Growing Pain:
The Expense of Drugs for the Elderly
By Dean Baker and John Schmitt[1]
February
10, 2003
Executive Summary
Prescription drug costs are placing a large and growing burden on the elderly population. The Congressional Budget Office (CBO) estimates that drug costs have risen at a 19.1% annual rate over the last eight years and projects that they will rise at a 10.1% rate annually over the next decade. This paper assesses the extent of this burden by constructing projections for the increase in drug costs over the decade by income quintile of the elderly. The paper then examines the costs of government drug programs of varying degrees of generosity, assuming that the projections of growth in drug expenditures are not affected. Finally, the paper examines options that would contain the cost of drug growth.
The projections show that:
As Baker and Chatani (2002) have argued, such a system would eliminate many of the distortions created by the patent system, such as research money wasted in developing copycat drugs, large marketing expenses, and secrecy – even falsification – of research findings. Projections from Baker and Chatani indicate that increasing government financial support for research, with the findings placed in the public domain, could save the public as a whole (not just the elderly), more than $200 billion a year in by 2013, even after subtracting the cost of the additional public research.
Growing
Pain: The Expense Of Drugs for the Elderly
Expenditures on prescription drugs have been rising at double-digit rates for more than a decade. This rate of increase had not originally posed much of a problem for most of the population because drugs were a small item in total consumer expenditures. However, after the sustained rapid pace of growth, drug expenditures have now become a significant share of household spending. This is especially true for the elderly, who have the greatest need for drugs and are the segment of the population least likely to have employer-paid insurance coverage for prescription drugs.
This paper examines the
projected path of prescription drugs spending for the elderly over the next ten
years. It also discusses the potential impact of a Medicare prescription-drug
benefit, as well as alternative solutions to the problem of rising drug costs.
The Path of Prescription-Drug Spending by the Elderly
CBO estimates that the elderly currently spend about $100 billion per year on prescription drugs not covered under the traditional Medicare program.[2] According to CBO projections, spending on prescription drugs by the elderly will rise at an annual rate of 10.1% over the next decade, leaving total prescription-drug spending by the elderly at $278 billion, or more than 1.6% of GDP in 2013.
A portion of this spending on prescription drugs by the elderly is currently covered by the government. Medicare HMOs account for about 6.0% of drug expenditures by the elderly.[3] Medicaid, state-based programs, and the Veterans Administration together cover another 21.0% of drug expenditures. This leaves more than 70% of drug expenditures to be covered by the elderly themselves, either directly out-of-pocket or indirectly through private insurance.
At present, 26% of drug expenditures by the elderly are paid by employer-sponsored insurance, in most cases through plans that maintain insurance for retirees. This source of support for drug payments is likely to dwindle in the near future as employers cut back on the generosity of benefits for retirees.[4] Also, in many industries, retiree health-care benefits were coupled with defined-benefit pension packages. With coverage by defined-benefit plans slipping rapidly, it is likely that coverage by retiree health-insurance plans will fall at a similar pace. This means that through 2013, the period of interest here, a rising percentage of prescription-drug costs will be paid directly by the elderly in their retirement years, rather than being covered under an insurance benefit earned during their working lifetime.
Table 1 shows our forecasts, based on CBO projections and our own analysis of data from the annual Consumer Expenditure Survey for 2000, of average annual prescription drug spending by the Medicare population by income quintile for the years 2000 to 2013. The dollar estimates in the table refer to the sum of direct spending on prescription drugs and indirect spending on those drugs paid for by private insurance purchased by the household. The table excludes expenditures paid for by the government or employer-sponsored insurance.[5] (For a complete description of how Tables 1 and 2 were constructed, see the appendix .)
Table 1
Average Annual Spending by Elderly Households on Prescription-Drugs by Income
Quintiles
|
Bottom |
Second |
Middle |
Fourth |
Top |
2000 |
$632 |
$822 |
$1,042 |
$1,144 |
$1,208 |
2003 |
1,137 |
1,477 |
1,873 |
2,057 |
2,172 |
2004 |
1,276 |
1,659 |
2,102 |
2,309 |
2,438 |
2005 |
1,432 |
1,861 |
2,359 |
2,591 |
2,736 |
2006 |
1,606 |
2,088 |
2,646 |
2,906 |
3,069 |
2007 |
1,801 |
2,341 |
2,968 |
3,259 |
3,442 |
2008 |
2,019 |
2,625 |
3,327 |
3,654 |
3,858 |
2009 |
2,263 |
2,941 |
3,728 |
4,094 |
4,324 |
2010 |
2,535 |
3,295 |
4,177 |
4,587 |
4,844 |
2011 |
2,839 |
3,691 |
4,678 |
5,137 |
5,426 |
2012 |
3,179 |
4,132 |
5,238 |
5,752 |
6,075 |
2013 |
3,558 |
4,625 |
5,863 |
6,439 |
6,800 |
Source:
CBO, CEX and authors’ calculations, see appendix.
