Title: The Effect of an Employer Health Insurance Mandate on Insurance Coverage and Labor Market Outcomes:
1The Effect of an Employer Health Insurance
Mandate on Insurance Coverage and Labor Market
Outcomes Evidence from Hawaii
- Tom Buchmueller
- Ross School of Business, University of Michigan
- John DiNardo
- Ford School of Public Policy, University of
Michigan - Robert Valletta
- Federal Reserve Bank of San Francisco
2Background and MotivationEmployer-Sponsored
Insurance
- Employer-sponsored insurance (ESI) is a
cornerstone of the US health care system. - Roughly 90 of private coverage is through the
workplace. - ESI coverage has been declining gradually for
over 2 decades. - Policy Question Should health care reforms build
on or reduce the role of employers?
3Background and Motivation Employer Mandates
- One approach mandate employers to prove health
insurance. - Proposals by Presidential Candidates
- Edwards Require employers to provide or help
finance health insurance for employees. - Obama Require employers to offer meaningful
coverage or contribute a percentage of payroll
toward the costs of the public plan. - Richardson Employers would be required to
provide insurance or help finance coverage for
their employees. - Clinton Large employers must provide an employee
plan or contribute to the cost of coverage. - source www.health08.org
4Background and MotivationPrevious Mandate Bills
- Comprehensive health insurance is an idea whose
time has come in America. Let us act now to
assure all Americans financial access to high
quality medical care. - -- Richard Nixon, whose 1974 Comprehensive
Health Insurance Plan included an employer
mandate. - Other prominent mandate legislation and
proposals - Massachusetts (1988)
- Oregon (1989)
- Washington (1993)
- Clinton Plan (1994)
- Californias SB 2 (2002)
- None of these have been enacted
5Background and MotivationPrevious Mandate Bills
- The only employer health insurance mandate ever
enacted is Hawaiis Prepaid Health Care Act
(PHCA). - Received considerable attention in the health
care reform debate of the early 1990s. - Some saw it as a model to be emulated.
- Critics questioned the impact on coverage and
claimed it would distort labor markets and reduce
employment. - But it has received very little attention from
researchers. - Little evidence on coverage or labor market
effects.
6Hawaiis PHCA Timing Issues
- Enacted in 1974, the same year as Federal
legislation (ERISA) that prohibits state coverage
mandates. - PHCA went into effect 1/75, but was soon
challenged in court. - Legal battle made it to the US Supreme Court,
which in 1981 ruled that the PHCA was
unconstitutional. - Congress granted PHCA an ERISA exemption in 1983.
- What is t0?
7Hawaiis PHCA Data Limitation
- Hawaii is not separately identified in the
Current Population Survey (CPS) until 1977. - Health insurance is not measured in a national
survey until March 1980 (with information on
coverage in 1979) - If t0 1975, it is not possible to do a pre/post
analysis. - If t0 1983, this is possible.
8Timing and Data Issues Implications for Research
- Despite these problems, we think it is possible
to learn from Hawaiis experience with the PHCA. - Coverage data suggests that the ESI rate in
Hawaii and the rest of the US began to diverge
around 1983. - Suggests that t0 1983
- Costs of the mandate have increased over time as
health insurance premiums have risen in real
terms. - Even if t0 1975, a post/more post analysis is
of interest
9Overview
- Economics of employer benefit mandates
- Hawaiis PHCA
- Previous Research
- Data and Empirical Strategy
- Preliminary Results
- Tentative Conclusions and Next Steps
10The Economics of ESI
- Employer provision of health insurance can be
explained by - Preferential tax treatment of employer premium
payments - Economies of scale from group purchase
- Reduced loading charges due to risk pooling
- These factors lower the cost of ESI relative to
purchasing insurance in the non-group market. - Let B employer cost of health insurance aB
employee valuation of those benefits. - If a 1, workers are willing to accept lower
wages in return for health insurance benefits. - On average, we would expect a lt 1 for workers
without ESI. - (for low wage workers, min. wage may prevent
offset even if a 1.)
11The Economics of an Employer Mandate
S
S
wage
B
B
If a 1
Full wage offset No change in employment
B
D
D
of workers
12The Economics of an Employer Mandate
S
S
aB
wage
B
If a lt 1
Less than full wage offset Decline in covered
employment
ltB
D
D
of workers
13The Economics of an Employer Mandate
S
S
wage
If minimum wage binds
Min wage
Larger decline in employment
D
D
of workers
14Empirical Implications
- Effect of a mandate should be strongest on
workers who are least likely to receive ESI in a
voluntary market. - Low skill workers
- Younger workers
- minorities
- Employees of small firms.
- If wages of these workers cannot fully adjust, we
would expect covered employment to fall. Either - Employment ?
- Employment in exempt categories ?
15Hawaiis PHCA Coverage
- Covers all private sector employers (no firm size
exemption) - Employers must provide insurance for all workers
except - Part-time (less than 20 hrs/week)
- New hires (employed less than 4 consecutive
weeks) - Very low wage (86.67 x min wage)
- Seasonal employees
- Paid by commission
- Employees may decline if they have coverage from
another source or for religious reasons.
16Hawaiis PHCA Benefits and Cost Sharing
- Minimum benefit package
- Inpatient outpatient hospital
- Physician services
- Lab X-ray
- Maternity services
- Employee contribution for single coverage capped
at 50 of premium or 1.5 of monthly gross wage.
