Impressive Past and Present of Medical Research Funding
Uncertain Future for Medical Research Funding
Characteristics of Trust Funds and Entitlements
of Mandatory Spending Designs for Medical Research
Allocation and Management of Funds
First Program of the Mary Woodard Lasker Trust and the American
Association for the Advancement of Science, Wye River Conference
Centers, Queenstown, Maryland, February 14-16, 1999. Thanks to Namratha
Swamy for research assistance.
Most federal funding for medical research is provided through the
annual appropriations process, which is commonly know as "discretionary"
spending. This paper analyzes how discretionary federal funding
for medical research could be converted into two forms of "mandatory"
spending: trust funds and entitlements. Mandatory spending is enabled
by laws other than annual appropriations acts; that status is generally
perceived as providing greater funding stability.(1)
Converting medical research from discretionary to mandatory funding
would be a very difficult task. Most members of the appropriations
committees would be reluctant, at best, to cede their authority
over this area of the budget. Members of the budget committees would
be concerned that mandatory status for medical research funding
would decrease their flexibility to manage the aggregate budget.
And budget analysts might be skeptical that mandatory status is
well-suited to the allocation and management of medical research
paper begins by comparing the impressive record of medical research
funding with the risk that this record will not be matched in the
future. Next it describes some essential characteristics of trust
funds and entitlements. The paper then analyzes some complexities
of using trust funds and entitlements to finance medical research.
Impressive Past and Present of Medical Research Funding
In its statement of objectives, Funding First argues that medical
research is "a national treasure which may be jeopardized by current
funding practices" (Funding First, 1997). Given the life-saving
and enhancing impacts of U.S. medical research, it certainly deserves
its description as a "national treasure." It is also clear that
this treasure was bought with sustained and enduring support from
the federal government, beginning with the 1945 Bush report's call
for national investment in science. Of course, there were a few
years when federal funding for medical research either failed to
grow or was cut, most recently in 1982 and 1996 (OMB, 1998b). But
in comparison to most other categories of domestic discretionary
spending which were reduced significantly from 1990 to 1997, medical
research did very well.(2)
Clinton's pledge in the fiscal year 1999 budget to increase the
National Institutes of Health (NIH) budget by 50% over a multiyear
period, symbolized by NIH Director Varmus's gallery seat of honor
for the 1998 State of the Union Address, indicates that this funding
success is likely to continue. There is bipartisan Congressional
support for reaching this goal and even going beyond it, which is
the historical Congressional reaction to Presidential requests for
medical research funding.
reasons for the generous funding of medical research are numerous.
Most importantly, the average American is confident that investments
in medical science will be productive, an assumption that is fully
justified by the remarkable advances of the past century. Those
investments have been enabled by effective administration within
NIH. An essential element of this administration is the strong scientific
norm in favor of peer review, which allows appropriators to be generally
confident that funds will be spent efficiently. The medical research
establishment has advocated skillfully for federal support. That
quest has been aided by the design of NIH's extramural program,
which grants funds to more than 25,000 recipients, broadly distributed
across the country, who are well-respected members of major institutions,
and who are not shy. Consequently, most Members of Congress know
that medical research is politically popular, and they often proclaim
their support of funding for local institutions or targeted diseases.(3)
(Strickland, 1972; Weston, 1994)
Uncertain Future for Medical Research Funding
Despite the impressive growth of discretionary appropriations
for medical research, another of Funding First's objectives is taking
"funding for research beyond the annual appropriations process"(Funding
A skeptic might compare such concern about future appropriations
to the fear of a patient who is scheduled for an operation at world-famous
Johns Hopkins Hospital, but who wants to transfer to the world-famous
Mayo Clinic on the grounds that he heard the jello served at Hopkins
wasn't very good. On the other hand, one could come up with a more
frightening analogy-a patient who might need an emergency operation
within the year, but who is scheduled to travel through the developing
countries during that period. Prudence might dictate changing plans.
about the future of medical research is very reasonable given the
recent changes in the health sector (Barfield and Smith, 1997).
For example, the widespread adoption of managed care has reduced
important cross-subsidies for teaching and research-will this continue?