As can be seen, all income
quintiles were spending a substantial sum of money on prescription drugs even in
2000. The fifth quintile spends nearly twice as much as the first quintile. This
partly reflects that many elderly households in the bottom quintile are eligible
for Medicaid or other government programs, but also that wealthier households
are probably receiving somewhat more treatment on average. Total spending is
projected to more than quintuple over this period. In just the years from 2000
to 2003, spending per household is projected to have increased by more than 80%.
The figures in table 1 actually imply a slower rate of growth in spending on
prescription drugs in the future than in the recent past. CBO estimates that
combined private and public spending by the elderly on prescription drugs, on a
per capita basis, rose from $600 in 1995 to $2,400 in 2003, a 19.2% annual rate
of increase.[6]
The projected rates of
increase in spending will impose a substantial burden on elderly households at
all income levels, which is apparent when these increases are expressed as a
share of projected after-tax income. Table 2 shows prescription drug spending by
income quintile as a percentage of after-tax income.
Table 2
Shares of Elderly After-Tax Income Used for Prescription Drugs, by Quintile
|
Bottom |
Second |
Middle |
Fourth |
Top |
2000 |
9.4% |
7.0% |
6.1% |
4.6% |
2.1% |
2003 |
15.2 |
11.3 |
9.9 |
7.4 |
3.4 |
2004 |
16.4 |
12.3 |
10.7 |
8.0 |
3.7 |
2005 |
17.8 |
13.3 |
11.6 |
8.7 |
4.0 |
2006 |
19.3 |
14.4 |
12.6 |
9.4 |
4.3 |
2007 |
20.8 |
15.6 |
13.6 |
10.1 |
4.7 |
2008 |
22.6 |
16.8 |
14.7 |
11.0 |
5.1 |
2009 |
24.4 |
18.2 |
15.9 |
11.9 |
5.5 |
2010 |
26.4 |
19.7 |
17.2 |
12.8 |
5.9 |
2011 |
28.5 |
21.3 |
18.6 |
13.9 |
6.4 |
2012 |
30.8 |
23.0 |
20.1 |
15.0 |
6.9 |
2013 |
33.3 |
24.9 |
21.7 |
16.2 |
7.5 |
Source:
CBO, CEX and authors’ calculations, see appendix.
Measured as a share of
after-tax income, the projected increases in prescription-drug spending are
large. In the case of the bottom income quintile, the projected increase in
spending between 2000 and 2013 is equal to 23.9% of after-tax income. Even if we
start the analysis from 2003, the increase in spending by 2013 would be equal to
18.1% of after-tax income. The projected increase in the share of income going
for prescription drugs spending is less for higher income elderly, but still
large relative to their after-tax income. In the case of the middle-quintile
elderly, for example, the projected increase in spending from 2000 to 2013 is
equal to 15.6% of after-tax income; from 2003 to 2013 alone, the increase is
equal to 11.8% of after-tax income. Even the wealthiest quintile will see an
increase in spending equal to 5.4% of after-tax income over this thirteen-year
period. These increases will leave the elderly devoting a large share of their
after-tax income to prescription drugs. By 2013, prescription-drug costs paid
directly out-of-pocket and indirectly through private-insurance plans would
account for about 33.3% of all after-tax income by the poorest fifth of elderly households;
24.9% of after-tax income for the next poorest fifth; 21.7% for the middle
fifth; and 16.2% by the next fifth.
The Cost of a
Prescription Drug Program
The potential impact of rising prescription-drug costs on the living
standards of the elderly provide the rationale for a prescription-drug benefit
to help defray some of these costs. A major problem with any reasonable
prescription-drug plan, however, is that it is likely to be quite costly. Since
2000, both Democrats and Republicans have made a series of proposals for
prescription-drug benefits. While the cost of these particular proposals varies
greatly depending on their structure and generosity, if we assume that programs
don't affect the rate of increase in drug prices, a relatively simple exercise
yields a reasonable range for the potential costs of a new prescription-drug
program.
