer h ealth
insurance, and all individuals wh o are ine ligible for public or employer-provided
insurance. Other important factors s haping the efficacy of proposed tax i ncentives
to ex pand heal t h insurance coverage incl ude the t ype of health insurance policies
eligible for t he incentives , t he shares of individual and family premiums they cover,
and t heir ultimate policy objectives.
Contents
RationaleforGovernment InterventionintheMarket forHealth Insurance .....5
Im portant Considerations in Designing Tax Incentives to Ex pand
Health InsuranceCoverage ......................................7
Current FederalTax SubsidiesforHealthInsurance ...................7
Relevant Characteristics o f t he Non-Elderly Uninsured Population .......8
Relevant Features ofHealth InsuranceMarketfortheNon-Elderly .......9
Price S ensitivity of the Dem and for Health Insurance ................11
Potential Effectiveness and Cost of Some Proposed Tax S ubsidies To
Expand Health InsuranceCoverage ...............................13
LewinGroupStudy ...........................................13
Lewin Group’s Findings o n C ost-Effectiveness .................14
Lewin Group’s Findings o n Effectiveness ......................15
Lewin Group’s Findings o n C ost .............................15
Gruber 2000 Analys is .........................................15
Gruber’s FindingsonCost-Effectiveness ......................17
Gruber’s FindingsonEffectiveness ...........................17
Gruber’s FindingsonCost ..................................18
Gruber’s FindingsonEquity ................................18
Gruber’s FindingsonEmployerHealth InsuranceCoverage .......18
Gruber’s FindingsonKeyAdministrativeIssues ................19
Gruber 2001 Analys is .........................................19
NewSubsidiestoEmployers ................................19
NewSubsidiestoEmployees ................................20
NewSubsidiesforNon-Group Coverage ......................20
Gruber 2002 Analys is .........................................21
Conclusions .....................................................22
Tax Subsidies for Expanding Health
Insurance C overage: Selected Policy
th
Issues for the 108 Congress
An issue o f c o n t i n u i ng concern t o C ongress is the number o f Americans who
lack health insurance coverage. In 2001, the m ost recent year for which estimates are
available, 41.2 million p ersons were uninsured 1, 40.9 million (or 99%) o f whom were
under age 65. Among the non-elderly population , this is the highest total ever
reported b y t he U.S. Census Bu reau, and it marks t he first time since 1998 that the
number o f uninsured under age 65 ex ceeded 40 million. The number o f non-elderly
uninsured rose each year from 1987 to 1998 and t hen d eclined in both 1999 a n d
2000, before rising again i n 2001. Much of the ex p ansion since t he mid-1980s has
bee n d u e t o sust ai ned i ncreases i n t h e cost o f h eal t h care, whi ch i n t urn d ri ve up2
privat e health insurance premiums.
Congressional concern about the s ize o f t he uninsured population h as multiple
root s. One i s t he adverse h e a l t h e f fect s associ at ed wi t h a l ack of heal t h i n surance
coverage. C ompared t o i nsured individuals , i ndividuals with no or inadequate health
insurance are more likely t o postpone or do without needed health care, incur
burdensom e p ersonal d ebt s for care t hey recei ve, b e d eni ed access t o h eal t h care, be
hospitaliz ed for p reventable health problems, and h ave a greater likelihood of being3
di agnosed wi t h l at e-st age breast , prost at e, col orect al , o r s ki n cancer. Another s ource
of congressional concern i s t he economic costs arising from b eing uninsured.
Accordi n g t o recent s t udy of t h e t radeoffs i n usi n g t a x c r e d i t s t o ex pand heal t h
insurance coverage, t he uninsured pay only 30% of the actual cost o f t heir care.4 The
cost of uncom pensat ed m edi cal care recei ved b y uni nsured i ndi vi dual s i s passed o n
to the i nsured in the form of higher health insurance premiums ; t o l o cal , s tate, and
1 T here i s s ome question a bout how ma ny of these i ndivi duals can be considered chronically
uninsured. T h e e s timate i s i ntended t o capture those who were without health insurance
during all of 2001. Nonetheless, in recent t estimony before the Senate Special Committee
on Aging, former Congressional Budget Office Direct o r Dan Crippen c laimed that the
number of Americans who are uninsured for more t han one year at a s tretch “is s ubstantially
lower t hen 40 million, perhaps as much as 20 million l ower than 40 million. ” ( See “Ex-
CBO Head Urge s Focus on Uninsured W ho Rema in Without Cove rage Over T i me ,” Daily
Report f or Executives , Bureau of National Affairs, March 11, 2003, p. A-37.)
2 Ha r ve y S. Ros e n, Public Finance,6th edition (New York: McGraw-Hill Irwin, 2002), p.
3 K a iser Commi ssion on the Uninsured a n d M edicaid, The Uninsured: A Primer
(Washington: Henry J . K aiser Family Foundation, March 2002), pp. 6 a nd 8.
4 Mark Pauly a nd Bradley Herring, “ Expanding Cove rage V i a T ax Credits: T rade-Offs and
Outcomes,” Health Affairs, vol. 20, no. 1, J a n./Feb. 2001, p. 15.
federal governments i n t he form of high er subsidies for uncompensated care; and t o
heal t h care p rovi ders i n t h e form o f reduced i n com es.5 Fi nally, c o n gr e s s ional
concern about the uninsured population also r eflects a view held by many Americans
that health care i s a merit good. According t o t his v iew, all i ndividuals, regardless
of ability to pay, ought to have access t o adequate health care. Ex ponents also argue
that if choices must be made, t hey s hould b e done so on the b asis not of a p atient’s
i n com e or weal t h but of fact ors rel at ed t o t h e cost -effect i v eness o f recom m ended
m edi cal care, such as t h e age and s ex of a p at i ent or t h e appropri at eness and cost o f
prescri b ed t reat m ent s.
These concerns have led policym akers at t he state and federal l evels o f
government to search for effect ive, affordable, and politically feas ible ways to
ex pand access t o adequate health insurance coverage. W hile a v ariety of proposed
sol u t i ons have been anal yz ed and d ebat ed – i ncl udi ng t h e cr e at i o n o f unrest ri ct ed
medical savings accounts and a s ingl e-payer h ealth care s ys tem – recent C ongresses
have shown a growing i nterest i n t ax -bas e d approaches t o i n crease t hi s coverage. 6
In the 107 th Congress, the Trade Act o f 2002 (P .L. 107-210) established a new
refundable t ax credit for t he purchase o f health insurance b y i ndividuals who receive
or are el i gi b l e t o recei ve t rade adj ust m ent assi st ance and who are b et ween t h e ages
of 55 and 6 4 and recei ve benefi t s from t he P ensi o n Benefi t s Guarant y C o rporat i on.7
A n u m ber o f p roposals t o establish n ew or ex pand ex isting t ax incentives fo r
heal t h i n surance coverage have surfaced i n t h e 108 th Congress.8 For ex ample, i n l at e
5 There i s some uncertainty over the a mount of uncompensated care received by t he
uninsured. A maj or difficulty with many available estimates is that the data used t o generate
them are f lawed. The l atest attempt t o avoid this difficulty is a study by two senior analysts
from the Urban Institute. According to the study, uninsured individuals received an
estimated $35 billion i n uncompensated care in 2001, or about 3% of total personal health
care s pending. Federal, s t a t e , and local governments covered an estimated 87% of that
amount through a combination of grants, direct provision of care, appropriated f unds, and
special payments under M edicare and Medicaid. See J ack Hadley and J ohn Holahan, “How
Much Medical Care Do the Uninsured Use, and W ho Pays for It?,” Hea l th Affairs,Web
Exclusive: [http://www.healthaffairs.org] , visited Feb. 12, 2003, pp. W3-78.
6 T his is not to imply t hat t here are no s ignificant differences between congr e s sional
Republicans a nd Democrats ove r t he use of t ax subsidies t o r educe t he uninsured population.
The t wo s i d e s do not fully agree on some critical issues, such as t he merits of targeting
subsidies at non-gr oup coverage and t he need for r eforms of the private health insur a n c e
industry t o discourage r ating and underwriting practices that make it difficult for high-risk,
low-income indivi duals to obtain adequate, affordable cove rage . See Robert Cunningham,
“J oint Custody: Bipartisan Inte rest Expands Scope of T a x-Credit Proposals,” Health Affairs,
Web Exclusive [ http://www.healthaffairs.org] , visited Sept. 18, 2002, pp. W290-W298.
7 T h e c r e d i t i s equal t o 65% of health insurance premi ums paid by eligible persons for
policies covering t hems elves and qualifyi ng family members. It can be paid in one of two
ways. First, t he Internal Revenue Service can pay t he credit directly to taxpayers when they
file their a nnual t ax returns. Second, the IRS ca n pay the credit directly to a health insurer
on a monthly basis, provi ded t he eligible taxpayers first pay their 35% share of t he premium.
8 For a brief overview of l egislative proposals i n t he 108th Congress to expand tax benefits
for health insurance, see CRS Issue Brief IB98037, Tax Benefits fo r H e a l t h Insurance:
(continued...)
