View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

May 2019, EB19-05

Economic Brief

Lifetime Medical Spending of Retirees
By John Bailey Jones and Aaron Steelman

Retirees face considerable medical expenses during their remaining lives.
Model simulations suggest that although a large amount of that spending
can be predicted — based on attributes such as income, health, and marital
status — there remains significant dispersion. Households with heads who
turned seventy in 1992 will incur $122,000 in medical spending on average,
including out-of-pocket expenditures and Medicaid payments. But the top
5 percent of households will incur more than $300,000 in such spending.
The level and dispersion of this spending diminish only slowly with age.
Most elderly Americans face the risk of catastrophic health care expenses despite near
universal enrollment in the Medicare program.
There are many gaps in Medicare coverage,
such as long hospital and nursing home stays.
Moreover, Medicare requires copayments for
many medical goods and services. Affluent individuals have reported being worried about
rising health care costs more than any other
financial issue.1
Although there is a large literature documenting
annual medical spending at older ages, relatively
little work has been done documenting cumulative lifetime spending, particularly the distribution of that spending. Yet in many ways, lifetime
totals are most important to savings decisions
and household well-being because forwardlooking individuals base their consumption less
on current income than on the average income
they expect to receive over their lifetimes. The
same logic applies to medical spending. Households care not only about the risk of catastrophic
expenses in a single year, but also the risk of

EB19-05 – Federal Reserve Bank of Richmond

moderate but persistent expenses than can build
into catastrophic lifetime costs.
In a recent paper, John Bailey Jones and Justin
Kirschner of the Richmond Fed, Mariacristina De
Nardi of the Minneapolis Fed, and Eric French
and Rory McGee of University College London
estimate the distribution of lifetime medical
spending for retired households whose heads
are seventy or older.2 They focus on out-of-pocket
spending, but because high out-of-pocket
expenses can leave households indigent and
dependent on Medicaid, they also include Medicaid payments in their benchmark spending
estimates. In accounting terms, the benchmark
estimates measure the medical spending not
covered by Medicare or supplemental private
insurance, although they do include Medicare
and supplemental private insurance premia.
In economic terms, the benchmark estimates
measure the medical spending risk that households of all wealth levels would face were Medicaid not available (absent any other changes in
their insurance).

Page 1

Figure 1: Unconditional Distribution of Annual and
Lifetime Medical
Annual Expenditures
Spending of Surviving Households
625
625

All Households,
Following
HRS Distribu�on
Annual
Spending of Surviving
Households
(Percentiles)

125
125

2014 Dollars (000s)

2014 Dollars (000s)

250

50

2525
10

55
2

11

72
72

76
76

Mean

Mean

625
625

80
80

84
84

Age
Age

88
88

Chart
50thTitle

92
92

96
96

90th

Q_50

100
100

95th

Q_90

Q_95

Annual Spending in the Period before Death (Percentiles)

250

2014 Dollars (000s)

125
125
50

2525
10

55
2

11

72
72

76
76

Mean

Mean

625
625

80
80

84
84

88

92
92

Age 88

Chart
50thTitle
Q_50

90th

Q_90

96
96

100
100

95th

Q_95

Lifetime Spending of Surviving Households (Percentiles)

250

2014 Dollars (000s)

125
125
50

Data and Methodology
The authors use data from the Health and Retirement Study (HRS), which has high-quality information on out-of-pocket medical spending during the
period 1995–2014. Because the HRS does not have
Medicaid payment data, the authors impute Medicaid payments using the Medicare Current Beneficiary
Survey. Ideally, these data would permit the authors
to estimate medical spending directly by calculating
discounted sums of household spending histories.
But even though the HRS has a long panel dimension
for a survey of its type, it is not long enough to track
all seventy-year-olds through the ends of their lives.
The authors employ models instead.
They combine two models in their analysis, both
dynamic. The first is a model of health and mortality estimated from the HRS. The second is a model
of medical spending (out-of-pocket and Medicaid)
given health and household composition, estimated
using the HRS and the medical spending measures
described above. In addition to health and household composition, the second model allows medical
spending to depend on age, permanent income (PI),
and idiosyncratic shocks. Simulating the estimated
models permits the construction of household histories, the calculation of discounted sums, and ultimately the distribution of lifetime medical spending.3
Results
Figure 1 shows the implications of the authors’ model for the cross-sectional distribution of their preferred medical spending measure, the sum of costs
paid either out of pocket or by Medicaid, expressed
in 2014 dollars. Mean expenditures are shown, along
with the 50th, 90th, and 95th percentiles.