Might the likely return to higher rates of health care cost growth
reduce money available for research? Will technological advances
further increase the cost of the research process?
specifically, discretionary appropriations for medical research
are threatened by uncertainty at both the macrobudgetary and microbudgetary
levels. At the macrobudgetary level, discretionary spending caps
loom. From 1990 to 1997, these caps greatly limited increases above
a nominal dollar base for total discretionary spending (in other
words, there was far less money than necessary to cover the costs
of inflation, resulting in a real funding decrease). The caps have
been a major factor in eliminating federal deficits. If applied
in the future, they could force even those who have vocally pledged
to increase medical research funding to do the opposite when budgets
are requested or appropriations bills are written.
assumption that the caps will continue is a major contributor to
the large surpluses that are now projected. Yet in 1998, the Congress
and the President used loopholes and gimmicks to exceed the caps
by over $20 billion. Complying with the caps in 1999 will be very
difficult, and given the large surplus projections, it is possible
to envision a benign scenario in which the caps would be increased
the other hand, domestic discretionary appropriations are not the
only claimants to the projected surpluses. Alternative major uses
include tax cuts (the Republicans' top agenda item), defense spending
(the current consensus is that some increase is needed), and Social
Security, Medicare, and Medicaid (it will be extremely difficult
to reach agreements, but any agreement will be very costly).
addition, the longest peacetime expansion of the economy will eventually
end-perhaps sooner than later--converting surpluses into deficits.
The resultant unbalanced budget would disturb most Americans (though,
in many cases, without much thought about why), and an easy (though
counterproductive) policy would be to enforce strict discretionary
of these macrobudgetary scenarios will occur? I don't have the foggiest
idea. A common feature of budgetary politics is that many participants
make assumptions about the course of the American economy, and about
the balance of power in American politics, which become incorrect
in just a year or two. But a defensible precautionary approach might
consider the macrobudgetary risk significant enough to justify a
strategy of escape from discretionary appropriations.
secondary risk associated with discretionary funding is at the microbudgetary
level. The dysfunctionalities of the appropriations process are
widely known: bills are often enacted long after the beginning of
the fiscal year, there is usually turfing between committees, and
bills are often loaded down with micromanagement directives-especially
earmarks to favored programs, organizations, and locations.
to the experiences of many other programs, such flaws have been
relatively minor for the NIH, but nothing guarantees that will continue.
Perhaps the most significant fear is that public pressure for disease-specific
projects could lead the Congress to write very detailed appropriations
that would imperil funding for more-productive basic science. So
a reasonable precautionary approach here might be to seek autonomy
from the annual appropriations process. Protecting the current level
of flexibility could be worth the cost of less funding.
of Trust Funds and Entitlements
There is no technical guidebook or overarching law on mandatory
funding in general, and on trust funds and entitlements in specific.
Rather, the details of funding for each mandatory program is governed
by the specific law that creates that program. Consequently, the
characteristics of mandatory spending designs vary significantly,
and they are usually quite complex. For example, many (but not all)
trust funds earn interest on their fund balances, but at different
essence of both trust funds and entitlements is that, by law, either
the receipt of funds by a program or the obligation of those funds
becomes automatic. For example, the typical trust fund law requires
that specific revenues be reserved in a fund for a particular purpose.
But note that labeling a fund a "trust" does not, as implied by
the name, create the kind of stability created by a bequest from
a will. In this private trust case, a fiduciary must gain the permission
of a court to legally ignore the conditions of the bequest, and
such permission is granted only in extraordinary circumstances.
In the federal trust case, the Congress and the President need only
rewrite the law to recapture or shift the funds, and they often
do just that. Also note that placing receipts into a trust fund
does not automatically make them available for obligation; the governing
law must be written to require that. For some important trust
funds, the appropriations committees have retained their ability
to appropriate moneys out of these funds (meaning that spending
remains discretionary), and the appropriators' reluctance to appropriate
all available receipts and interest earnings have created large
balances in the funds.
the typical entitlement program, the government guarantees payment
of funds to any eligible beneficiary who meets specified criteria,
and it does so by formula. This usually means that when the number
of those entitled beneficiaries increases unexpectedly from what
was anticipated in the budget, the government must spend more. Again,
however, it should be noted that on a few occasions (for example,
with food stamps), laws gave the appropriations committees authority
to refuse to appropriate all the moneys necessary to finance expected
entitlement benefits. If they appropriated less than the full amount,
the administering agency would be expected to ration access. In
contrast, most entitlement formulas automatically direct varying
amounts of benefits to different beneficiaries, thus reducing or
eliminating the amount of discretion that might otherwise be exercised
by appropriators and administrators.