J une, t he House p assed a bill (H.R. 2596 as incorporated into the v ersion of H.R. 1
passed b y t he House) that, am o n g o t h e r t hings, would create t wo new p ersonal
savings accounts t o p ay for m edical ex pens es not reimbursed under h ealth insurance
plans with tax advantages similar t o t hose available t o current hold e rs o f A r cher
medical savin gs accounts: health savings accounts (HS As) and health savings
security accounts (HSSAs). 9 The former would apply t o i ndividuals covered under
health plans with an annual d eductible of at least $1,000 for i ndividual coverage and
at least $2,000 for family coverage; t he latter would apply t o uninsured individuals
or to individuals covered under h ea l t h p lans with an annual d eductible of at least
$ 5 0 0 for individual coverage and at l eas t $1,000 for family coverage. For b o t h
accounts, individual contributions would be tax -deductible, even i f t he tax p ayer does
not itemize; em ployer contributions would be t ax -ex empt; family member
cont ri but i ons woul d b e m ade o n an aft er-t ax basi s; i nvest m ent earni ngs woul d accrue
free o f t ax ; and distributions would b e t ax -ex empt i f t hey a r e u s e d f o r qualified
medical ex penses. H.R. 2596 (as i ncorporat ed into the v ersion of H.R. 1 p assed b y
the House) also would p ermit up to $500 in unused health care flex i ble s pending
accounts t o b e carried over, contribu t e d t o a qualified retirement o r d eferred
compensation p lan, or contributed to a HSA.
In addition, Representative C li f f S t earns has i ntroduced a b ill (H.R. 198) to
per m i t i n d i viduals to deduct 100% of out-of-pocket p ayments for health insurance
cove r a ge a n d p rescription d rugs , regardless o f whether they itemiz e. H.R. 583,
introduced by Representative M ark Ke nnedy, would establish a refundable,
advanceabl e t ax credi t for t h e purchase of qual i fi ed h eal t h i n surance b y i ndi vi dual s
not covered b y public insurance p rograms or s ubsidiz ed employer health plans. And
at least five bills (S. 10, S. 53, S. 86, S. 414, and H.R. 450) to establish t ax credits for
sm al l fi rm s t h at offer h eal t h i n surance t o em p l o yees have been i n t roduced; i n each
case, t h e credi t w oul d equal a cert ai n percen tage of the cost i ncurred by an eligible
employer in providing health insurance coverage to employees.
The Bush Administration also favors a tax -based approach to redu ci n g the
uninsured population. In his budget req u e s t f o r FY2004, President Bush p roposes
that Congress pass a limited, means-tested refundable t ax credit for t he purchase o f
non-group health insurance b y i ndividuals not covered under employer o r public
heal t h pl ans.10
8 (...continued)
Current Legislation, b y Bob Lyke , pp. 9-13.
9 For more d e t a i l s on the design of t he two proposed health savi ngs a ccounts a nd the
si milarities and di fferences between them and existing health savings account s , see CRS
Report, Tax-Advantaged Accounts f or Health Care Expenses: Side-by-Side Comparison,by
Bob Lyke and Chris L. Peterson.
10 More specifically, t he Bush Admi nistration wants to establish a refundable t ax credit for
the p u r c hase of health insurance by i ndivi duals under a ge 65 who a re not cove red under
public or employer health plans. T he credit would s ubsidize up t o 90% of the health
insurance premi um and would be capped a t $ 1 ,000 per a dult a nd $500 per c hild for up t o
two children. Single filers with no depende n t s and a modified adj usted gr oss i ncome
(MAGI) of up t o $15,000 would be e ligible for t he ma ximum c redit; the c redit phases out
(continued...)
The u se of tax policy t o s hrink t he uninsured population appeals t o s ome for a
variety o f reasons – although p roponents o f t ax subsidies t o ex p and h ealth insurance
coverage do not necessarily agree o n which reasons are p aramount. Foremost among
them are a p reference for t a x i n c e n tives to government spending in achieving
important social goals; a recognition that existing federal tax subsidies for health
insurance coverage ar e i n e q u itable i n t hat t hey ex clude the non-elderly who are
unem p l o yed o r d rop out of the workforce; and t he belief t hat p rivate insurance
com p ani es w oul d b e m ore effi ci ent and effect i v e t h a n governm ent agenci es i n
providing Americans with a range of desirable health insurance plans. Ot h ers,
however, doubt that tax s ubsi d ies are an efficient o r effective way to achieve
substantial reductions in the s iz e of the uninsured population. T h e y t e n d to favor
more direct solutions, s uch as public health insurance p rograms t a r ge t e d at t he
uninsured.
This report s ummariz es what i s known about the k ey factors s haping the cost-
effect i v eness o f t ax subsi d i es for ex pandi ng health insurance coverage. It opens with
an ex amination of t he economic justification for government intervention i n t h e
m arket for h eal t h i n surance. The re port t hen m ove s o n t o d iscuss important
considerations in designing t ax subsidies for ex panding health insurance coverage
and concl udes wi t h a revi ew o f t he pri n ci pal fi ndi ngs o f recent s t udi es assessi ng t h e
cost-effectiveness o f a variety o f p ropos ed tax s u b s idies, many of which mirror
proposals considered in recent C ongresses. The report will be updated o r revised to
refl ect l egi sl at i o n t o ex p and federal t ax i ncent i v es for h eal t h i n surance p assed b y a
full committee i n t he House or S enat e, or to incorporat e t he findings o f significant
new s t udi es of t h e cost -effect i v eness o f s uch i ncent i v es.
10 (...continued)
completely at a M A G I o f $30,000. All other filers with a M AGI up to $25,000 would be
eligible for t he maximum credit. In the case of a household consisting of t wo adults and t wo
children, the credit would phase out completely at a MAGI of $60,000. Taxpayers could
claim t he credit either on their f ederal tax r eturns or through r eductions in their premi um
payments equal t o t he credit they are due. If t hey choose t he second option, health insurers
would r eceive reimburseme nts from t he T r easury Department e qual t o t he amount of the
credit. Eligibility for t he advance credit would be based on an indivi dual’s t ax return for t he
previous year. Health plans eligible for t he credit would need to meet mi nimum coverage
c r i t eria. Indivi duals could purchase t hese plans t hrough t he non-gr oup ma rket, pri va t e
purchasing groups, s tate-s ponsored purchasing pools, or state high-risk pools.
Rationale for Government I nterve ntion i n
the M arket for H ealth Insurance
Mainstream economic theory holds that market s are most likel y t o achieve thei r
best possible outcomes when governments refrain from i ntervening in th eir
operations. Yet one of the d efining character istics of t he U.S. health insurance m arket
is the dominant role played by s tate and federal heal t h i n s urance program s and tax
subsidies for health insurance. 11 In 2001, for ex ample, s pending on Medicare and
Medicaid (which p rovide health insurance for the elderly and a large number o f t he
poor and d isabled) and t he State C hildren’s Health In surance P rogram totaled $454.8
billion; and t he federal government lost an estimated $65.7 b i l lion i n t ax revenues
because of t h e t ax ex cl usi o n for em pl oyer-provi ded h eal t h i n surance, accor d i n g t o
the J oint Committee o n Tax ation. On what economic grounds can such involvement
bejustified?
Conventional economic analys is suggests t wo possible grounds. On t h e one
han d , m a i n st ream econom i c t h eory hol ds t h at soci al wel fare i s enhanced by t h e
widespread availability of act uarially fair insurance. Most individuals are risk-averse
to some degree, and risk-averse individuals are t hought to be better o ff when they can
reduce t heir financial risks by transferring them t o en t ities with a comparative
advantage i n risk bearing b ecau s e of thei r capacity for risk pooling and
diversification ( e.g., private insurance companies). Thus, m ost i ndividuals would
want to buy enough h ealth insurance, provided t he cost reflects t heir ex pected annual
losses from health problem s. Health economists seem to agree t hat health insurance
is likel y t o yield max imum s ocial wel fare gains when coverage is widespread am ong
al l ri s k group s , t h e p o t ent i al l osses from h eal t h probl em s are l arge, and t he
probabilities o f s uffering such problems cannot be known for specific i ndividuals.
On th e o t h er h a n d , if the m arket for health insurance were l eft t o its own
devices, t here would b e n o certainty that health insurance coverage would p roduce
a n et wel fare gai n. In st ead of t h e em ergence o f broad, act uari al l y fai r i n surance
coverage, one might find that available coverage would b e ex cessive in many cases,
and t oo few i ndividuals would h ave adequate a n d affordable insurance coverage.
Undesi rabl e out com es s uch as t hese m ay ari se because t h e p rovi si on of heal t h care
is vulnerable t o f o u r market failures: adverse s el ect i on, m o ral h az ard,
incompleteness, and t he ex istence of free riders i n t he utilization of m edical care. 12
11 For an overview of existing f ederal tax subsidies f or health insurance, see CRS Issue Brief
IB98037, Tax Benefits for Health Insurance: Current Legislation, b y Bob Lyke , pp. 2-7.
12 The health care market appears r iddled with imperfections. In addition t o t h e three
problems discussed here, consumers a nd h ealt h insurance companies face imperfect
information r elating t o t he skills of doctors, health profiles of patients, and t he benefits of
treatments and procedures; competition i n t he provision of health care servi ces is limited;
and provi ders whose principal o b j e c t i ve goes beyond the minimization of costs or
maximi zation of profits play a l arge role in the delivery of health care. For more details onrd
these market f ailures, see J oseph E. Stiglitz, Economics of t he Public Sector ,3 edition
(New York: W .W. Norton, 2000), pp. 308-311.