Table 3 considers the cost in the year 2013, both in 2013 dollars and as
a percentage of GDP in 2013, of a set of plans offering varying degrees of
coverage and generosity. The coverage of the plan decreases from the first
column, which assumes coverage for all seniors regardless of income, through the
last column, which assumes coverage for only the poorest fifth of elderly
households. The generosity of the plan (in the rows of the table) is set at two
levels. The plan would keep prescription drug-expenditures as a share of
after-tax income at either their 2000 (more generous) or their 2003 (less
generous) level. The most expensive plan ($134.2 billion or 0.75% of GDP) would
cover all the elderly (column one) and maintain their total prescription-drug
expenditures at their 2000 (row one) level relative to after-tax income. The
least expensive plan ($11.9 billion or 0.07% of GDP) would cover only the
poorest fifth of seniors (last column) at the 2003 share of income (last row).
Table 3
The Annual Cost of a Prescription-Drug Plan for the Elderly in 2013
|
Protect |
Protect |
Protect |
Protect |
Protect |
Base |
all |
bottom 4 |
bottom 3 |
bottom 2 |
bottom |
Year |
quintiles |
quintiles |
quintiles |
quintiles |
quintile |
|
Billions
of 2013 dollars |
|
|
||
2000 |
134.2 |
100.3 |
68.1 |
38.9 |
15.8 |
2003 |
101.7 |
76.0 |
51.6 |
29.4 |
11.9 |
|
Percent
of 2013 GDP |
|
|
|
|
2000 |
0.75 |
0.56 |
0.38 |
0.22 |
0.09 |
2003 |
0.57 |
0.42 |
0.29 |
0.16 |
0.07 |
Source:
Authors’ calculations, see appendix.
The table demonstrates that any serious drug benefit will be expensive.
The most expensive plan in the table, $134.2 billion or 0.75% of GDP, for
example, is substantially larger than President Bush’s recent tax-cut
proposal, which is projected to cost approximately 0.43% of GDP over the next
decade.[7]
Even this plan is not overly generous, since it would still have seniors paying
as large a share of their income on prescription drugs in 2013 as they did in
2000 –a year when the political pressure resulting from high drug costs was
sufficiently strong that the incumbent president and both major party
presidential candidates proposed prescription-drug benefits.
Even a substantially less generous benefit, for example, one that holds
the spending of the bottom three quintiles at their 2000 levels, is still
expensive in the context of the larger federal budget. Such a plan, which does
nothing to offset the drug spending of the middle-income families in the fourth
quintile (average income $25,000 in 2000), would still cost just under 0.4% of
GDP.[8]
It is possible to whittle back the number of beneficiaries and the size
of the benefits so that the costs become more manageable, but then most seniors
face the substantial increases in drug spending described in the prior section.
In short, given the CBO projections for drug spending, either a large segment of
the elderly will face crushing expenses from prescription drugs, or the
government will be faced with an expensive new entitlement program.[9]
Containing Drug Costs – An Alternative Route?
The previous sections show that the projected increases in drug
expenditures by the elderly will pose a substantial problem over the next
decade, regardless of whether the elderly themselves shoulder these costs or the
federal government picks up most or all of the tab through a senior
prescription-drug benefit. The high costs of either route raise the obvious
question of whether it is possible to reduce the size of the projected burden by
containing the cost of prescription drugs.
Consumers in the United States currently pay far more for prescription
drugs than do consumers in any other advanced economy. According to a recent
analysis by the Australian Productivity Commission (2001), for example, drug
prices were far higher in the United States than they were in all other OECD
countries, with US consumers often paying more than twice as much for drugs than
did consumers in other nations.
The main explanation for the higher US prices is fairly straightforward. The United States is the only nation that guarantees drug companies long monopolies through patent protection, and then does nothing to restrict the price of the drug. Every other industrialized nation has some form of price controls that limit the ability of drug companies to exploit their monopoly position. Only in the United States can firms rely on the government to enforce their monopoly position in the market, while firms maintain unrestricted power to set prices in order to maximize profits.
One obvious answer to the problem of rising drug costs, then, would be to
follow the lead of the rest of the developed world and place price controls on
prescription drugs. Price controls could either take the form of prices that are
explicitly set by some governmental body or prices that are set through a
process of negotiation between a bulk buyer and individual pharmaceutical
company. From an economic standpoint, these two mechanisms would produce largely
the same result, although there may be political reasons why one route would
appear more desirable than the other.