2525
10

55
2

11

72
72

76
76

Mean

80
80

84
84

50th

Age

88
88

92
92

90th

96
96

100
100

The top graph summarizes the health care expenditures of surviving households in annual terms. Medical expenses rise rapidly with age. For example, mean
medical spending rises from $5,100 per year at age
seventy to $29,700 at age 100. At the upper tail, the
95th percentile increases from $13,400 to $111,200.

95th

Mean
Q_50
Q_90
Q_95
Source: John Bailey Jones, et al. (2018)
Notes: All households following HRS distribution characteristics.
Data are graphed using a logarithmic scale.

The middle graph shows annual end-of-life costs,
including burial expenses. The results for age seventy-

Page 2

two describe the expenses incurred by households
who die between ages seventy-two and seventy-four.
On average, end-of-life medical expenses exceed
those of survivors. Mean end-of-life expenses range
from $11,000 at age seventy-two to $34,000 at age
100. For the 95th percentile, expenses range from
$35,000 at age seventy-two to $114,000 at age 100.
The bottom graph plots the authors’ variable of greatest interest, lifetime expenditures. At each age, they
calculate the present discounted value of remaining
medical expenditures from that age forward, using
an annual real discount rate of 3 percent. The lifetime
totals are considerable. At age seventy, households
will, on average, incur more than $122,000 of medical expenditures during the remainder of their lives.
The top 5 percent of spenders will incur expenses
greater than $330,000. A noteworthy finding is that
lifetime totals do not fall rapidly as households age

and approach the ends of their lives, as one might
expect. A couple at age ninety will, on average,
spend more than $113,000 before they die. The 95th
percentile of remaining lifetime spending is higher
at age ninety than at age seventy. The slow decline
of lifetime costs is due mostly to the tendency of
medical costs to rise with age. Households that live
to older ages have shorter remaining lives but higher
annual expenditures.
In total, the graphs in Figure 1 show that medical
costs of older households are high, rise with age, and
are widely dispersed.
Figures 2 and 3 show the means and 90th percentiles
of lifetime medical expenses for different values of PI
and initial health and marital status. Figure 2 shows
the results for households at the very bottom of the
income distribution (PI = 0). Lifetime spending varies

Figure 2: Distributions
of Lifetime
Medical Expenditures by Initial Health
and Household
PI = 0*
PI Rank
0, 90th State,
Percen�les
PI Rank
0, Means

Means

180
180
160
160

350
350

140
140

2014 Dollars (000s)
2014 Dollars (000s)

300
300

2014 Dollars (000s)

120
120

250
250

100
100

200
200

8080

150
150

6060

100
100

4040

5050

2020
00

90th Percentiles

400
400

smg smb smn

sfg

sfb

sfn

cgg

cbb

cgn

cng

cbn

cnb

smg smb smn sfg s� sfn cgg cbb cgn cng cbn cnb

Household State at Age 70
Household State at Age 70

00

smg smb smn

sfg

sfb

sfn

cgg

cbb

cgn

cng

cbn

cnb

smg smb smn sfg s� sfn cgg cbb cgn cng cbn cnb

smg = single male, good health
sfg = single female, good health
cgg = couple, both good health

smb = single male, bad health
sfb = single female, bad health
cbb = couple, both bad health

cng = couple, male in nursing home,
female in good health

cbn = couple, male in bad health,
female in nursing home

Household State at Age 70

smn = single male, nursing home
sfn = single female, nursing home
cgn = couple, male in good health,
female in nursing home
cnb = couple, male in nursing home,
female in bad health

Source: John Bailey Jones, et al. (2018)
* PI = 0 indicates the bottom percentile of the permanent income distribution.