Adverse s el ect i o n refers t o a process w hi ch i s l i k el y t o em erge i n t he heal t h
insurance m arket when buyers of particular policies have great er risks of illnes s and
i n j u ry than the average potential buyer. It i s m ost likel y t o m at erialize when
participation i n h ealth insurance coverage is voluntary, risk pools a r e small and
dominat ed by ex perience rating, and a multitude of insurers vie for market share by
o f f e r i n g consum ers a range o f p l ans wi t h varyi n g b enefi t s . In general , t he average
buyer of a h ealth insurance p lan h as an a bove-average risk of developing serious
heal t h probl em s. S u ch a t endency i s root ed i n t h e fact t h at a v ery s m al l share o f t he
U.S. populati o n a c counts for a l arge share o f s pending on health care i n a gi ven
year. 13 This disparity, i f uncorrected by policy i ntervention, can touch o ff an upward
spiral in insu r a n c e p r e m i ums d riven b y efforts b y i nsurers t o raise rates t o cover
losses from unanticipated claims. In respons e, individuals with relatively l ow risks
(e.g., young, h ealthy adults) m ay choose t o b ecome uninsured or switch t o cheaper
policies offering less coverage, while individuals with relativel y high risks (e.g.,
m i ddl e-aged adul t s who s m oke, d o not ex erci se regu l arl y, and are overwei gh t ) face
the choice of payi ng premiums far above their ex p ected losses, switching to cheaper
policies offering inadequate coverage, o r b ecoming uninsured and relyi ng on charity
care as n eeded. T he prevale n ce of advers e s election i n non-group markets gives
insurance companies an incentive t o engage i n a practice known as “cherry picking, ”
where t hey focus on selling health insurance t o t hose with relativel y l ow risks.
Among health economists, m oral haz ard refers not to a character defect but to
the ex cessive utilization of health care f ostered by having health insurance and the
associ at ed effi ci ency l o sses. He a l t h i n s u rance can have t h i s effect m ai n l y because
insured i ndividuals bear only a fraction o f t he actual cost o f t h e h e a l t h care t hey
recei ve, w eakeni n g t hei r i n cent i v e t o avoi d b ehavi o rs t h at ar e known to contribute
to health problem s and dulling t heir awareness of t he relationship bet w een the
m argi n al cost s and benefi t s of t h e care t hey recei ve. In t heory, i n sured i ndi vi dual s
will consume health care until the m argi nal benefits match t he marginal out-of-
p o c k e t costs, which under m ost policies are equivalent to deductibles and co-
paym ents . G e n e r ous health insurance coverage can push t hese costs far below t he
a c t u a l m argi nal cost s of t h e m edi cal servi ces. Of course, i nsurance com p ani es a r e
well aware of t he damage moral haz ard can inflict on t heir financial viability; s o t hey
try to contai n or limit it through t he imposition of deductibles, co-payments, coverage
limits and ex clusions, m anaged care plans, m an d a t o ry prior approvals of certain
treatments a n d p rocedures, and financial i ncent i v es for h eal t h -care p rovi ders t o
restrict the u se of costly procedures.
Moreover, it appears t hat t he health insurance m arket fails to satisfy a necessary
condition o f economic efficiency: t he ex is tence o f what economists call “complete
markets.” W hen a market fails to provide a good or service even t hough i ndividuals
would be willing t o pay more than the aver a ge cost of providing it, economists
consi d er t h e m arket t o b e i ncom pl et e. And t hey regard t he l ack of com p l et eness as
a m arket failure. S ome argue that the p rivate health insurance m arket i s i ncomplete
because it usually does not provide gu aranteed renewability or i n s u r a n ce against
13 In 1992, fo r e x a mp l e , 10% of Americans accounted for 72% of personal health
expenditures.
future health risks, only current one s.14 Most individuals cannot purchase h ealth
insurance at fix ed rates and under fix ed terms for more than one year in advance. Yet
it is thought th at m a n y i n dividuals would be willing t o purchase i nsurance against
future health risks at a reasonable cost, if it were provided.
Fi nal l y, i n every st at e, cert ai n heal t h car e providers – m ainly public hospitals –
have a l egal obligation t o provide treatment to someone suffering from a life-
threat ening or serious health pro b l em , i rrespective of his or her ability to pay.
A l t hough t hese mandatory-care l aws can enhance s ocial welfare by ensuring t h a t
everyone recei ves at l east a m odi cum o f h eal t h care w hen i t i s u rgent l y needed, t hey
can also undermine welfare by diluting t he inc e n t i v e t o purchase actuarially fair
health insurance, es peci ally for l ow-income i ndividuals with few financi al assets to
prot ect . A s a resul t , m andat o ry-care l aw s h a v e t he unintended effect of layi ng the
foundation for a classic m arket failure known as t he free rider problem. In general,
this problem can be seen as a b yp roduct o f t he e x i s tence o f public goods, one of
wh o s e d efining t raits is the impossibility of ex cluding anybody from consuming
t h em . Laws grant i n g everyone t h e ri ght t o recei ve em ergency care, regardl ess of
ability to pay, effectivel y endo w health care with this trait, thereby giving rise t o a
free-ri der p robl em . It i s not cl ear how t h e w el fare gai n s from t hese l aws com p are
with the wel fare losses.
For m ost h eal t h econom i s t s , t he presence of these m arket failures p rovides
am pl e j u s t i fication for government intervention i n t he health insurance m arket. In
t h ei r v i ew, a cent ral i ssue for pol i cym akers i s not whet he r t o i nt ervene. R at her, i t
is to devise politically acceptable policy measures that attempt t o s trike a b alance
between the welfare losses arising from m or al haz ard and t he welfare gains inherent
in the risk s preading provided by act uarially fair health insurance coverage.
Important Considerations in Designing Tax
Incentive s to E xpand H ealth Insurance C overage
If the poli c y o b j ective i s t o achieve a s iz able reduction i n t he uninsured
population and the p referred policy i nstrument i s t ax policy, then a critical task for
policym akers i s t o d evise t ax subsidies t hat accomplish t he go al at an accep table
budgetary cost. Recent research i n t o t h e probable effects o f t ax policy o n h ealth
insurance coverage sheds light on several factors t hat s hould b e carefully considered
in designing s uch s ubsidies. Each is ex amined in some detail below.
Cur r e nt Feder a l Tax Subsi di e s f or Heal th I nsur a nce
T h i s m a y s eem obvious, but one important factor to weigh i n d evising a ta x -
based approach to ex panding health insurance coverage is the current mix of federal
tax s ubsidies for health insurance coverage. Any such approach cannot avoid b eing
constructed around a d ecision about whet her t o j ettison, modify, o r retain t hese
14 See Leonard E. Burman a n d A me l i a Gr uber, “First, Do No Harm: Designing T ax
Incentives for health Insurance,” National Tax J ournal , vol. 59, no. 3, Sept. 2001, p. 480.
subsidies. The reas on lies i n t he central role they play in the current system of health
insurance coverage for t he non-elderly.
Although t he present t ax code contai ns a v ariety o f subsidies for health
insurance, three i n particular have facilitated t he em ergence of t his s ys tem: the t ax
ex cl usions for employer-pai d health insurance premiums, s o-called cafet eria plans,
and flex i ble s pending accounts . 15 T h e y are r el at ed i n t h at each pert ai ns t o heal t h
insurance coverage obtained t hrough employers. Under t hese ex clusi o n s , i t i s
possi bl e for i ndi vi dual s who recei ve heal t h i n surance t hrough em p l o yers t o pay n o
federal or s tate income and payroll tax es on t heir o w n and thei r employers’
contributions for t he premiums. In combination, the ex clusions substantially lower
the after-tax price of health insurance rel ative t o other good s and services workers
may consume and subsidi z e t h e purchas e of employm ent-based group health
insurance relative t o non-group coverage.16 These e x c l u s i ons are s o d eepl y
embedded i n t he current structure o f h ealth insurance coverage for t he non-elderly
that the i r r e m o v al could t rigger a sign if icant i ncrease i n t he uninsured population.
In a recent s tudy, J onathan Gruber, an economist at M IT who h as written ex t ensively
on the impact of tax policy o n h e a l t h insu rance coverage among the non-elderly,
estimated t hat t he elimination of t he ex cl usions at the federal and state levels would
lead to a 14% drop in th e n u m b e r of employers offering health insurance t o
emp l o ye e s , which in turn would result i n a decrease i n t he number o f i ndividuals
covered under employm ent-based h ealth plans o f over 2 2 million and a 50% decline
in total s pending on these plans.17
Releva nt Characteristics of the
Non-El der l y Uni nsur ed Popul ati on
P o l i cym akers s e eki ng t o craft cost -effect i v e t ax subsi d i es t o ex p and h eal t h
insurance coverage should also t ake i nto account charac t e r i s t i cs o f t he non-elderly
uninsured population t hat are relevant to tax policy.
Several s uch characteristics d eserve mention. Fi rst, the non-elderly uninsured
are d isproportionately concentrated in low-income households: i n 2001, nearly 38%
of uninsured individuals resided i n hous eholds with annual i ncomes below $20,000.
It comes as n o s urprise, then, t hat a very hi gh proportion o f uninsured tax p ayers fall
in lower t ax brackets, and t hat a sign ificant p roportion h ave n o federal income tax
liability.18 Second, while most uninsured individuals under age 65 live i n households
15 For an overview of current federal t ax benefits for health insurance, see CRS Issue Brief
IB98037, Tax Benefits for Health Insurance: Current Legislation, pp. 1-8.
16 J onathan Gruber, an economist at M IT , has estimated that the t ypical worker in the 15%
federal i ncome t ax bracket, f acing a s tate income tax r ate of 5% and a combined payroll tax
rate of 15.3%, pays $0.65 after t axes for $1 of health insurance, meaning t hat t he tax price
is 0.65. See J onathan Gruber, Taxes and Health Insurance, W orki ng P a p e r 8657
(Cambridge , M A: National Bureau of Economic Research, Dec. 2001), p. 4.