While price controls could bring drug prices down to affordable levels,
they do raise important issues that have not been seriously addressed. Most of
the price reduction would come at the expense of profits and marketing costs,
both of which are substantial components of drug prices. However, whatever
entity sets the price – either a price control board or a bulk-buying agent
– would also effectively be setting the drug companies research agenda through
the financial incentives implicit in the controlled (or bargained) prices. The
categories of drugs for which the government allowed the greatest mark-up over
marginal costs would be the ones where the industry would have the greatest
incentive to carry through further research.
This is important to recognize, since it means that once the price is set
by a governmental body, the direction of research is also being determined by
this body. There is nothing necessarily wrong with this situation –the
pharmaceutical industry’s research decisions are already affected by the
industry's own assessment of whether it will be able to get the government or
private insurers to pay for specific drugs– but this reality does mean that
the pricing agency must consciously set prices with the understanding that it is
determining the course of drug research. The failure to set prices with this
consideration in mind will lead to the wrong incentive structure for future
research.
Also, if the government is guiding
research through its pricing mechanism, then it is not clear that the current
patent system is the most efficient way to finance drug development. Already,
close to half of all biomedical research is not dependent on the patent system,
being supported either by the National Institutes of Health or other government
agencies or foundations, universities, and private charities. The patent system
is inherently wasteful, since it promotes secrecy and duplicative research
efforts. According to the Food and Drug Administration (FDA), more than 70% of
the new drugs approved in the last decade do not constitute qualitative
improvements over existing treatments (U.S. FDA, 2001). Patents can also provide
substantial incentives for falsifying research findings, or at least withholding
the disclosure of findings that reflect negatively on a firm’s drugs. The
patent process itself, and the potential it creates for legal disputes, also
adds additional costs that are passed on to consumers.[10]
For these reasons, and others, it is likely to be more efficient to
simply finance all drug research directly, through some combination of expanded
government support and increased incentives for research by the non-profit
sector. The products of this research could be placed immediately in the public
domain. Then all firms could compete to produce newly developed drugs at the
lowest price – without the government stepping in to impose either a patent
monopoly or price controls. In effect, all drugs could be produced and sold in
the same way as generic drugs at present.
Baker and Chatani (2002) have produced a set of projections of the
potential savings under such an alternative mechanism for financing drug
research. Their projections show that the annual savings from such a system
could easily exceed $200 billion by 2013 (see their Tables 10B and 11). The fact
that there is a large and growing gap between the price for which new drugs are
sold and their actual cost of production virtually guarantees that there will be
large gains associated with a system of direct financing of drug research.
Perhaps the most basic result in economic theory is that efficiency is
maximized when the price is equal to the marginal cost of production. The
distortions that have resulted from the gap between price and marginal cost in
prescription drugs – massive marketing efforts, falsified or misleading
research results, and wasteful copycat research – have been well documented
(e.g. Public Citizen, 2001; Bodenhiemer, 2000; Ernst & Young LLP, 2001).
Given this evident inefficiency, good economics dictates that policymakers
should explore mechanisms that bring price and marginal cost into line in the
pharmaceutical industry by undertaking efforts to introduce normal competition
into an industry long characterized by government protection.
In this paper, we have examined the implications of the CBO’s
projections for prescription drug spending by the elderly. CBO's projections
imply large increases in spending on prescription drugs, especially when
measured as a share of projected after-tax income. In the absence of efforts to
control and even reduce prescription-drug prices, these rising costs will fall
either on seniors themselves in the form of higher out-of-pocket expenditures
and higher private-insurance premiums, or on taxpayers more generally in the
form of a prescription-drug program. At an additional 0.75% of GDP per year by
2013, whether the incremental cost of prescription drugs over the next 10 years
are borne privately or publicly, rising drug costs are large relative to other
politically important benchmarks such as President Bush's recently proposed
tax-cut plan (about 0.44% of GDP over the next decade).
A prescription-drug plan that tried to limit price increases, or even
reduce prices, would go a long way toward alleviating the fiscal pressure. One
option is to follow the lead of other advanced economies and implement direct
price controls that prevent pharmaceutical companies from exploiting their
government-granted patent monopolies. An alternative would be to use government
money to fund expanded university, private, and public drug research
(non-pharmaceutical company research already accounts for just under half of all
drug research in the United States). The results of this publicly funded
research would all enter the public domain. At that point, pharmaceutical
companies would participate in a largely deregulated market that would reward
companies producing and distributing drugs at the lowest price. Such a system
could reduce overall private spending on drugs (for the whole population, not
just the elderly) by about $200 billion or 1.1% of GDP per year by 2013, with
the additional research expenditures by the government almost completely offset
by cost savings on existing government health care programs.