of Mandatory Spending Designs for Medical Research
This section discusses two major complexities of designing trust
funds and entitlements. The first part considers the implications
of some alternative revenue sources for a trust fund-asking "who
would pay how much into the fund and why?" The second part addresses
how the allocation and management of funds could be affected by
these mandatory designs-asking "who would get how much and for what?,"
and "what would guarantee that they spend efficiently what they
Three of the alternative revenue sources for a trust fund for medical
research are: a specified amount of general revenues; taxes on health
insurance premia; and taxes and/or royalties on sales from medical
general revenues source has the significant advantage of consistency
with the widely-accepted "public good" justification for federal
funding of medical research. That is, federal funding is thought
- the potential
benefits of medical research are so large that relying on a limited
flow of charitable giving would delay progress (Harden, 1996),
- the private
sector should underinvest in basic medical research because of
the uncertain rate of scientific advance and because of its inability
to capture all benefits flowing from these investments, and
with rare diseases will not provide a sufficiently large market
for private research investments to be profitable, and the result
is that these patients are unfairly neglected.
a general revenue set-aside into a trust fund presents the same
difficult question currently faced in the annual appropriations
process: how much should be set-aside for each year? The only difference
is that this question would be answered for a multiyear period-for
example, a specified amount of funds for the base year with scheduled
increases for each of the following years (or "outyears"). In other
words, the "trust fund" would provide advance appropriations over
a series of years.(5)
second revenue alternative listed above is a tax on health insurance
premia. (A similar mechanism was discussed for financing medical
education during the 1993-4 health care reform debate.) As with
the previous alternative, a policy decision would be required to
determine what rate(s) to impose. But once this choice was made,
the tax on insurance premia would "solve" the problem of having
to set funding levels for each outyear; that would be determined
by the amount of health insurance in force.
would not guarantee rising or stable funding levels from year to
year, however. What if, the trust fund tax rate were set as a percentage
of the health insurance premium paid? Premium increases would reduce
the number of insurance policies in force, but produce higher payments
from the insurance policies remaining in force. The balance of these
effects would be uncertain from year to year because of the complexity
of the health insurance market, which would complicate budget forecasts.(6)
Receipt instability could also be caused by the imperfect collection
and accounting abilities of the government (GAO, 1998) Similar problems
have resulted from the per-gallon tax on gasoline that finances
the Highway Trust Fund.
final effect of a premium tax is that the burden of financing medical
research would shift. How it would shift would be very sensitive
to the tax rate(s) that would be applied, to the types of insurance
covered (would Medicare and Medicaid pay?), and to the responses
to the tax. Clearly, the tax would be avoided--at least in the first-order--by
those who did not have health insurance. Since access to health
insurance is positively correlated with higher income, a premium
tax should be roughly progressive in incidence. However, it is likely
to be less progressive than the current tax system from which federal
medical research is now financed.
The third revenue alternative would be to capture and dedicate taxes
and/or royalties on sales from medical products. For example, some
manufacturers of proprietary pharmaceuticals recently proposed that
if they were granted up to ten years of additional market exclusivity
for certain drugs and antibiotics, they would make royalty payments
to the government--but only 3% of net U.S. sales--to be used for
medical research. The Congressional Budget Office (CBO) found that
much of the new revenue would be offset by increased federal health
care payments for these non-generics, and that private costs would
also increase (1997; see also U.S. Congress, Senate Subcommittee
on Labor-HHS, 1998).
the extent that private profits result from federal investments
in basic research, the private sector might owe a duty to pay, and
the federal government might deserve to capture some of these profits.
This alternative, like the previous one, raises positive and normative
questions that are too complicated to answer in detail here. Some
- How would
changes to patenting/intellectual property laws affect trust fund
- To what degree
and how should the prices of taxed medical products be regulated?
- What would
be the incidence of the tax-what shares would be borne by shareholders,
by employees, and by customers of taxed firms?