17 Ibid., pp. 26-27.
18 Gr uber has also es t i mated that in 1997, 90% of uninsured taxpayers with tax liabilities
(continued...)
headed by individuals holding full- or part-time jobs, over 75% of uninsured workers
are not offered h ealth insurance b y t heir employers.19 Third, many uninsured workers
are employed b y s mall firms: in 2001, over 50% of uninsured workers h eld j obs with
fi rm s h avi n g fewer t h an 100 em pl oyees. 20 Fi nal l y, t he previ ous charact eri s t i cs
notwithstanding, t he uninsured under age 65 do not constitute an isolated, easily
i d e n t i fied or targeted, and stable popula tion group. Rather, t hey are scattered i n
varyi n g and shifting d egrees throughout all i ncome groups and firm s iz es. For
instance, i n 2001, 24% of uninsured individuals lived in households with annual
incomes o f $50,000 or more, and 23% of uninsured workers were employed b y firms
with 500 or more employees.21
Releva nt Features of Health Insurance
M a r ket for t he Non-El der l y
It is al so u s efu l to consider certain feat ures of the ex i sting s tructure of health
i n surance coverage for t he non-elderly i n d esigning tax -based s ubsidies t o ex p and
health insurance coverage. Of particular interest for policym akers are the dominant
r o l e of employers as s ources of health insurance coverage, t he advantages and
disadvantages of group coverage provided t hrough t he workplace, and t he limitations
of coverage provided i n t he non-group market.
In 2001, nearly 66% of non-elderly Americans received h ealth insu r a n c e
coverage through employers; anot her 15% were covered t hrough public health
i n surance p rograms; and about 7% obtained coverage through i nsurance p l a n s
purchased in the non-group or individual m a r k et. Gruber h as noted that in recent
decades, m ore t han n ine out of 10 privatel y i nsured non-elderly i ndividuals received
their coverage from an emp l o yer, either thei r o wn, t heir spouse’s o r t heir parents’.
Most workers who are o ffered i nsurance by t h ei r em p l o yers, regardl ess o f wage and
salary income, d ecide to take it, although the take-up rate appears t o h ave d eclined
som ewhat i n recent years as em p l o yers have passed o n t o em p l o yees i n creases i n t h e
cost of enrolling i n health plans.22 The o ffering of insurance v aries considerably by
firm siz e. In general, s maller firms are m uch l ess likely t o o ffer coverage: i n 2001,
a worker i n a firm with fewer t han 1 0 employees was almost three times as likel y t o
be uninsured as a worker i n a firm with 1,000 or mo r e e m p l o yees. Nevertheless,
sm al l fi rm s payi ng hi gh average w ages are as l i k el y t o o ffer coverage as l arge fi rm s ;
18 (...continued)
belonged to the 15% tax bracket, and 50% had no f ederal tax liability. See J onathan Gr uber,
Tax Subsidies f or He alth Insurance: Evaluating t he Costs and Benefits, W orking Paper 7553
(Cambridge , M A: National Bureau of Economic Research, Feb. 2000), p.7.
19 Gruber, Taxes and Health Insurance,p.7.
20 CRS Report 96-891, p. 6.
21 EBRI, pp. 11 and 12.
22 DavidM.Cutler,Employee Costs and the Decline i n H ealth Insurance Coverage,
Working Paper 9036 (Cambridge , M A: Nationa l Bureau of Economic Research, J uly 2002),
pp.25-27.
and eligibility, coverage, and take-up rat es are s imilar among em ployees in smal l and
large firms that offer h ealth insurance coverage. 23
There are noteworthy advantages and d is advantages to tying health insurance
coverage to employment. On t he one hand, the work p lace serves as a convenient,
almost logi cal way t o pool health insurance risks, s ince most individuals choose t heir
em pl oyers for reasons ot her t h a n ex p ect ed usage o f m edi cal care i n com i n g years.
In general, the l arger t he employment group, the greater the s cope for t he risk sharing
and com m uni t y rat i n g t hat enhances soci al w e l fare. S een from t hi s p erspect i v e,
employment-based insurance h a s the pot ent i al t o p recl ude t h e adverse s el ect i o n
prev ailing i n t h e n on-group insurance m arket. Fu rthermore, insurance
administrative, underwriting, and m arketing costs are l ower for l arge groups than
individuals.24 Large groups also have gr e a t e r b a r gaining power in ex tracting p rice
and o ther concessions from i nsurers and health-care p roviders.
On t h e o t h er hand, em pl oyer-based heal t h i n surance s erves as an i m p erfect ri sk-
p o o l i n g m echani s m i n t he case o f s m al l fi rm s. In a fi rm wi t h 20 em pl oyees , f o r
instance, if two i ndivi d u a l s suffer a serious illness, the average cost of health
insurance can rise markedly the following year. In addition, making employers t he
prima r y source of health insurance coverage for t he non-elderly can contribute t o
effici ency losses by ret ardi n g j o b m o bility. Insured workers m ay be disinclined to
work for another employer o u t o f f ear of losing their coverage or having less
desirable options for coverage. Another d rawback t o em pl oyer - p r o v i ded h eal t h
insurance i s t hat i t offers limited renewability. Those who become unemployed m ay
lose their coverage, and there are no federal l aws t o k eep employers from d ropping
coverage when faced wi t h huge p rem i u m i ncreases. Fi n al l y, a syst em of
employment-based health insurance coverage can foster production i nefficiencies by
implicitly favoring l arge firms over s mall firms. The s ource of the i neffici enci es is
the considerable cost advantages that large firms typically enjoy over s mall firms i n
offeri ng t h e s am e p ackage o f h eal t h benefi t s t o em pl oyees. 25
Some proposals t o establish n ew t ax s ubsidies t o ex p and h ealth insurance
coverage would rely o n t he non-group mark et to achieve their objectives. In 2001,
16.4 millio n i n d i viduals were covered under non-group policies. Coverage under
these policies d eclined gr a d u ally between 1993 – when i t reached 17.5 million
23 J a son Lee, Are Health Insurance Premiums Higher f or Small Firms?,ResearchSynthesis
Report No. 2 ( Princeton, NJ : Robert Wood J ohnson Foundation, Sept. 2002), pp. 4-6.
24 According t o o n e a uthoritative estimate, health insurance l oading fees as a percent of
benefits pa i d f a l l i n the r ange of 60% to 80% for i ndivi dual policies, compared to 8% to
15% for groups of 201 to 1,000 persons. T hese fees cove r t he operating c osts of insurance
companies a nd represent a ma rk-up ove r t he expected payout of benefits under a policy. Seerd
Charles E. Phelps, Health Economics,3 Edition ( Boston: Addison Wesley, 2003), pp. 342-
25 Small f irms pay more t han l arge firms f or the s ame health insurance coverage because the
costs of marketing and admi nistration are s p r ead over fewer workers in small f irms and
underwriting costs typically are greater for small firms. See U.S. Congressional Budget
Office, The Tax Tr eatment of Employment-Based Health Insurance (Washington: March
persons – and 2000. Many of those who purchase i nsurance in the non-group market
are s el f-employed, retired but not yet eligible fo r M ed i c are, working part-time, or
divorced or w i d o w e d . 26 Access t o and the cost o f non-group coverage depend
critically on a p erson’s h ealth status, age, and place of residence. Fo r young adults
in good health, p remiums can be lower t han i n t he employer-based group market; but
for older adults with ch ronic health problem s, premiums tend to be much higher.
Generally, i ndividuals in po o r health can face form i d abl e obst acl es t o obt ai ni ng
adequate and affordable coverage. Insurers o ften decline t o cover p ersons with pre-
ex isting conditions, impose s evere limits on the coverage they will provide for t hose
conditions, o r add high premium s urcharges t o cover t hose conditions. 27 Many states
have responded to thes e pract i ces by es t abl i s hi ng hi gh-ri s k i n s u rance pool s ; and s om e
have enacted law b arring insurers from ex cluding c o v e r age of pre-ex isting
conditions. But th ey seem to have had little impact on trends in the uninsured
population i n t he past 10 to 15 years.
Price Sensitivi ty of the Dem and f or Health Insurance
Tax s ubsidies attempt t o boost t he demand for h ealth insurance b y reducing its
tax price (or after-tax cost). This relationship implies t hat i n des igning tax s ubsidies
to shrink the uninsured population, it is useful to know how sensitive t he decision to
purchase h ealth insurance i s t o reductions in its tax p rice. Given t he structure o f t he
health insurance coverage for t he non-elderly, tax policy can work through t hree
basic channels to ex pand coverage: (1) subsidies t o employers to offer coverage; (2)
subsi d i es t o em p l o yees t o accept coverage t h at i s offered; and (3) subsi d i es t o
uninsured individuals to purchase non-group coverage. These options suggest that
policym akers might consider the responsiveness t o t ax price changes of the d ecision
by employers t o o ffer coverage, t he decision by employees to take up coverage if it
is offered, and t he decision of uninsured individuals to purchase non-group coverage,
t o get h er wi t h t h e ways i n whi ch t h ese d eci si ons m ay feed back on one anot her.
Considerable research has b een done on the b ehavioral responses of firms and
i n d i vi dual s t o changes i n t he cost of heal t h i n surance. A recent s t udy by Grub e r
reviewed this literature. 28 One o f h is findings was a s triking l ack of consensus o n t he
price s ensitivit y o f t h e t h ree d ecisions shaping t he efficacy of tax s ubsidies as a
policy t ool to ex pand health insurance coverage. But such a l ack was not as
problematic for pol i c ym a k e r s as i t m ay have seemed upon first glance b ecause of
important differences in focus, methodology, data, and applicability of results among
t h e s t u d i es consi d ered by Gruber. He com p ared t h e s t rengt hs and w eaknesses o f
available estimates o f p rice sensitivity and i dentified t hose h e t hought were useful for
the purpose o f forecasting t he impact of propos ed tax s ubsidies o n h ealth insurance
coverage.