Page 3

greatly across the distribution of initial health and
marital status. The following are some of the most
apparent trends. First, women have higher lifetime
medical expenditures than men. Second, people
who are initially in good health have higher lifetime
expenditures than those who are initially in bad
health. This result is due to their longer life expectancies in combination with the tendency of medical
costs to rise with age. Third, households in nursing homes have the highest lifetime expenditures,
despite their high rates of mortality, because of the
high cost of such care. Most people who are not in
nursing homes at age seventy never have extended
nursing home visits, however.
Figure 3 shows the results for households at the
very top of the income distribution (PI = 1). These
households spend considerably more than those at

the bottom, shown in Figure 2, often well in excess
of 50 percent more. For instance, consider a seventyyear-old couple with both members initially in
good health and a PI rank of 0. That couple would,
on average, spend $104,000 during their remaining
lives. With a PI rank of 1, they would spend more
than $165,000. Households with higher incomes
may have higher lifetime expenditures because
they live longer or because they have higher expenses at any given age.
Figure 4 (on the following page) compares lifetime
medical spending of couples in initial good health
with and without Medicaid payments. The top two
graphs show those at the bottom of the income
distribution. Medicaid covers, on average, 57 percent
of lifetime costs at age seventy. At older ages and
higher spending percentiles it covers even more. The

Figure 3: Distributions of Lifetime Medical Expenditures by Initial Health
and Household
State, PI = 1*
PI Rank
1, 90th Percen�les

PI Rank 1, Means
Means

300
300

200
200

400
400

2014 Dollars (000s)

500
500

2014 Dollars (000s)

250
250

300
300

150
150

200
200

100
100

100
100

5050
00

90th Percentiles

600
600

smg smb smn

sfg

sfb

sfn

cgg

cbb

cgn

cng

cbn

cnb

00

smg smb smn sfg s� sfn cgg cbb cgn cng cbn cnb

Household State at Age 70

smg smb smn

sfg

sfb

sfn

cgg

cbb

cgn

cng

cbn

cnb

smg smb smn sfg s� sfn cgg cbb cgn cng cbn cnb

smg = single male, good health
sfg = single female, good health
cgg = couple, both good health

smb = single male, bad health
sfb = single female, bad health
cbb = couple, both bad health

cng = couple, male in nursing home,
female in good health

cbn = couple, male in bad health,
female in nursing home

Household State at Age 70

smn = single male, nursing home
sfn = single female, nursing home
cgn = couple, male in good health,
female in nursing home
cnb = couple, male in nursing home,
female in bad health

Source: John Bailey Jones, et al. (2018)
* PI = 1 indicates the top percentile of the permanent income distribution.

Page 4

Discussion and Conclusions
The simulations by Jones, De Nardi, French, McGee,
and Kirschner show that lifetime medical spending
is high and uncertain — and that the level and dispersion of such spending diminish only slowly with
age. Although permanent income, initial health, and
initial marital status have large and predictable effects, much of the dispersion in lifetime spending is
due to events at older ages. The poorest households
have the majority of their medical expenses covered by Medicaid, which significantly reduces their

bottom two graphs show results for households at
the top of the income distribution. Medicaid covers, on average, 21 percent of lifetime costs at age
seventy and rises to nearly 30 percent at age 100.
While most high-income households do not receive
Medicaid, those that qualify do so under the “medically needy” provision, which assists households
whose financial resources have been exhausted by
medical expenses. Those households tend to have
high medical expenses and tend to receive large
Medicaid benefits.4

Figure 4: Lifetime Medical Expenses of Couples Initially in Good Health with and without Medicaid Payments

Lifetime OOP/Medicaid Spending of Surviving Households
Chart
Title
Couples with PI Rank
O and
Initially in Good Health

Lifetime OOP Spending of Surviving Households
Chart
Title
Couples with PI Rank
O and
Initially in Good Health