- If almost
all of the burden was borne by customers-by and large, by the
sick and disabled---would this be fair?
for the sake of argument that the assumption in the first part of
the last question was accurate. That the sick and disabled would
pay for medical research might be considered fair because it is
consistent, at face value, with the "benefit principle"--those who
benefit should pay. This principle often justifies the dedication
of funds. On the other hand, sickness and disability is often due
to bad luck. Healthy people-many of whom are healthy because of
previous medical research-may have a greater moral obligation to
finance medical research.
taxes are often politically attractive because they are perceived
as falling on only a small group of taxpayers. This perception of
the burden is often inaccurate because of tax shifting; for example,
much the burden of the corporate income tax is shifted to consumers.
Similarly, a tax on medical products may be perceived--largely incorrectly--as
falling primarily on the shareholders of manufacturing firms.
occasion, the initial payers of targeted taxes accede to their "burdens"
in return for the dedication of receipts to specified purposes.
That is, the benefit principle has a common political corollary:
those who pay should have a special say on the allocation of these
funds. Again, the complications of this approach for medical research
require more space than available here. But here's one question
to consider-if taxes on medical products were dedicated to medical
research, would the manufacturers have an incentive to suggest that
less research be done on preventative health?
related budgetary issue affects whether the tax/royalty payments
would be on- or off-budget. If payments were paid voluntarily (as
in a royalty agreement) and allocated by an institution not under
the direct control of the federal government (that is, not by NIH),
OMB and CBO might decide to treat them as off-budget (budget concepts
and practices on this issue are in flux). A contrasting case was
the health alliances proposed by the Clinton administration in 1993.
The quasi-mandatory nature of payments and extensive indirect government
controls led CBO to classify the health alliances as on-budget,
a decision that was disputed by the Office of Management and Budget.
conception of the budget's proper could become dated if health policy-making
ever leaves the incrementalist torpor in which it now appears trapped.
America's health care system offers great quality, but at an indefensibly
high cost and to too few people. Reforming the system will never
be easy, but it should not be impossible. If an ambitious and skillful
political leader ever takes the risk of proposing major institutional
reforms, adoption might be more likely if there is a simultaneous
redefinition of "the budget" on the order of the global (or sectoral)
budgeting of the type proposed by Herb Stein (1989).
Allocation and Management of Funds
Mandatory status often does reduce annual conflicts over the allocation
of funds. Because of benefit schedules, our society does not debate
how much individual Social Security beneficiaries should receive
each year. Could mandatory status for medical research have a similar
not. Begin with entitlement status. Most entitlements enjoy this
status because of a mix of political commitments and the necessities
of program design. There may be a legal or quasi-legal "right" (Indian
trust funds, employee pensions), or a "moral" one because of repeated
government promises (Social Security). It is difficult to imagine
how some entitlement programs could work without relative stability;
for example, long-term planning for retirement is aided by confidence
that Social Security benefits will provide an income base.(7)
the necessities of program design suggest that medical research
funding should be an entitlement? Assuming the accuracy of the common
description of medical research as a public good, Joseph White implies
the answer is no.
On the whole,
an entitlement should be for a benefit that can be quantified
and divided up. Protection from crime, a classic "public good,"
would be hard to deliver in this way. Space exploration, the census,
in short most bureau programs in the real world-do not have benefits
that can be divided so neatly. A pension, however, is quite precise.(White,
1998, p. 513)
It is true
that investigator grants are as precise as pensions in amounts awarded.
And a common complaint is that many successfully peer-reviewed research
projects go unfunded. This might suggest that all investigators
who pass peer-review should be entitled to funding.
Of course, that
is a totally unrealistic option. The cost would be tremendous, and
the incentive effects for the peer-review process would be dangerous.
Not all good projects are the best. While expectations about the
probable benefits from most grants are imprecise, grant competition
can reduce that uncertainty in order to direct resources towards
projects that are more likely to be cost-effective.
be created instead at the institute level? Not if allocating funds
for medical research is an inherently judgmental task that requires
some year-to-year flexibility (NIH, 1997; Institute of Medicine,
1998, 1990). Arguably, allocations could be improved by permitting
even more administrative flexibility than is currently allowed.
Such flexibility is especially valuable to all programs that rely
on human and physical capital, as is the case with medical research.