26 Karen Pollitz, Richard Sorian, and Kathy T homas, How Accessible i s I ndividual Health
Insuranc e f or Consumers i n Less-Than-Perfect Health? (Washington: K aiser Family
Foundation, J une 2001), p. 1.
27 Ibid., p. iv.
28 See Gruber, Taxes and Health Insurance, pp. 12-25.
On the question o f how r e s ponsive t he decision by firms t o o ffer h ealth
insurance t o employees was t o changes in its tax p rice, Gruber concluded t hat t he
best available estimate of this responsiveness was one of his o wn. In a 2000 study
done with Michael Lettau, he estimated t hat t he price elasticity of employer offering
of health insurance fell b etween -0.3 an d -0.4. This implied t hat a 10% rise in the t ax
price of health insurance could l ead to a decline i n t h e n u m b e r of firms offering
coverage to employees of three t o four percentage points.29
On the question o f how responsive t he decision of employees to accept h ealth
insurance coverage offered b y t heir employers w as t o changes i n i t s t ax p ri ce, Gruber
found that the d ecision seemed to be independent of such changes. In other words,
available evidence i ndicated that employees did not measurably alter t heir take up of
employer health plans i n response t o changes in their t ax price. He noted that this
finding was c o n s i s t e n t with “a growing body of evidence” that the d ecisions of
employers appear to have much more infl uence t h an t he preferences of employees
over t he availability of fringe benefits like health insurance.30
And o n t he question o f h ow responsive t he decision of individuals who l ack
public or emp l o ye r -provided h ealth insurance coverage to purchase non-group
coverage was t o changes in its tax p rice, Gruber noted that only one meaningful study
had b een done, and that it was t oo flaw e d to provide helpful guidance for
policym akers. The s t u d y estimated a tax -price elasticity of non-group coverage of
-0.3, implying t hat a 10% drop in the aft er-tax cost of non-group health insurance
would l ead to a 3 % i ncrease i n t he number o f i ndiv i d u a l s c o v e red under s uch
policies. In his v iew, the s tu dy’s shortcomings meant t hat t he estimate should be
const r u e d a s a “lower bound” on the t ax -price sensitivity of the d emand for non-
group coverage among those not eligible for public or employer health insurance. 31
Gruber’s stud y i ndicated that neither fi rms nor individuals were especially
sensitive t o reductions in the t ax price of health insurance. But i t did offer fresh
evidence that the o ffering decisions of fi rms t end e d t o b e m ore responsive t o s uch
reduct i ons t h an t h e accept ance d eci si ons of em pl oyees. T he st udy al so served t h e
purpose of highlighting s ome of t he significant gaps i n our understanding o f t he
forces shaping b ehavioral responses to tax s ubsidies for health insurance coverage.
Of particular interest was a paucity of conclusive evidence on the price el as ticity of
demand for non-group insurance among those currently lacking public or employer
health insurance. Little is also known about the p rice sensitivity of the uninsured
who are not offered h ealth insurance b y t heir employers.
29 Ibid., p. 16.
30 Ibid., p. 19.
31 Ibid., p. 22.
Potential E ffectiveness and C ost of
Some Proposed Tax S ubsidies
To Expand Health Insurance C overage
Recent p roposals t o u se tax s ubsidies t o achieve sub s t a n tial reductions in the
uninsured population raise the question o f what approach is likely t o b e m ost cost-
effect i v e. S everal recent s t udi es have addressed t hi s quest i on. Am ong ot her t hi ngs,
t h ey shed l i ght on t h e cri t i cal fact ors d et erm i n i n g t he cost -effect i v e n e s s o f t ax
su b s idies t o ex pand health insurance coverage. The principal findings o f t hese
st udi es are revi ewed b el ow.
LewinGroupStudy
In 1999, Lewin G r o up issued a s tudy which assessed t he effectiveness and
revenue cost of nine proposed tax s ubsidies for the purchase o f h ealth insurance. The
proposals ranged from s omething as simp le as a t ax deduct i o n for the purchase o f
non-group health insurance t o s omething as complex as t he replacement o f current
fed eral t ax subsidies for health insurance with individual i nsurance mandates a n d
refundable t ax credits for t he p u r c h a s e o f non-group health insurance. Some were
modeled after l e gi s l a t i v e i nitiatives then being considered in Congress. For each
proposal, t he Lewin Group use d a m i cro-simulation m odel o f t he U.S. health care
system (known as t he Lewin Group Health Benefits Simulation M odel) to estimate
the number of persons who would be eligible f o r t h e p r o posed tax s ubsidy, the
change in the number o f i nsured persons in response t o its adoption, and its federal
revenue cost in 20 00 dollars. These es timates t hen were used t o compute t he
percentage change in the uninsured population relative t o 1997 – t he most recent year
at the time for which figu res o n t he uninsured population were av ailable – and t he
revenue cost per n ewly insured p erson.
The micro-simulation m odel generat ed thes e estimates on t he basis of s everal
key assumptions. Two are worth mentioning here.
One assumption addressed t he sensitivity of consumers i n general to changes i n
the after-tax price of health insurance – or the t ax -p ri ce el asticity of demand for
h e a l th insurance. The p roposed tax s ubsidies l owered the after-tax price o f h ea l t h
insurance for eligible individuals, which in turn sparked an overall increas e i n health
i n s u r a nce coverage among these i ndividuals. Nat urally, t he magnitude of the
increas e depended critically on the price el as ticity of demand. The s tu dy used an
e l asticity of -0.2, which implied t hat a 10% decrease i n t he after-tax price o f
insurance would t rigger a 2% rise in coverage nationwide. Nonethel ess, considerable
uncertainty surrounded t he actual p rice elasticity. Health econom ists who h ad
studied the i ssue came u p with estimates rangi ng from 0.0 to -2.7. 32
32 T he coefficient f or the price elasticity used in the Lewin Group study came from a 1998
analys is by the s ame or ga n i za t i on t hat analyzed t he effect of changes i n employee
contributions to employer health insurance on t he number of workers and dependents
purchasing employer-provided health insurance. Other s tudies have come up with different
(continued...)
Another important assumption concerned t he number o f i ndividuals who would
claim o r t ake u p t he proposed tax s ubsidies. Fo r s ubsidies s upplem e n t i ng ex isting
federal t ax subsidies for health insurance, the s tudy assumed every eligible individual
currently covered b y non-group insurance would claim them, whether a d eduction o r
refundable t ax credit. The rate at which a t ax subsidy was taken u p m attered b ecause
i t s revenue cost and overal l effi cacy depended i n p a r t o n how m any i ndi vi dual s
actually took advantage o f it. Available evidence i n 1999 suggested, however, t hat
the t ake-up rate for t hese individuals was s ignificantly less than 100%.33
Several o f t he study’s findings h ave important implications for the potential cost
and effect i v eness o f t ax subsi d i es t o ex p and h e a l t h i n surance coverage. T hey are
high lighted below.
Lew i n G roup’s Findings on Cost-Effectiveness.34
! The m ost cost-effective s ubsidy was a refundable t ax credit for t he
purchase o f non-group insurance b y t ax payers not covered b y
em pl oym ent -based heal t h i n surance, Medi care, or Medicaid that was
capped at $500 for s ingl e p ersons and $1,000 for families; the cost
per n ewly insured p erson was $1,246 (2000 dollars).
! The l eas t cost-effective s ubsidy w as one that scrapped ex i sting
federal t ax subsi d i e s for t h e purchase o f h eal t h i n surance – ex cept
for m edical savings accounts – an d s ubstituted a fix ed refundable t ax
credit of $800 per adult and $400 per child (cap p ed at $2,400 per
family) for the purchas e of health insurance by all tax payers ex cept
those who are covered under M edicare o r M edica i d ; t h e cost p er
newly i nsured person was $10,541 (2000 dollars).
! Refundable t ax credits for t he purch ase o f h ealth insurance b y l ow-
income working and non-working i ndividuals were more effective
but more costly (per newly i nsured person) than an abov e -the-line
tax d eduction f o r t h e purchase o f non-group health insurance b y
32 (...continued)
estimates of the price elasti c i t y o f d e ma nd for health insurance, but in doing so they
employed different estimation methods, f or the most part. For a discussion of these s tudies,
see Gruber a nd Poterba, “Fundame ntal T a x Reform a n d Employer-Provi ded Health
Insurance,” pp. 159-162.
33 In a 1993 study of the r esponse of s elf-employed indivi duals to the creation of a partial
tax deduction f or their health insurance expenditures b y t he T ax Reform Act of 1986,
economists J onathan Gruber a nd J a me s Poterba found that from 1986 to 1989, between 15%
a n d 2 0 % o f the s elf-employed with income s below $20,000 and a bout 50% of those wit h
income s a bove $50,000 claime d t he deduction. Fr om these r esponse r ates, t hey c oncluded
that a l ack of awareness of t he presence of the s ubsidy undercut its effectiveness. See U.S.
Congress, House Committ e e o n W ays and M eans, Subcommittee on Health, Healthth st
Insurance Premium Deductions for t he Self-Employed, hearing, 104 Cong., 1 sess., J an.
34 Cost-effectiveness i n t his context denotes the r evenue cost per newly insured person.
tax p ayers who were not covered b y employm ent-based i nsurance,
Medi care, or Medi cai d.
