Federal Assistance: Illustrative Simulations of Using Statistical
Population Estimates for Reallocating Certain Federal Funding
(23-JUN-06, GAO-06-567).
Decennial census data need to be as accurate as possible because
the population counts are used for, among other purposes,
allocating federal grants to states and local governments. The
U.S. Census Bureau (Bureau) used statistical methods to estimate
the accuracy of 1990 and 2000 Census data. Because the Bureau
considered the estimates unreliable due to methodological
uncertainties, they were not used to adjust the census results.
Still, a key question is how sensitive are federal formula grants
to alternative population estimates, such as those derived from
statistical methods? GAO was asked to identify (1) the top 20
formula grant programs based on the amount of funds targeted by
any means, and (2) the amount of money allocated for Medicaid and
Social Services Block Grant (SSBG), and the prospective impact of
estimated population counts from the 1990 and 2000 Censuses on
state allocations for these two programs. Importantly, as agreed,
GAO's analysis only simulates the formula grant reallocations. We
used fiscal year 2004 Medicaid state expenditure and 2005 SSBG
state allocation data, the most recent data available.
-------------------------Indexing Terms-------------------------
REPORTNUM: GAO-06-567
ACCNO: A55843
TITLE: Federal Assistance: Illustrative Simulations of Using
Statistical Population Estimates for Reallocating Certain Federal
Funding
DATE: 06/23/2006
SUBJECT: Allocation (Government accounting)
Block grants
Census
Data integrity
Formula grants
Medicaid
Population statistics
Statistical data
Statistical methods
1990 Decennial Census
2000 Decennial Census
Census Bureau Accuracy and Coverage
Evaluation Program
Census Bureau Post Enumeration Survey
Social Services Block Grant
******************************************************************
** This file contains an ASCII representation of the text of a **
** GAO Product. **
** **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced. Tables are included, but **
** may not resemble those in the printed version. **
** **
** Please see the PDF (Portable Document Format) file, when **
** available, for a complete electronic file of the printed **
** document's contents. **
** **
******************************************************************
GAO-06-567
* Report to Congressional Requesters
* June 2006
* FEDERAL ASSISTANCE
* Illustrative Simulations of Using Statistical Population
Estimates for Reallocating Certain Federal Funding
* Contents
* Results in Brief
* Background
* The Top 20 Formula Grant Programs Represented Two Thirds of All
Federal Grant Programs
* Prospective Impact of Statistical Population Estimates on
Medicaid and SSBG Allocations
* States Would Have Shifted 0.23 Percent of Medicaid Funds
Using Statistical Population Estimates Based on the 2000
Census to Recalculate State Medicaid Allocations
* Statistical Population Estimates from the 2000 Census Would
Have Reallocated 0.25 Percent of SSBG Funds Among the States
* Statistical Population Estimates from the 2000 Census Would
Have Shifted a Smaller Percentage of Funding Compared to
Those from the 1990 Census
* Scope and Methodology
* Medicaid
* Social Services Block Grant
* Estimated Reallocations Using Statistical Population Estimates Based
on the 1990 and 2000 Censuses
* Estimated Percentage Changes in State Funding Using Statistical
Population Estimates
* GAO Contact and Acknowledgments
Report to Congressional Requesters
June 2006
FEDERAL ASSISTANCE
Illustrative Simulations of Using Statistical Population Estimates for
Reallocating Certain Federal Funding
Contents
Tables
Figures
June 22, 2006Letter
The Honorable Tom Davis Chairman Committee on Government Reform House of
Representatives
The Honorable Michael Turner Chairman Subcommittee on Federalism and the
Census Committee on Government Reform House of Representatives
Data from the decennial census are used to apportion congressional seats,
redraw congressional districts, and allocate federal assistance to state
and local governments through certain formula grant programs, as well as
for a number of other public- and private-sector purposes. As a result, it
is essential that census data be as complete and accurate as practicable.
Although the U.S. Census Bureau (Bureau) puts forth tremendous effort to
conduct a complete and accurate population count, some amount of error is
inevitable because of the nation's size and demographic complexity, and
the inherent limitations of census-taking methods.
To estimate the accuracy of the 2000 Census and possibly adjust for any
errors, the Bureau conducted the Accuracy and Coverage Evaluation
(A.C.E.), which was an independent sample survey designed to estimate the
number of people that were over- and undercounted in the census, a problem
the Bureau refers to as "coverage error."