References
Australian
Productivity Commission. 2001. International
Pharmaceutical Price Differences, Australian Productivity Commission,
Commonwealth of Australia, <http://www.pc.gov.au/research/commres/pbsprices/finalreport/pbsprices.pdf>.
Baker, Dean and Chatani TK. 2002. “Promoting Good Ideas
on Drugs: Are Patents the Best Way? The Relative Efficiency of Patent and Public
Support for Biomedical Research.” Washington, D.C.: Center for Economic and
Policy Research. <http://www.cepr.net/promoting_good_ideas_on_drugs.htm>.
Bodenhiemer, TK. 2000. "Conflict of Interest In
Clinical Drug Trials: A Risk Factor For Scientific Research." Paper
presented at the National Institutes of Health Conference on Conflicts of
Interest in Scientific Research, August 15, 2000.
Congressional Budget Office. 2002. “Testimony
on Projections of Medicare and Prescription Drug Spending, March
2002,” Testimony before the Committee on Finance, U.S. Senate, Congressional
Budget Office. <http://www.cbo.gov/showdoc.cfm?index=3304&sequence=0>.
Congressional Budget Office. 2000. “Testimony
on a CBO Analysis of the Administration's Prescription Drug Proposal, May
2000,” Washington, D.C.: Congressional Budget Office, <http://www.cbo.gov/showdoc.cfm?index=1985&sequence=0>.
Ernst & Young LLP. 2001. Pharmaceutical Industry R&D Costs: Key Findings about the Public
Citizen Report. Pharmaceutical Research and Manufacturers of America. <http://www.phrma.org/press/newsreleases/2001-08-11.277.pdf>.
Public Citizen. 2001. RX
R&D Myths: The Case Against the Drug Industry's R&D "Scare
Card." Washington, D.C: Public Citizen, <http://www.citizen.org/congress/drugs/R&Dscarecard.html>.
U.S. Food and Drug Administration. 2001. “NDAs Approved
in Calendar Years 1990-2001 by Therapeutic Potentials and Chemical Types,”
December 31, 2001. <http://www.fda.gov/cder/rdmt/pstable.htm>.
Appendix
We base the projections in Tables 1 and 2 on published data from the
Congressional Budget Office (CBO) and on our own analysis of microdata from the
Bureau of Labor Statistics' public-use version of the Consumer Expenditure
Survey (CEX) for 2000.
To construct Table 1, we started with the CBO's estimate of average
per-person private expenditure on prescription drugs for the elderly in 2003
($2,400) and deflated this figure to what it would have been in 2000 ($1,421).
We multiplied the latter figure by 61%, which is the percentage of drug
expenditures that CBO estimates are paid either directly by the elderly, or
through insurance they currently purchase. Separately, we used the CEX to
calculate per-person private expenditure on prescription drugs and private
health-insurance for "all-elderly" households (households where all
members are 65 or older) on average and for each before-tax income quintile.[11]
We combined direct expenditures on prescription drugs and private health
insurance in recognition that an important part of the expenditures made by
private health insurance for the elderly is for prescription drugs. Since the
per-person levels of prescription-drug expenditures (plus health insurance)
calculated using the CEX don't exactly match the CBO estimates, we then scaled
the CEX-based quintile expenditures by the ratio of our CEX-based average
expenditure to the CBO average, in order to make our projections as compatible
as possible with those of the CBO. In effect, we have used the CBO to set the
average level of expenditures and the microdata from the CEX to look at the
implied spread of expenditures around the CBO average. The resulting ratios of
quintile expenditure to average expenditure were: first quintile, 66.2%; second
quintile, 82.0%; third quintile, 109.1%; fourth quintile, 119.8%; and fifth
quintile, 126.5%.