A forthcoming report from the President's Commission on Capital
Budgeting is expected to propose some rules that should allow more
efficient investments in labs and information technology.
of mandatory funding is that it gives program advocates absolute
independence from a squalid political process. Program managers
need not lobby each year-they can focus on efficiency and effectiveness--because
they enjoy the trust of elected officials and a stable funding stream.
is quite different, for there is simply no way to avoid political
and budgetary oversight. Conversion from discretionary to mandatory
status would eliminate the role of the appropriators, but would
probably increase the activities of authorizers. If revenues were
dedicated, this would allow the revenue-raising committees to also
claim a jurisdictional right to oversight of medical research (Cogan,
1994). And the creation of new mandatory spending is controlled
by the "paygo" rules of the Congressional Budget Act (as amended),
which require that this spending be offset by cuts in existing mandatory
spending and/or by revenue increases.
It may also
be that the annual appropriations process, despite its well-known
flaws, promotes responsive and economical program management. There
are enough examples of independent agency horror-stories (for example,
the Synfuels Corporation) to suspect that there is merit to this
argument. NIH's grants and contract management practices were criticized
in the past, and the annual prospect of being blamed for similar
lapses may keep administrators on their toes. More broadly, the
annual appropriations process may promote the agency's legitimacy
by forcing it to respond to the demands of citizens about the purposes
of medical research.
of the 1993 Government Performance and Result Act (GPRA) may change
these accountability procedures. GPRA requires agencies to engage
in strategic planning and outcomes measurement, and to relate the
products of these processes to operating procedures and to budget
requests. Research agencies like NIH recognize that measuring the
impact of research activities on social outcomes is a very complicated
task, but this difficulty may be ignored or minimized by some influential
GPRA advocates. If implementation of GPRA continues at its strong
pace, research administrators and their overseers will have to negotiate
what kinds and amounts of benefits are appropriately linked to budget
allocations. The basic analytical challenge will be the same: how
to determine the point at which a dollar spent on medical research
results in diminishing benefits (Ginzberg and Dutka, 1989).
The nature of
this negotiation will be affected particularly by the types of potential
benefits that will be compared to each other. These vary significantly
within the NIH, the executive branch, and the Congress. In the latter,
the basic competition is structured by appropriations subcommittee
jurisdictions, which forces comparisons of inherently incommensurable
outcomes. I can only sympathize with the Members of Congress who
must now choose between funding for medical research or for foster
care and adoption services.
But the reality
of budgetary competition is that there is no way to avoid such comparisons
by shifting from discretionary to mandatory funding. In fact, the
trust fund form often seems to attract competitors. For example,
the Highway Trust Fund now finances bike paths, a policy successfully
advocated by cyclists despite the opposition of auto and truck interests.
Imagine a "Medical
Research Trust Fund" established to promote desirable health outcomes.
Might it attract social science and clinical practice researchers
that would portray themselves as symbionts to biomedical research,
but might they be perceived by biomedical researchers as parasites?
Or imagine that the trust fund form would be applied to research
in general, an extension of President Clinton's "Research Fund for
America" data aggregation from the 1999 budget. In this case, the
intended outcome would be knowledge advances from research of all
types. That trust fund would allow direct competition between NIH's
medical research and the basic science funded by NSF (on which NIH's
medical research is so dependent).
Useful guides to federal budget concepts and process include Schick,
1995; OMB 1998a.
Data from the 1980s and 1990s on economy-wide spending for health
research and development also show impressive growth; see Neumann
and Sandberg, 1998.
Note also that investments in medical research are consistent with
the American fascination with technological advances and are much
less threatening politically than some other types of medical spending.
Consider the recent opposition to funding for research on cost-inefficient
clinical practices, or to the expansion of health care access that
would reduce the market shares of insurers.
"Special funds" do not earn such interest, even though
they resemble trust funds by receiving dedicated receipts for restricted
purposes. For more details on prototypical complex accounting and
spending designs, see Meyers, 1994, chapters 4-6; GAO, 1991, 1992,
1994b. On trust funds, see Patashnik, 1997; on entitlements, see
White 1998, 1999; CBO, 1994: GAO, 1994a.
"trust funds" have been financed by set-aside government
revenues--General Revenue Sharing and the Violent Crime Reduction
Trust Fund. The practical effect of these provisions was to "fence"
off, or protect, these parts of the discretionary appropriations
total from reduction. A symbolic effort towards this result was
the Specter amendment (#2254) to Congressional Budget Resolution
for FY 1999 (considered in the Senate April 2, 1998), which would
have "dedicated" $2 billion to medical research. The quotes
reflect the fact that such budget resolution directives are only
advisory to appropriators. The source of the assumed funding was
originally to be from proceeds of the then-expected tobacco settlement,
but was changed to a cut in all other discretionary appropriations
example of this difficulty was the Harkin amendment (#579) to Revenue
Reconciliation Act of 1997 (considered in the Senate June 27, 1997).