In 1999, the Supreme Court ruled that the Census Act prohibited the use of
statistical sampling for purposes of apportioning seats in the House of
Representatives, but the Supreme Court's decision did not specifically
address the use of statistical sampling for other purposes, including
adjusting the data used to allocate federal assistance via formula
grants.1 However, after conducting a number of evaluations of the A.C.E.
samples and methodology, the Bureau concluded that the A.C.E. population
estimates were flawed and that all potential uses of these data would be
inappropriate. As a result, the Bureau's parent agency, the Department of
Commerce, determined that the A.C.E. estimates should not be used for any
purpose that legally requires data from the decennial census. We have also
stated that the A.C.E. statistical population estimates are unreliable.2
Still, a key question that arises is how sensitive are federal formula
grants to alternative population estimates, such as those derived by
statistical methods? With that in mind, you asked us to examine how
statistical population estimates might have redistributed federal
assistance among the states had they been used to calculate formula grants
rather than the actual population counts. In response, and as requested,
we recalculated certain federal assistance to the states using the A.C.E.
population estimates from the 2000 Census, as well as the population
estimates derived from the Post-Enumeration Survey (PES)-the independent
sample survey the Bureau administered to evaluate the accuracy of the 1990
Census. While only the actual census numbers should be used for official
purposes, our analysis shows the extent to which alternative population
counts would impact the distribution of federal grant funds and can help
inform congressional decision making on the design of future censuses. As
agreed with your offices, we identified (1) the top 20 formula grant
programs based on the amount of funds targeted by any means, and (2) how
much money would have been allocated using census data for certain formula
grant programs, and the prospective impact of using estimated population
counts from the 1990 and 2000 Censuses to recalculate state allocations
for these grant programs.
To address the first objective, we used expenditure and obligations data
from the fiscal year 2004 Consolidated Federal Funds Report (CFFR), the
most recent year for which data were available. For the second objective,
we recalculated the amount of federal funding allocated to the states for
Medicaid using the 2005 fiscal year Federal Medical Assistance Percentage
(FMAP) with 2004 state expenditure data, and recalculated the Social
Services Block Grant (SSBG) state allocations using 2005 fiscal year
allocation data.3 In both cases we used the most recent year for which
data were available. We based our calculation on the official population
counts from the 1990 and 2000 Censuses. We then recalculated the
allocations using the PES population estimates from the 1990 Census, and
the A.C.E. population estimates from the 2000 Census.
The federal share of total Medicaid program costs is determined using a
statutory formula that calculates each state's FMAP. SSBG receives an
annual appropriation that is distributed in proportion to each state's
population. We selected these programs for our analysis because they would
be particularly sensitive to alternative population estimates. Medicaid is
the largest formula grant program; thus any changes would redistribute
more money compared to other programs. Further, any changes in SSBG would
have a proportional impact on the distribution of state funds.
Importantly, our analysis only simulates the formula grant reallocations
using A.C.E. data and was done for illustrative purposes-to show the
sensitivity of these two formula grant programs to alternative population
estimates. (See app. I for a more detailed description of our
methodology.) We use the term "allocation" to include Department of Health
and Human Services (HHS) reimbursement to states of Medicaid expenditures
subject to the FMAP formula (net of administrative costs) and SSBG state
allotments. (Appendix II has additional details on Medicaid and the
formula used to allocate money; app. III has the same information for
SSBG.) We use the term "statistical population estimates" to refer to the
results of the coverage measurement programs that the Bureau conducted
following the 1990 and 2000 Censuses (but were not used to adjust the
actual census counts).
We conducted our work between November 2005 and May 2006 in accordance
with generally accepted government auditing standards. Because we did not
evaluate the policies or operations of any federal agency to develop the
information presented in this report, and because we are not making any
recommendations, we did not seek agency comments. However, we discussed a
statement of facts for this report with Census Bureau officials, who
provided us with technical comments that we have incorporated where
appropriate.
Results in Brief
In fiscal year 2004, the top 20 formula grant programs together had $307.9
billion in obligations, or 67 percent of the total $460.2 billion
obligated by the 1,172 federal grant programs. Medicaid was the largest
formula grant program, with obligations of $183.2 billion, or nearly 40
percent of all grant obligations.