Lew i n G roup’s Findings on Effectiveness. 35
! T h e m ost effective s ubsidy was a p lan d eveloped b y t he Heri t a ge
Foundation which would m ake s weeping changes in the s tructure of
health insurance coverage for t he non-elderly. Among other t hings,
it abolished ex i sting federal tax s ubsidies for health insurance,
substituted a refundable t ax credit for h ealth insurance and medical
care ex p enditures, and require d a l l individuals to purchase a
minimum l evel of health benefits and all employers t o convert their
health benefit plans to wages. The plan res ulted i n t he elimination
of the entire uninsured population, which was assumed t o t otal 43.3
millionin2000.
! The l east effective s ubsidy was a 30% refundable, means-tested tax
credit for non-group heal t h insurance purchased by tax p ayers
without access t o emplo ym ent-based h ealth insurance and not
co v e red under M edicare o r M edicaid. The credit phased out for
single persons with adjusted gross incomes between $25,000 and
$35,000, and for married couples with adjusted gross i n c o m es
between $40,000 and $50,000. It reduced the uninsured population
by 1.5 million persons.
Lew i n G roup’s Findings on Cost.
! The m ost costly subsidy was the Heritage Foundation plan: its net
revenue cost was $55.3 billion (2000 dollars).
! The l east costly subsidy was the 30% refundable t ax credit for t he
purchase o f non - g r oup health insurance b y i ndividuals without
access t o em p l o ym ent -based h eal t h i n suran c e and not covered b y
Medicare o r M edicaid: its net revenue cost was $3.3 billion (2000
dollars).
Gruber 2000 Analysis
G r u b e r’s 2000 study and t he Lewin Group study had m uch i n common. The
Gruber s tudy had a similar focus: t he potential effect s of alternative t ax subsidies for
health insurance o n t he number o f uninsured and t he federal budget. Like t he Lewin
Group s t udy, m any o f t he tax s ubsidies Gruber evaluated were modeled after
l e gi s l at i v e proposal s s t i rri n g i n t e res t i n recen t C o n gres s es . An d t h e Gru b er s t u d y al s o
employed a computeriz ed model o f t he li nks between tax policy and the U.S. health
i n surance m arket t o anal yz e t h ese effect s.
35 Effectiveness i n t his context denotes the t ot al decrease i n t he uninsured population, if any.
Bu t t he Gruber s tudy went beyond the Lewin Group study in that it addressed
two important issues raised by tax -based approaches to ex panding health insurance
coverage which w ere ex cluded or given marginal treatment in the Lewin Group
study. One issue was the ex t ent t o which different income groups benefitted from t he
proposed subsidies; and t he second con cerned t he impact of the s ubs i d ies o n t he
market for group health insurance. The Gr uber analysis also d iffered s omewhat i n
approach (but not in basic met hodology). It simulated the effects of a “base case”
along a number o f d imensions, i n c l u ding the federal revenue cost, t he siz e of the
uninsured population, the number o f i ndividuals with empl o ym e n t -based health
insurance, and t h e d i stribution o f t he net cost o f t he subsidy among major i ncome
groups.36 Gruber t hen s imulated t he effects o f other proposed tax s ubsidies (e.g. , a
non-refundable t ax credit for t he purchase of non-group insurance and an above-the-
line t ax deduction f o r the s ame purpose) on the s ame v ariables and compared t he
results to those o f t he base case.
For each proposed tax s ubsidy, Gruber es timated t he revenue cost to the federal
government, how many individuals woul d become insured, how the benefits would
be distributed among income groups, and how many individuals with employment-
based health insurance would drop it or lose it in response t o the subsidy.37 In
simulating t hese effects , h e h a d t o m ake assumptions about a number o f k ey
behavioral variables, including the ex t ent t o which those who were uninsured took
up the s ubsidies t o purchase non-group coverage, t he ex tent to which t hose already
covered b y non-group health insurance claimed t he proposed su b s i d ies, and t he
ex tent to which firms reacted to the subsidies by eliminating or cuttin g h e a l t h
benefits for employees. Not surprisingly, t he results of Gruber’s analys is hinged on
the s ensitivity of the dem and for health insurance coverage to declines in its after-tax
price. Gruber assumed a high er price elasticity than the Lewin Group study: -0.53
compared to -0.2. This difference was hardly trivial: a 1 % fall i n t he after-tax price
of health insurance i n both s tudies led t o an i ncrease i n i nsurance cov e r a ge a m o n g
the uninsured that was 2.6 times great er in Gruber’s anal ys is. W hile the validity of
Gruber’s assumed p rice elasticity is open t o question, it should b e noted that it was
cl oser to the middle of t he range of available elasticity estimates t h an t he one used
in the Lewin Group study.
S everal results of G r uber’s analys is also should b e h ighlighted because they
have important implications fo r t he effect i v eness and cost of t ax s ubsi d i es t o ex p and
health insurance coverage:
36 T he base case i nvolve d a refundable t ax credit for t he purchase of non-gr oup health
insurance by i ndivi duals not covered by M edicare or employment-based health insurance.
T he credit was limited t o $1,000 for s ingl e f i l e r s and $2,000 for j oint filers and head-of-
household f ilers, and i t p h a s e d o ut for s ingl e f ilers with adj usted gr oss i ncomes between
$45,000 and $60,000 and f or j oint a nd head-of-household f ilers with adj usted gr oss i ncomes
between $75,000 and $100,000.
37 Gruber, Tax Subsidies f or He alth Insurance: Evaluating t he Costs and Benefits, p. 9.
Gruber’s Findings on Cost-E ffectiveness.38
! The m ost cost-effective s ubsidy was a m eans-tested refundable t ax
credit for t he purchase o f non-group health insurance b y i ndividuals
not covered b y M edi care o r em p l o ym ent -based h eal t h i n surance.
Under t erms specified by Gruber, the credit could not ex ceed $500
for s ingl e filers and $1,000 for j oint and h ead-of-household f i l e rs;
and i t phased out for s ingl e filers wit h adjusted gross i ncomes
between $45,000 and $60,000 and for joint and head-of-household
filers with adjusted gross i ncomes between $75,000 and $100,000.
The cost p er newly i nsured person was $2,239 (1999 dollars).
! The l east cost-effective s ubsidy was a refundable t ax credit for out-
of-pocket h ealth insurance ex p enditures b y i ndividuals not covered
by Medic a r e : the cost p er newly i nsured person was $5,003 (1999
dollars).
Gruber’s Findings on Effectiveness. 39
! The m ost effective t ax s ubsidy was a refundable t ax credit for all
out-of-pocket h ealth insurance ex p enditures b y i ndividu a l s not
covered b y M edicare t hat was limited t o $1,000 for s ingl e filers and
$2,000 for j oint and h ead-of-hou s e h o l d filers: i t reduced the
uninsured population by 12.4 million.
! Running a close second was a refundable t ax credit for t he purchase
of non-group health insurance by i ndivid u a l s not covered b y
Medi care o r em p l o ym ent -based h eal t h i n s u r a nce. The credi t w as
limited t o $2,000 for s ingl e filers and $4,000 for j oint and h ead-of-
household filers and was paid dir ectly to insurers when p r emium
paym ents were due. It reduced the uninsured popul ation by 12.1
million.
! The l east effective s ubsidy was an above-the-line t ax deduction for
the purchase o f non-group health insurance b y i n d i v i duals not
covered b y M edi care o r em p l o yment-based insurance: i t l o wered t he
uninsured population b y 250,000.
38 In a 1993 study of the r esponse of s elf-employed indivi duals to the creation of a partial
tax deduction f or their health insurance expenditures b y t he T ax Reform Act of 1986,
economists J onathan Gruber a nd J a me s Poterba found that from 1986 to 1989, between 15%
a n d 2 0 % o f the s elf-employed with income s below $20,000 and a bout 50% of those wit h
income s a bove $50,000 claime d t he deduction. Fr om these r esponse r ates, t hey c oncluded
that a l ack of awareness of t he presence of the s ubsidy undercut its effectiveness. See U.S.
Congress, House Committ e e o n W ays and M eans, Subcommittee on Health, Healthth st
Insurance Premium Deductions for t he Self-Employed, hearing, 104 Cong., 1 sess., J an.
39 Cost-effectiveness i n t his context denotes the r evenue cost per newly insured person.
Gruber’s Findings on Cost.
! The m ost costly subsidy was a refunda b l e t ax credit for out-of-
pocket h ealth insurance ex p enditures b y i ndividuals not covered b y
Medicare t hat was limited t o $1,000 for s ingl e filers and $2,000 for
joint and head-of-household filers: its net reven u e c o s t was $62.2
billion (1999 dollars).
! The l east costly subsidy was the above-the-line t ax deduction for the
same ex penditures: it carried a n et revenue cost o f $0.9 billion
(1999dollars).
Gruber’s Findings on Equity.
! The s ubsidy engendering the l argest share o f b enef its for l ow-
income hou s e h o l ds was a refundable t ax credit for t he purchase o f
non-group health insurance b y i ndividuals not covered b y M edicare
or employment-based health insurance. It was limited t o $1,000 for
single filers and $2,000 for j oint and h e a d -of-household filers and
ph a s ed out for s ingl e filers with adjusted gross i ncomes between
$18,000 and $25,000 and for joint and head-of-household filers with
adjusted gross i ncomes between $30,000 and $50,000. Households
with incomes b elow 200% of the federal poverty l e v e l in 1999
recei ved 69% of t h e n et revenue cost of t h e s ubsi d y.
! The s ubsidy granting t he lowest share of its benefits to low-income
households was a non-refundable t ax credit for t he purchase o f non-
group health insurance b y i n d i viduals not covered b y M edicare o r
employm e n t -based health insurance. It was limited t o $1,000 for
single filers and $2,000 for j oint and h ead-of-househol d f i l e rs.