250

250

125
125

125
125

2014 Dollars (000s)

625
625

2014 Dollars (000s)

625
625

50

2525

10
55

11

50

2525

2

10
55

2
72
72

76
76

80
80

84
84

Age

Mean
Q_50

Mean

88
88

92
92

96
96

11

100
100

50th Percentile
Q_95

625
625

76
76

80
80

84
84

Age

88
88

92
92

96
96

100
100

95th Percentile

Lifetime OOP Spending of Surviving Households
Couples with PIChart
Rank 1Title
and Initially in Good Health
625
625
250
125
125

2014 Dollars (000s)

250

125
125
50

2525

50

2525

10

55
2

11

72
72

90th Percentile

Q_90

Lifetime OOP/Medicaid Spending of Surviving Households
Chart
Title
Couples with PI Rank
1 and
Initially in Good Health

2014 Dollars (000s)

*

10

55
2

72
72

76
76

80
80

Mean

84
84

Age

88
88

92
92

96
96

50th Percentile

100
100

11

72
72

76
76

90th Percentile

80
80

84
84

Age

88
88

92
92

96
96

100
100

95th Percentile

Source: John Bailey Jones, et al. (2018)
Notes: OOP stands for out of pocket. PI rank 0 indicates the bottom percentile and PI rank 1 indicates the top percentile
of the personal income distribution. Data are graphed using a logarithmic scale.

Page 5

spending volatility. Medicaid also reduces the level
and volatility of medical spending for high-income
households but to a much smaller extent.
The authors note a few caveats to their analysis.
Their research assumes, as do many other empirical
studies, that medical spending is exogenous when
in fact it is a choice variable. Although the demand
for some medical goods and services is extremely
inelastic, the demand for others might be quite
elastic. For instance, nursing home care is a bundle
of medical and nonmedical commodities, and the
latter can vary greatly in quality and type.
In addition, the authors’ estimates are for the cohort
that turned seventy in 1992. In the interim, medical spending has risen at every age. As a result, the
estimates are lower than they would be if they had
followed a more recent cohort.
Finally, their analysis excludes payments made by
Medicare and private insurers. Medicare spending
substantially reduces out-of-pocket medical expenses throughout the retiree population.5 While
the combination of out-of-pocket and Medicaid
expenditures considered by the authors may be
sufficient for some analyses, such as studies of
household saving, other analyses require that all
health care costs be accounted for.

Endnotes
1

M
 errill Lynch Wealth Management, “Merrill Lynch Affluent
Insights Survey National Fact Sheet,” February 2012.

2

J ohn Bailey Jones, Mariacristina De Nardi, Eric French, Rory
McGee, and Justin Kirschner, “The Lifetime Medical Spending of Retirees,” Federal Reserve Bank of Richmond Economic
Quarterly, Third Quarter 2018, vol. 104, no. 3, pp. 103–135.

3

T he paper uses the model found in a paper titled “Couples’
and Singles’ Savings after Retirement” by De Nardi, French,
Jones, and McGee. It is currently a work in progress. It builds
on a previous paper by French and Jones, “On the Distribution and Dynamics of Health Care Costs,” Journal of Applied
Econometrics, November 2004, vol. 19, no. 6, pp. 705–721, as
well as two papers by De Nardi, French, and Jones, “Why Do
the Elderly Save? The Role of Medical Expenses,” Journal of
Political Economy, February 2010, vol. 118, no. 1, pp. 39–75,
and “Medicaid Insurance in Old Age,” American Economic Review, November 2016, vol. 106, no. 11, pp. 3480–3520.

4

See De Nardi, French, and Jones (2016).

5

S ee Silvia Helena Barcellos and Mireille Jacobson, “The Effects
of Medicare on Medical Expenditure Risk and Financial Strain,”
American Economic Journal: Economic Policy, November 2015,
vol. 7, no. 4, pp. 41–70.

This article may be photocopied or reprinted in its
entirety. Please credit the authors, source, and the
Federal Reserve Bank of Richmond and include the
italicized statement below.
Views expressed in this article are those of the authors
and not necessarily those of the Federal Reserve Bank
of Richmond or the Federal Reserve System.

John Bailey Jones is a senior economist and research
advisor and Aaron Steelman is director of publications in the Research Department at the Federal
Reserve Bank of Richmond.

FEDERAL RESERVE BANK
OF RICHMOND
Richmond Baltimore Charlotte

Page 6