The federal government allocated $159.7 billion to states in Medicaid
funds (not including such administrative costs as processing and making
payments to service providers) based on fiscal year 2004 state
expenditures and $1.7 billion in SSBG funds based on fiscal year 2005
state allocations.4 Recalculating these allocations using the A.C.E. and
the 1990 Post-Enumeration Survey statistical population estimates would
have produced the following results:
o A total of 0.23 percent ($368 million) of federal Medicaid funds would
have been shifted overall among the states in fiscal year 2004 and 0.25
percent ($4.2 million) of SSBG funds would have shifted among the states
in fiscal year 2005 as a result of the simulations using statistical
population estimates from the 2000 Census.5
o With respect to Medicaid, 22 states would have received additional
funding, 17 states would have received less funding, and 11 states and the
District of Columbia would have received the same amount of funding using
statistical population estimates from the 2000 Census. Based on a fiscal
year 2004 federal Medicaid allocation to the states of $159.7 billion,
Nevada would have been the largest percentage gainer, with an additional
1.47 percent in funding, and Wisconsin would have lost the greatest
percentage-1.46 percent.
o With respect to SSBG, 27 states and the District of Columbia would have
gained funding, and 23 states would have lost funding using statistical
population estimates from the 2000 Census. Based on a fiscal year 2005
SSBG allocation of $1.7 billion, Washington, D.C. would have been the
biggest percentage gainer, receiving an additional 2.05 percent in
funding, while Minnesota would have lost the greatest percentage
funding-1.17 percent.
o Statistical population estimates from the 2000 Census would have shifted
a smaller percentage of funding compared to those from the 1990 Census
because the difference between the actual and estimated population counts
was smaller in 2000 compared to 1990. For example, using statistical
estimates of the population following the 1990 Census, a total of 0.45
percent of Medicaid funds would have been gained overall by the states in
fiscal year 1997, and 0.37 percent of SSBG funds would have shifted among
the states in fiscal year 1998.
Background
To help measure the quality of the 2000 Census and to possibly adjust for
any over- or undercounts of various demographic groups, the Bureau
designed the A.C.E. program, a separate and independent sample survey
conducted as part of the 2000 Census. When matched to the census data,
A.C.E. data were to enable the Bureau to use statistical estimates of net
coverage errors to adjust final census tabulations. However, in March
2003, after much research and deliberation, the Bureau decided against
using any A.C.E. estimates of coverage error to adjust the 2000 Census,
because of several methodological concerns.
The Bureau measured the accuracy of the 1990 Census as well, and
recommended statistically adjusting the results. However, the Secretary of
Commerce determined that the evidence to support an adjustment was
inconclusive and decided not to adjust the 1990 Census. In 1999 we
examined how these statistical population estimates might have
redistributed federal assistance among the states had they been used to
calculate formula grants.6
Looking toward the 2010 Census, the Bureau plans to use statistical
population estimates to (1) produce estimates of components of census net
and gross coverage error (the latter includes misses and erroneous
enumerations) in order to assess accuracy, (2) determine whether the
strategic goals of the census are met, and (3) identify ways to improve
the design of future censuses. The Bureau does not plan to use statistical
estimates of the population for adjusting census data based on its belief
that the 2000 Census demonstrated "that the science is insufficiently
advanced to allow making statistical adjustment to population counts of a
successful decennial census in which the percentage of error is presumed
to be so small that adjustment would introduce as much or more error than
it was designed to correct."
The Top 20 Formula Grant Programs Represented Two Thirds of All Federal
Grant Programs
In fiscal year 2004, the federal government administered 1,172 grant
programs, with $460.2 billion in combined obligations. However, as shown
in table 1, most of these obligations were concentrated in a small number
of grants. For example, Medicaid was the largest formula grant program,
with federal obligations of $183.2 billion, or nearly 40 percent of all
grant obligations, in fiscal year 2004. The top 20 grant programs
comprised around two-thirds of all federal grant programs, with $307.9
billion in obligations for fiscal year 2004 (SSBG is not included in table
1, because with obligations of $1.7 billion, it is not among the top 20
formula grant programs).