Households with incomes b elow 200% of the federal poverty l evel
in 1999 received 22% of the n et revenue cost of the s ubsidy.
! On the whole, refundable t ax credits for t he purchase o f non-group
heal t h i n surance w ere o f m uch greater benefit t o l ow-income
hous eholds than either a non-refundable t ax credit or a t ax
deduction.
Gruber’s Findings on Employer Health Insurance Coverage.
! Only one subsidy engendered an i ncrease (6.6%) i n t he num b e r o f
individuals covered b y employment -based h e a l t h i n surance: a
refunda b l e t a x c r e d i t f o r o u t - o f - p o c k e t h e a l t h i n s u r a n c e ex p enditures
by all i ndividuals not covered b y Medi care. Those al ready covered
by em pl oyer pl ans were el i gi b l e for t he credi t .
! The s ubsidy l eading t o t he smallest decline (-0.9%) i n t he number o f
individuals covered b y employment-based i nsurance was a t ax
deduction for the purchase o f non-group health insurance.
! Two almost identical subsidies generated the l argest decline (-9.6%)
in the numb e r of i ndividuals covered b y employm en t-based
insurance: (1) a refundable t ax credit for t he purchase o f non-group
health insurance b y i ndividual s n o t covered b y M edi care o r
employment - b ased health insurance t hat was limited t o $2,000 for
single filers and $4,000 f o r j oint and h ead-of-household filers and
phased out for s ingl e filers with adjusted gr oss i ncomes between
$45,000 and $60,000 and for joint and head-of-household filers with
adjusted gross i ncomes between $75,000 and $100,000; (2) a version
of the s am e cred i t w h i ch could be pai d directly to insurers in
advance o f t he t ax year i n whi ch i t w as cl ai m ed.
Gruber’s Findings on Key Administrative I ssues.
! Avoiding mismatches b et ween the timing of health insurance
premium payments and the timing o f s ubsidy t ransfers to eligible
individuals with minimal or no household s avings had a sign ificant
i m p act on t h e effi cacy of t h e b ase case. The reduct i o n i n t he
uninsured population was 37% great e r w h en the b ase case was
simulated on t he assumption t hat s uch mismatches were prevented
than when it was s imulated without such an assumption.
Gruber 2001 Analysis
In a follow-up to his 2000 study, Gruber ex p lore d t h e p o t ential cost and
effect i v eness o f w hat h e v i ewed as t hree l i k el y d i rect i ons for t he use o f t ax pol i cy t o
enlarge t he insured population. 40 The d irections involved o ffering new t ax subsidies
to employers; offering new t ax subsidies t o employees; and offering new t ax subsides
for t h e p u r chase o f non-group insurance t o i ndividuals not eligible for public or
em pl oyer-provi ded i n s u r a n c e c o v e r a ge . G r u b e r a s s e s s e d a n d com p ared t h e effi ci ency
– as m eas ured by the estimated revenue cost per newly insured i ndividual – of thes e
basic options. His key findings are summariz ed b elow.
New S ubsidies to E mploye rs.
As Gruber h as not ed, m any fi rm s al ready o ffer h eal t h i n surance t o em p l o yees.
So an important consideration i n devising new tax subsidies for em ployers t o ex pand
health insurance coverage is which firms would be eligible. If t he subsidies are not
narrowly o r p roperly t argeted, the cost per newly i nsured individual could quickly
reach unsust ai n abl y hi gh l evel s . Avai l abl e d at a o n t he l i nks bet ween t h e uni nsured
a n d t h e workpl ace suggest a rel at i v el y s i m p l e and effi ci ent s ol ut i on: offeri ng t h e
larges t t ax subsidy t o t he smallest firm s p aying t he lowest wages and phasing it out
as fi rm si z es and average w ages ri se.
Gruber s imulated t he effect s of j ust s uch an approach. He assumed a max imum
tax s ubsidy rate of 0.4 for firms with 10 or fewer e m p l o ye e s a n d average annual
wages o f $10,000 or less. Such a rate was equivalent to a t ax credit for 40% of the
cost to the firm o f p roviding health i n s u r a n c e t o each employee. In addition, he
assumed t hat t he amount of the s ubsidy was reduced by 2.5% for every addit i o nal
40 See Gruber, Taxes and Health Insurance.
em pl oyee beyond 10, so t h at i t reached z ero at 50 em pl o ye e s ; a n d t h at i t was al s o
reduced by any i ncrease i n average annual wages beyond $10,000. Gruber found that
such a t ax subsidy added 2.3 million i ndividuals to the ran k s of the non-elderly
insured at an annual cost o f $4.7 billion, or $2,005 (2000 dollars) p er newly i nsured
individual.41 He also found that by increasing t he subsidy rate t o 0.6, the annual cost
rose by 70% (to $8 billion per year) but coverage ex panded by 50%, res ulting i n a
high er cost per n ewly insured i ndividual: $2,300.42
Ne w S ubsidies to E mployees.
Another policy option for ex panding h ealth insurance coverage among the non-
elderly analyz ed b y Gruber consisted o f a new t ax subsidy t o employees to take up
coverage offered by employers . H e d et ermined that it would have little effect on
coverage rates and would be r el at i v el y i neffi ci ent . 43 There were t wo reasons for t hi s
outcome. First, only 5 % o f t hose who are o ffered i nsurance by employers were
un i n s u r e d . Second, the d ecision to take up coverage was i nsensitive t o changes in
its tax p rice. Gruber d id note t hat t his option o ffered t he advantage o f t argeting l ow-
income workers; but it had t he disadvantage o f not being able t o t a rget l ow-wage
firms, as many low-wage work e r s are empl oyed by firms t hat pay relativel y high
average w ages and al ready o ffer i nsurance.
New S ubsidies for Non-Group Cove rage.
The final alternative analyzed by Gruber consisted of a new t ax subsidy for the
purchase o f non-group insurance b y i ndividuals not covered under publ ic or
em ployment-bas ed insurance. He identified t hree major hurdles lying i n t he path of
such an approach. One was t hat about half of uninsured households pay n o i ncome
tax es. This f a c t suggested that if the s ubsidy were a t ax credit, it should b e m ade
refundable. A s econd hurdle was that mo st uninsured households lack access t o t he
l i qui d asset s requi red t o p ay i n surance p rem i u m s before recei vi ng a t ax subsi d y s uch
as a refundable t ax credit. As a result, t h e e f f e c t i v e n e s s o f a tax s ubsidy would b e
greatly enhanced if it were made payable i n advance. Fi nally, non-group coverage
was costly; s o any tax s ubsid y w ould need to cover a large s hare of premiums if it
were to be high ly effective.
Gruber s imulated t he effect s of a tax s ubsidy s imilar t o what t he Bush
Administration p roposed in its budget re quest for FY2002. He assumed t hat t he
subsidy i nvolved a refundable c r e d i t o f $1,000 for s ingl e filers with incomes u p t o
$75,000, and $2,000 for joint and head-of-household filers with incomes up t o
$100,000. In addition, he assumed t hat i t was not payable i n advance. Swayed by
such a s u b s i d y, the uninsured population s hrank b y 4 million p ersons, and the
estimated cost p er newly i nsured came t o $3,300 (2000 dollars). 44 Gruber p erformed
another s imulation under t he assumption t hat t he credit was p ayable in advance and
41 Ibid., p. 29.
42 Ibid., p. 30.
43 Ibid., p. 31.
44 Ibid., p. 33.
found that the cost p er newly i nsured fe ll to about $2,500. Bo th estimates were well
above his estimated cost o f a new t ax subsidy for employers.
Gruber 2002 Analysis
Still another recent s tudy by Gruber i s worth considering, since i t focused on the
cost and effectiveness o f t he tax credit for health insurance coverage proposed by the
Bu sh Administration i n its budget request for FY2003. He presented h is findings i n
a s tatement made to the House W ays a n d Means C ommittee i n connection with a
hea ri n g o n t ax credi t s for ex p andi ng health insurance coverage it held on February
13, 2002. 45 The credit h ad the following features : i t was refundable, available only
for t he purchase o f non-group insu rance b y t hose not covered under public or
employer-provided h ealth insurance, payable i n advance, and limited t o $1,000 for
single individuals with adjusted gross i ncomes of up to $30,000 and t o $3,000 for
families with adjusted gross i ncomes of either $40,000 (if only one adult purchased
insurance) or $60,000 (if m ore t han one adult purchased insurance).
Gruber’s analys is again i nvolved t he use o f a micro-simulation m odel. This one
was similar in design to the one he employed in his 2000 analys is and t ook into
account what was known about the b ehavioral responses of non-elderly i ndividuals,
fi rm s, and i nsurance com p ani es t o changes i n t h e t ax pri ce o f h eal t h i n surance. One
of his assumptions w a s t h a t i ndividuals and families who purchased non-group
coverage in response t o t he credit paid average m arket p rices for t hat i nsurance.
He estimated t hat t he subsidy would have a t otal annu al cost of $5.2 billion
(2001 dollars) and result in a n et decline i n t he uninsured population of 1.9 million,
yi e l d i n g a cost p er newly i nsured individual o f n early $2,800. 46 In addition, an
estimated 10.5 million i ndividuals took up the new credit, 3.3 million of whom were
previously uninsured. Nearly 2.5 million of t he individuals taking up the credit had
been covered by employm ent-based coverage. Of t hes e , 1.5 million persons
voluntarily switched t o non-group coverage because they found it to be a b etter d eal
with the credit , a n d 1.0 million persons involuntarily switched b ecause their
employers d ropped h ealth insurance coverage in response t o t he credit. Some of the
individuals who l ost employer-provided i nsurance ended u p uninsur ed, which
ex plained why the n et decrease i n t he uninsured population was 1.9 million p ersons
and not 3.3 million . For t hese newly uninsured, t he cost of non-group coverage
proved t oo high even with the credit.