It would have dedicated to medical research one-half of projected
increased revenues resulting from a cut in capital gains tax rate-an
especially difficult to predict and probably ephemeral amount.
contrast, one of the traditional justifications for entitlement
status-that the federal government has a counter-cyclical responsibility
to provide a safety net for the unemployed-was not reflected in
the fine details of the 1996 welfare reform law.
Barfield, Claude E. and Bruce L. R. Smith, editors,
1997. The Future of Biomedical Research. Washington, D.C.:
American Enterprise Institute and the Brookings Institution.
Vannevar, editor, 1994. Science-The Endless Frontier. Washington,
D.C.: Office of Scientific Research and Development.
John F., 1994. "The Dispersion of Spending Authority and Federal
Budget Deficits," in Cogan, Timothy J. Muris, and Allen Schick,
The Budget Puzzle. Stanford: Stanford University Press, pp.
First, 1997. "Objectives," http://www.fundingfirst.org/objectives.html.
Eli and Anna B. Dutka, 1989. The Financing of Biomedical Research.
Baltimore: Johns Hopkins University Press.
Victoria A., 1986. Inventing the NIH: Federal Biomedical Research
Policy, 1887-1937. Baltimore: Johns Hopkins University Press.
of Medicine, 1998. Improving Priority Setting and Public Input
and the National Institutes of Health. Washington, D.C.: National
of Medicine, 1990. Funding Health Sciences Research: A Strategy
to Restore Balance. Washington, D.C.: National Academy Press.
Meyers, Roy T., 1994. Strategic Budgeting. Ann Arbor: University
of Michigan Press.
Peter J. and Eileen A. Sandberg, 1998. "Trends in Health Care R&D
and Technology Innovation," Health Affairs, 17:6, pp.111-119.
Patashnik, Eric, 1997. "Trust Funds and the Politics of Precommitment,"
Political Science Quarterly, 112: pp. 431-452.
Allen, 1995. The Federal Budget: Politics, Policy, and Process.
Washington, D.C.: Brookings.
Herbert, 1989. Governing the $5 Trillion Economy. New York:
Oxford University Press.
Stephen P., 1972. Politics, Science, and Dread Disease. Cambridge,
Massachusetts: Harvard University Press.
Congress, Senate Committee on Appropriations, Subcommittee on Departments
of Labor, Health and Human Services, Education, and Related Agencies,
1998. "Legislative Proposal to Increase Funding for Medical Research,"
Hearing 105-524, Government Printing Office.
Congressional Budget Office, 1997. Cost Estimate of Proposed Demonstration
Project to Fund Biomedical Research, November 18.
Congressional Budget Office, 1994. Reducing Entitlement Spending.
General Accounting Office, 1998. Excise Taxes: Internal Control
Weaknesses Affect Accuracy of Distributions to the Trust Funds.
General Accounting Office, 1994a. Budget Policy: Issues in Capping
Mandatory Spending. September.
General Accounting Office, 1991, 1992, 1994b. Principles of Appropriations
Law, volumes 1-3, http://www.gao.gov/special.pubs/vol1.pdf,
Office of Management and Budget, 1998a. The Budget for Fiscal
Year 1999, Analytical Perspectives.
Office of Management and Budget, 1998b. The Budget for Fiscal
Year 1999, Historical Tables. pp. 55-59, 274.
National Institutes of Health, 1997. "Setting Research Priorities
at the National Institutes of Health," http://www.nih.gov/news/ResPriority/priority.htm.
Richard Clarke, 1994. The Legislative Politics of Appropriations
for Biomedical Research. Dissertation, School of Public Health,
University of Michigan. Ann Arbor: University Microfilms.
Joseph, 1998. "Entitlement Budgeting vs. Bureau Budgeting," Public
Administration Review, 58: November, pp. 510-521.
Joseph, 1999. "Budgeting for Entitlements," in Roy T. Meyers, Handbook
of Government Budgeting, San Francisco: Jossey-Bass, pp. 678-698.
1999 American Association for the Advancement of Science