Medical%20Expenditure%20Burdens:%20The%20Impact%20of%20Tax%20Subsidies,%20Within-Year%20Expenditure%20Concentration,%20and%20More

1
Medical Expenditure Burdens The Impact of Tax
Subsidies, Within-Year Expenditure Concentration,
and More

• Thomas M. Selden, Ph.D.
• Division of Modeling Simulation
• Center for Financing, Access and Cost Trends

2
What is a High Burden?

• Out-of-pocket expenditures
• ------------------------------------- gt
  threshold (e.g., 20)
• Resources (income)
• Various terminology
• Catastrophic expenditures
• High expenditure cases
• High expenditures relative to income
• High burdens

3
Why Study Burdens?

• Most families healthy in given year
• Illness, however, is a fact of life
• Medical bills pour in just when families must
  grapple with illness, seeking care, and
  possibility reduced earnings
• Accurate burden measurement a key ingredient for
  sound public policy

4
Overview of Talk

• Data
• 2 refinements to conventional results
• Impact of tax subsidies
• Intra-year burdens
• Neither previously studied
• Burden of uncompensated care (time permitting)

5
Data

• Medical Expenditure Panel Survey
• Sponsored by AHRQ NCHS
• 30K persons each year
• Civilian noninstitutionalized population
• Narrow family definition
• lt65 population
• MCR Part D likely to affect senior burdens
• Simulated tax subsidies
• Edited event and job data

6
Conventional Results 10 and 20 Burden
Frequency (OOPMEDOOPPREM) by Poverty Level, 0203
7
Impact of Tax Subsidy
8
Tax Subsidies

• Itemized deduction for out-of-pocket spending
• ESI premiums excluded from federal income,
  federal payroll, and state income taxes
• Retiree ESI coverage exemption
• Sales tax exemption
• Self-employment premium deduction from federal
  and most state income taxes

9
Tax Subsidies

• Tax subsidies total 16.3 of all spending on
  health care (across all ages)
• Nearly TWICE this share of private spending
• Ignoring subsidies leads to overstated burdens
• How large is effect?
• Whose burdens are reduced?

10
Family Tax Subsidies on OOPMED and OOPPREM, by
Poverty Level, 0203
Subsidy Rate
Average Subsidy
11
Impact of Tax Subsidy on 20 Burden Frequency
(OOPMEDOOPPREM) by Poverty Level, 20022003
12
Impact of Tax Subsidy on OOPMEDOOPPREM Burdens gt
20, by Poverty Level, 0203
13
Bias from Ignoring Subsidies

• Conventional studies ignore subsidies
• But reported spending is de facto net of sales
  taxes that would have been paid absent the
  exemption
• Thus, conventional measures are between
  pre-subsidy and post-subsidy measures
• Can one safely ignore subsidies?

14
Conventional vs. Subsidy-Adjusted 20 Burden
Frequency (OOPMEDOOPPREM) by Poverty, 0203
I.e., based on published MEPS data
15
Expanding Burden Definition to Include Wage
Offset from Employer Contributions Family Tax
Subsidies, by Poverty, 0203
Subsidy Rate
Average Subsidy
16
Expanding Burden Defn to Include Wage Offset from
Employer Contributions Impact of Tax Subsidy on
20 Burden Frequency, by Poverty, 0203
17
Expanding Burden Defn to Include Wage Offset from
Employer Contributions Impact of Tax Subsidy on
Burdens gt 20, by Poverty Level, 0203
18
Summary

• Tax subsidies modestly reduce out-of-pocket
  burdens
• Larger impact on burdens inclusive of cash wage
  offsets for employer contributions
• Little benefit for poor
• Middle income groups benefit more
• Conventional OOP burden estimates that ignore
  subsidies are fairly accurate
• OOP spending measured net of sales tax exemption
  (hence conventional measures already post-subsidy
  to a degree)

19
Do Annual 20 Burdens Tell the Whole Story? The
Impact of Within-Year Expenditure Concentration
20
20 Burden Frequency vs. Self-Reported Bill
Problems
Adults Age 19-64 in 2004
28
7.7
21
Bill Problem Responses Might Reflect Lower
Thresholds, Adults 2004
CMWF
22
Within-Year Burdens?

• Precautionary savings are low
• 24 of bottom quintile have no liquid assets
• Median among those with assets600
• Expenditures HIGHLY concentrated within year
• Earnings dip when expenditures spike?

23
Within-Year Family Expenditure Concentration,
20032004
Note Families with zero expenditures excluded
24
Peak Month as Percentage of Annual Total, by
Poverty and Expenditure
25
Quarterly vs Monthly Concentration

• Peak out-of-pocket month 44.5
• Peak out-of-pocket quarter 60.5
• For poor 76.2 of OOP occurs in a single quarter

26
Does Income Fall as Medical Expenditures Peak?

• Among poor families
• P(?Y lt -.333) in peak month 23.6
• P(?Y gt .333) in peak month 10.7
• P(?Y lt -.333) in low month 16.6
• P(?Y gt .333) in low month 15.6
• Diff-in-Diff 12.1 (plt.1)

Calculated for families with 1000 in earnings
and 5000 in medical expenditures half of which
occurred in single month
27
Does Income Fall as Medical Expenditures Peak?

• Among poor families
• P(?Y lt -.333) in peak quarter 19.7
• P(?Y gt .333) in peak quarter 9.6
• P(?Y lt -.333) in low quarter 11.5
• P(?Y gt .333) in low quarter 17.5
• Diff-in-Diff 16.1 (plt.05)

Calculated for families with 1000 in earnings
and 5000 in medical expenditures half of which
occurred in single month
28
20 Burdens among Nonelderlyby Poverty Level,
0304
43.6
34.6
27.0
21.5
16.7
6.9
1.1
29
Spending Distribution in High-Burden Month by
Poverty, 0304
Conditional on having 20 monthly burden
30
20 Burdens by Family Coverage, 0304
31
Spending in High-Burden Month by Family Coverage,
0304
Conditional on having 20 monthly burden
32
Within-Year Burden Conclusions

• gt40 poor have high monthly burden
• Income dips play small role
• Monthly burden rate for poor 41.3 vs 43.6
• Spikes in expenditures more important
• RX plays key role for poor and families with
  public coverage
• Monthly measure for adults in 2003 only slightly
  less than bill problem frequency (26.9 vs. 28)

33
Burden of Uncompensated Care
34
Uncompensated Care

• CWF bill problem includes inability to pay
• UC can indeed be burdensome
• Medical debt
• Credit problems
• Access problems
• Stigma
• All ignored by conventional burden analyses
• Not observed, but
• WTP(avoid UC burden) lt UC

35
Bounding 20 Annual Burdens for Uncompensated Care
36
Monthly Burdens (20)
37
UC Conclusions

• Modest increase in prevalence
• Concentrated among poor and pub/unin
• Families pay until it hurts
• Importance of measuring medical debt
• Monthly UC-adjusted burdens approximately same as
  bill problem frequency

38
Conclusions

• Tax subsidies modestly reduce burdens
• Little goes to poor
• Narrowing budget window from annual to
  quarter/month greatly increases burden
  prevalence, especially among poor
• Conventional measures relatively robust to
  inclusion/exclusion of uncompensated care
• More detailed analysis may help solve discrepancy
  between burden vs bill problem frequencies