This scaling procedure also has the advantage that it reduces any errors
that may be introduced by including CEX expenditures on private insurance in our
estimate of total prescription drug expenditures. The scaling ensures that
non-prescription-drug-related portions of private-insurance premiums can't push
our estimated level of prescription-drug expenditures above the presumably more
accurate CBO baseline. We then applied CBO's own estimate of average growth in
prescription-drug expenditures (10.1%) to the 2000 figures for each quintile to
calculate projected expenditures for the years 2003-2013.[12],[13]
We made one final adjustment to
produce the data in Table 1. The percentage (61% in 2000) of drug expenditures
paid either directly by the elderly or through currently purchased insurance is
assumed to rise to 74% in 2013. This path assumes that the share of the
prescription drugs purchased by the elderly that are paid by employer-provided
insurance falls from 26% in 2000 to 13% in 2013, dropping at the rate of 1.0
percentage point annually. While this assumption is consistent with recent
trends towards lower rates of employer-provided health insurance for retirees,
and for less generous benefits for those receiving coverage, as noted by CBO,
the assumed rate of decline in coverage could prove too fast or too slow.[14]
To construct Table 2, we used the expenditure figures in Table 1 and our own calculations of average quintile income. The projections assume that average real income for elderly households grows by 1.1 percentage points annually, the growth rate in average real wages projected in the 2002 Social Security trustees report (Table V.B.1). (Benefit growth for new retirees is linked to the growth in the average real wage.) [15] Our methodology implicitly assumes that the distribution of income among elderly households will remain unchanged between 2000 and 2013.
Note that our projections assume that for both Table 1 and Table 2 the distribution of spending on the combination of health insurance and prescription drugs is the same as the distribution of spending on prescription drugs alone. Since health insurance is also used to cover other health expenditures by seniors, there is likely to be some divergence in the distribution of these two categories of expenditures. Our projections also assume that the lower percentage of drugs expenditures covered by employer-provided insurance does not affect the distribution of spending by quintile. This is almost certainly wrong, since higher-income seniors are the ones most likely to currently be receiving drug coverage through an employer-provided plan. As a result, the projections in this paper may overstate the growth in prescription drug spending for the bottom quintiles and understate the growth in spending for the top quintiles. As discussed in the text, these projections also assume that the distribution of spending by quintile remains constant at the 2000 level. This is implausible, since many lower income households will find themselves unable to afford drugs recommended by their doctor, in the absence of additional government assistance.
[1] Dean Baker is co-Director and John Schmitt is Senior Research Associate at the Center for Economic and Policy Research.
[2] See CBO 2002, Table 3.
[3] See CBO 2002, Figure 4.
[4] The trend toward falling coverage among retirees and the diminishing generosity of benefits was documented as early as 1997. See Hewitt Associates, Kaiser Medicare Policy Project, Retiree Health Trends and Implications of Possible Medicare Reforms (Washington, D.C.: Hewitt Associates, September 1997), available at www.kff.org/content/archive/1318/retiree_r.html.
[5] We exclude expenditures from employer-provided insurance because the purpose of this calculation is to assess the cost of drugs as a share of household income. The value of employer-provided insurance to retirees is not included in the measure of after-tax income.
[6] The 1995 figure appears in CBO, 2000, Table 1.
[7] The tax cut is projected to cost $675 billion over the years from 2003 to 2013. GDP is projected to be approximately $156.8 trillion over this period.
[8] Even these estimates understate the cost of rising drug prices to the government, since they ignore the increased drug costs in the portion of senior drug spending (Medicaid, Medicare in-patient, Medicare HMOs, and Veterans Administration) that is already covered by the federal government. The projections also assume that states continue to assume the same share of the overall burden – a situation that may prove impossible as cash-strapped states face sharply rising drug costs.
[9] Of course, another possibility is that the elderly, particularly poor and middle-income elderly will simply choose to go without prescribed medication.
[10] See Baker and Chatani (2002) for a fuller discussion of these issues.
[11] In practice, when using the CEX data, we worked with households and made subsequent adjustments based on average household size by quintile. The average household size for our sample of "all-elderly" households in the CEX was 1.38.
[12] CBO, 2002, table 4.
[13] The per capita drug expenditure projection appears in CBO, 2002 table 4.
[14] See Hewitt Associates,
Kaiser Medicare Policy Project, Retiree Health Trends and Implications of
Possible Medicare Reforms (Washington, D.C.: Hewitt Associates,
September 1997), available at www.kff.org/content/archive/1318/retiree_r.html.
[15] The 2003 income levels assume that the real income of elderly households increased by 3.3 percentage points in total between 2000 and 2003. The projected nominal growth rate used in these projections is 3.6%, which includes an expected 2.5% average annual inflation rate.