17Previous Research Hawaiis PHCA
- Effects on Insurance Coverage
- Dick (1994)
- uses March CPS for 1986-1988 to analyze
uninsured - Cross-tabs and unreported regressions
- Results very small diff in uninsured rate
between Hawaii US - Claims this diff is explained by Xs
-
- Thurston (1997)
- cites 1969 data hospital insurance coverage
88 for Hawaii, 87 for US - March CPS, 1990-93 ESI rate for workers 71 for
Hawaii, 59 for US - Gap varies by industry
18Previous Research Hawaiis PHCA
- Effects on Insurance Coverage
- Kronick et al. (2004)
- March CPS for 1996-2002
- Document differences in ESI coverage across
states - Hawaiis ESI rate is 10 above rate predicted by
regression - Lee et al (2005)
- March CPS for 1994-2004
- Also find a significantly higher rate of ESI for
Hawaii
19Kronick et als Results
Hawaii
Michigan
20Previous Research Hawaiis PHCA
- Effects on Wages
- Thurston (1997) uses aggregated Census data (1970
1990) to test for differential wage growth
between industries that should have been more or
less affected by mandate. - Key explanatory variable ESI coverage in
Hawaii in 1990 less in US - Within Hawaii, wages grew less in industries
where gap was larger. - Consistent with a wage offset
- But offset disappears when other states used as
within-industry controls
21Previous Research Hawaiis PHCA
- Effects on Hours
- Thurston (1997) uses same Census data (1970
1990) - Dependent variable D low hours Hawaii D
low hours US - Explanatory variable ESI coverage in Hawaii -
ESI coverage US - Results imply an increase in low hour employment
in Hawaii - Lee et al (2005)
- Matched cross-tab no difference in the working
lt 20 hrs/week - CDF suggests spike at 20
- No evidence on employment effects.
22Data Current Population Survey, 1979-2006
- March Annual Demographic Files
- Insurance for prior year
- Wages for longest job held last year
- Employment, hours last week
- Demographic controls
- Merged Outgoing Rotation Group (MORG) Files
- Larger samples
- Better wage variable
- Employment, hours last week
- Demographic controls
-
23Testing for Heterogeneous Treatment Effects
- Simple approach stratify by education
- High school or less vs. College or more
- Adapted propensity score approach (next
iteration) - Use March data from rest of US
- Predict logit of own name ESI
- Use fitted values to calculate propensity of
obtaining ESI in full sample - Stratify analysis by predicted propensity
24Hypotheses
- Effects of mandate should be strongest for
workers who are least likely to receive ESI in a
voluntary market. - Little reason to expect an effect on workers who
have high rates of ESI coverage in the rest of
the US. - Cost of mandate has increased over time as health
care costs have grown faster than inflation and
total compensation. - Look for divergence between Hawaii and the rest
of the US.
25Single Coverage Premiums in Hawaii
(2006)1974-2007
Notes Premium data are from HMSA, Hawaiis
largest insurer. Single coverage premium amounts
are converted to an hourly amount assuming a
full-time schedule of 40 hours/week.
26Results Coverage
- Trends in ESI Coverage, 1979 to 2005
- All Private Sector Workers
- By Education (HS or less, College or more)
- Distribution of Coverage, 2002-2005
27(No Transcript)
28(No Transcript)
29(No Transcript)
30Table 1. The Distribution of Insurance Coverage
2002-2005
31Table 1. The Distribution of Insurance Coverage
2002-2005
32Analyzing Labor Market Outcomes
- Outcomes
- For private sector employees
- Log wages
- Short hours (usual hours lt 20/week)
- For all working age adults
- Employment
- Testing for a divergent trend in adjusted
outcomes - Run Yi Xib g Hawaii St dt Yeart St qt
Hawaii x Yeart - Collect qs
- Run weighted least squares regression of q on t
33DlnW .074 .002 x Hawaii
(.026) (.001)
34DlnW .137 .003 x Hawaii
(.026) (.001)
35DlnW -.009 .0023 x Hawaii
(.028) (.0016)
36Effect on Hours Worked
- Unadjusted Trends in Percent Low Hour Part-time
(lt20/week) - Results from OLS regressions
- Logit coefficients have similar pattern, but are
harder to interpret
37(No Transcript)
38DPT -.012 .0006 x Hawaii
(.004) (.0002)
39DPT -.013 .0007 x Hawaii
(.003) (.0002)
40DPT -.005 .0002 x Hawaii
(.003) (.0002)
41Employment Effects
- Unadjusted Trends
- OLS Regressions
42(No Transcript)
43DEmployed .059 .0006 x Hawaii
(.007) (.0004)
44DEmployed .064 .0013 x Hawaii
(.010) (.0013)
45DEmployed .040 .0001 x Hawaii
(.009) (.0005)
46Conclusions Coverage
- ESI rates in Hawaii and rest of US began to
diverge around 1982. - Effect of mandate was to limit erosion of
coverage. - Gap in overall coverage rate has been relatively
constant at around 10-12 points since the early
1990s. - Larger effect on workers that are least likely to
receive ESI. - Effect on ESI is not significantly offset by
reductions in other sources of coverage.
47Tentative Conclusions Wages
- Some evidence of a lower wage growth in Hawaii
since 1980 - Consistent with wage offset
- But timing is suspect
- (Regressions using March data yield similar
results for less educated workers, no effect for
highly educated workers)
48Tentative Conclusions Hours and Employment
- Maybe some evidence of shift to exempt
part-time workers - Significant for full sample and less educated
workers - But, again, timing is weird
- Magnitude is very small
- Little evidence of an effect on employment
49Next Steps
- Coverage
- Permutation tests definitively show that Hawaii
vs. US. diff is not driven by Xs - Labor Market Outcomes
- More parameterized models
- Control for industry, occupation
- Additional tests?
- Standard errors