Some analys ts have contended t hat the ado p t i o n o f a tax credit for health
insurance coverage modeled o n t he President’s p roposal but somewhat more
generous would t rigger a majo r overhaul o f t he non-group market, l eading t o t he
em ergence o f n ew l o w-cost pl ans w i t h fewer gaps i n coverage and l e s s p rem i u m
45 U.S. Congress, House Ways a nd Means Committee, Health Care Tax Credits to Decrease
the Number of Uninsured, hearing, 107th Cong., 2nd sess., Feb. 13, 2002 (Washington: GPO,
46 Ibid., p. 125.
variation b y age and geographic l ocation. 47 To ex pl ore t he cost and effi cacy of pl ans
such as t h ese, Gruber p erform ed anot her t wo simulations. In one, he assumed t hat
half of all p ersons buying non-group coverage with the credit were able t o d o s o at
prices 25% below average market prices; a nd in the other, he assumed t hat t he same
group of individuals were able to obta i n c o v erage at p rices 50% below average
m a r k et prices. His results suggested that the greater the d ecrease i n non- gr o u p
premiums under t he credit, the l arger t he decline i n t he uninsured population and the
gr eat er t h e d i s pl acem ent o f t hose p revi ousl y covered b y em p l o ym ent -b a s e d
i n s u rance. 48 Specifically, with a 25% reduction i n non-group premiums, t he total cost
per year was $5.4 billion (2001 dollars), and t he uninsured popul at i o n fell b y 2.2
million persons at a cost per n e w l y i n s u r ed of $2,503. By contrast, with a 50%
reduction i n p remiums, t he total annual cos t came t o $6.0 billion (2001 dollars), and
the uninsured population d eclined by 3.6 million p ersons at a cost p er newly i nsured
of$1,663.
Conclusions
A cent ral chal l enge for pol i cym akers i nt erest ed i n u si ng t ax pol i cy t o achi eve
a s u b s t a n tial and lasting reduction i n t he uninsured population i s t o d evise a t a x
subsidy t h a t c a n accomplish t his goal without having undesirable effects o n t he
allocation of economic resources , t he distribution of disposable income am ong
households, and the cost o f t ax administrat ion. Generally, t ax policy can and does
affect t h e p ri ce and t erm s of heal t h i n surance coverage. T he m at eri al ex am i n ed here
has important implications for t he use o f t ax policy t o ex p and t his co v e r age. Its
significance can be illuminat ed in a s eries of questions focusing on fact ors t o
consider in designing t ax subsidies t o ex pand health insurance coverage.
! Is the s ubsidy a tax deduction or a tax cr e d i t ? Many uninsured
individuals reside in low-income households. Gruber h as estimated
that up to 90% of uninsured tax payers with tax liabilities bel onged
to the 15% income tax b racket in 1997. Tax d eductions become
m o re val u abl e as a t ax payer’s m argi n al t ax rat e i ncrease s . B ut
generally, one dollar of a tax deduction h as a l ower revenue c o st
than one dollar of a tax credit.
! I f the s ubsidy i s a tax credit, i s i t ref undable? Gruber h as a l s o
estimated that 50% of uninsured taxpayers have no tax liability –
though t he proportion m ay be smaller t oday o wing to the reduction
in marginal tax rates and t h e creation o f 10% bracket under t he
Economic Growth and Tax Relief R e c o n c i l i ation Act of 2001 and
the i ncrease i n wages an d s alaries s ince 1997, the year to which
Gruber’s estimate applies. Non-refundable t ax credits are v aluable
47 For a discussion of policy debate ove r t his possible outcome , s ee Mark V . Pauly a nd Len
M. Nichols, “T he Nongroup Health Insurance M arke t: Short on Facts, Long on Opinions
and Policy Disputes,” He a l t h Affairs, Web exclusive [ http://www.healthaffairs.org] , Oct.
48 Gruber, Health Care Tax Credits to Decrease the Number of Uninsured, pp. 127-128.
only i f a tax payer has a tax liability agai nst which it can be cl aimed.
Refundable t ax credits, b y contrast , are valuable to all l ow-income
t ax p ayers b ecause t h ey can be cl ai m ed r egar d l e s s o f w het h er t h ere
is a current tax liability.
! If th e s u b s i d y i s a ref u n d a b l e t a x cred i t, i s i t f i x ed or
proportional in amount? Anot her fact or shapi n g t he effect i v eness
of a t ax subsidy t o ex p and h ealth insurance coverage is the s hare of
premiums it covers. As m any uninsured individuals reside in low-
income households, t he greater the s hare of individual and family
premiums met by t he subsidy, the great er the r at e at which the
subsidy i s likely t o b e t aken up by the uninsured. Fix ed refundable
tax credits may cover s maller and smaller portions of premiums as
health care costs rise over time. In addition, premiums in the non-
group market vary by age, sex , and geographic l ocation, and i n t he
group market by siz e of employer or risk pool. And premiums in the
non-group market tend to be lower for young adults in good health
and higher for older adults in poor health.
! Who is eligibl e for the subsidy? R ecent s t udi es suggest t h at t h e
m o st cost -effect i v e s ubsi d i es are apt t o b e ones whi ch are n arrowl y
targeted at groups with large p roportions of u n i nsured individuals.
Tax subsidies targeted at low-income households headed by
someone not covered b y employer-p rovided o r public insurance are
likely t o h ave a l ower revenue cost per n ewly insured i ndividual t han
subsi d i es wi t h a b roader reach, s uch as a t ax d educt i o n for al l out -of-
pocket ex p enditures fo r h e a l t h insurance. Bu t t here is some
evidence that targeting firms in the hope of induc i n g t h e m t o o ffer
c overage to employees may b e m ore effective t han t argetin g
individuals who l ack coverage. Gruber has es timated t hat over
three-quarters o f uninsured worker s are not offered h ealth insurance
by their employers. In a d dition, eligibility matters because of the
d i f f erent ways that firms, individuals, and private i nsurers c a n
respond to tax s ubsidies for health insurance.
! Does th e s u b s i d y a d d ress k ey ad mi n i strati ve i ssu es? Yet another
factor affecting t he cost and efficacy of proposed tax s ubsi d i es t o
ex pand health insurance coverage is administrative i ssues that could
impede t heir smooth implementation. At leas t t hree can be
mentioned here. One concern i s possible mismatches b et w een the
timing of health insurance premium payments and the timi n g o f
subsi d y t ransfers t o el i gi b l e i ndi vi dual s . Because rel at i v el y few
uninsured households appear to have sufficient liquid assets to draw
upon in payi ng health insurance premiums, m aking t he subsidies
payable i n advance of t he due date for filing t ax returns m ay enhance
t h ei r effi cacy. A rel at ed concern i s t he rat e at whi c h e l i gi bl e
individuals might take up tax s ubsidies t hat are payable i n advance.
E x peri ence wi t h t h e federal earned i ncom e t ax credi t suggest s t h a t
l o w-i n com e t ax p ayers are rel u ct ant t o cl ai m advanceabl e t ax credi t s
because many fear they will end u p facing a t ax liability when they
file their tax ret u rn s . 49 Fi nally, t here is some concern t hat t he IR S
l acks t he resources t o effi ci ent l y adm i n i s t er any advanceabl e t a x
subsidy for health insurance coverage.50 Itshouldalsobekeptin
m i n d t hat t here appear t o b e si gn i fi cant t rade-offs bet w een ease o f
administration and preci sion of targeting i n t ax subsidies t o ex pand
health insurance coverage.
! Does th e s u b s i d y a d d ress b arri ers to cov e r age of h i gh -ri s k
groups in the p rivate i nsurance market? Persons with relativel y
high risks of certain health problem s are likel y t o find i t difficult to
obtain affordabl e and adequate coverage in the non-group market.
Premiums vary widely by age, sex , and region. And i n m any s tates,
insurers are free t o ex clude pre-ex iting conditions from coverage or
to severely restrict coverage for t hese conditions. These
underwriting p ractices raise t he question o f whether a t ax subsidy t o
ex pand health insurance coverage should i nclude regu lations aimed
at curbi n g t hese pract i ces.
! What is th e u l t imate p olicy objective of the tax subsidy? The
design of cost-effective t ax subsidies t o ex pand health insurance
coverage al so depends in part on the ultimate policy objective. If the
m ai n go al i s t o i m p rove equi t y i n access t o federal t ax s ubsi d i es for
health insurance, a t ax deduction for personal ex p enditures o n h ealth
insurance by t hose i neligible for ex i sting s ubsidies m ay suffice. Or
if the m ain goal i s t o reduce s ubstantially the uninsured population,
then a proportional, advan ceable, refundable t ax credit target ed at
uninsured workers not offe red h ea lth insurance coverage by their
em pl oyers m ay suffi ce. But i f t he m ai n go al i s t o achi eve uni versal
coverage, t hen i t m ay be necessary to couple a s ubsidy s uch a s a
refundable, advanceable, m eans-tested tax credit with reforms o f t he
privat e health insurance m arket and a requirement t h a t a l l
individuals purchase h ealth insurance o r all firms p rovide h e alth
insurance t o t heir employees.
49 Gruber, Taxes and Health Insurance,p.32.
50 See George Guttman, “ Another IRS Burden: T he New Health Insurance Cr e d it,” Tax
Not e s , M arch 3, 2003, pp. 1322-1324.