Table 1: The Top 20 Formula Grant Programs and Their Fiscal Year 2004
Federal Obligations
Rank Catalog of Program Federal Percentage
Federal amount in
Domestic billions of top 20
Assistance formula
(CFDA) # grants
1 93.778 Medical Assistance $183.2a 59.5%
Program (Medicaid)
2 20.205 Highway Planning and 31.9 10.4
Construction
3 93.558 Temporary Assistance for 17.2 5.6
Needy Families
4 84.027 Special Education Grants 10.1 3.3
to States
5 84.010 Title I Grants to Local 8.3 2.7
Education Agencies
6 10.555 National School Lunch 7.4 2.4
Program
7 93.600 Head Start 6.6 2.1
8 10.557 Special Supplemental Food 5.0 1.6
Program for Women,
Infants, and Children
9 93.767 State Children's Health 4.9 1.6
Insurance Program
10 93.658 Foster Care Title IV E 4.7 1.5
11 20.507 Federal Transit Formula 3.7 1.2
Grants
12 20.106 Airport Improvement 3.4 1.1
Program
13 14.218 Community Development 3.0 1.0
Block Grants/Entitlement
Grants
14 93.563 Child Support Enforcement 2.9 0.9
15 84.367 Improving Teacher Quality 2.9 0.9
State Grants
16 93.596 Child Care Mandatory and 2.7 0.9
Matching Funds of the
Child Care and
Development Fund
17 84.126 Rehabilitation 2.6 0.8
Services-Vocational
Rehabilitation Grants to
States
18 10.561 State Administrative 2.5 0.8
Matching Grants for Food
Stamp Program
19 14.872 Public Housing Capital 2.5 0.8
Funds
20 17.225 Unemployment Insurance 2.4 0.8
Subtotal, $307.9 100.0%
top 20
programs
Total all $460.2
1,172
programs
grant
programs
Source: GAO analysis of CFFR data.
aCFFR uses direct expenditures or obligations, whereas in the rest of the
report, for our analysis of Medicaid allocations, we use state Medicaid
expenditures subject to the FMAP formula excluding administrative costs.
Prospective Impact of Statistical Population Estimates on Medicaid and
SSBG Allocations
Based on our simulations, recalculating allocations of key programs using
statistical population estimates, states would have shifted less than 0.25
percent of $161.4 billion in Medicaid and SSBG formula grant funding. The
two key programs analyzed-Medicaid and SSBG-together received federal
allocations of $161.4 billion in fiscal year 2004. Federal allocations for
Medicaid (excluding such administrative costs as processing, making
payments to service providers, and monitoring the quality of services to
beneficiaries) were $159.7 billion, by far the highest federal allocation
in fiscal year 2004.7 Using statistical population estimates to
recalculate federal Medicaid allocations to states, states would have
shifted 0.23 percent of $159.7 billion in federal Medicaid funds in fiscal
year 2004 and 0.25 percent of $1.7 billion in SSBG funds would have
shifted as a result of the simulations in fiscal year 2005. (Appendix IV
contains tables showing the difference between using estimated and actual
population data from the 1990 and 2000 Censuses for Medicaid and SSBG.)
Because the two programs allocate state funding using different formulas,
funding reallocations for the two programs may produce results that are
different from one another for a particular state. For example, using the
2000 statistical population estimates, which were lower for Minnesota than
the official census population count, Minnesota's Medicaid allocation
would have remained the same. This is because Medicaid allocations are
subject to a floor, and Minnesota was already receiving the minimum
required reimbursement. However, it would have lost funding under SSBG,
because the statistical population estimates from the 2000 Census, and the
subsequent recalculations, would have reduced funding. In another example,
the District of Columbia allocation would have remained the same for 2000
under Medicaid, because the District of Columbia receives a special rate
that is higher than its calculated rate, but it would have gained funding
under SSBG because its population, as measured by the 2000 Census, was
originally lower than the census population estimates. (For information on
how these formulas are calculated, see app. I.)
States Would Have Shifted 0.23 Percent of Medicaid Funds Using Statistical
Population Estimates Based on the 2000 Census to Recalculate State
Medicaid Allocations
Using statistical population estimates to recalculate federal Medicaid
allocations, states would have shifted 0.23 percent of $159.7 billion of
federal Medicaid funds overall in fiscal year 2004 as a result of the
simulation. If statistical population estimates had been used, of the
overall allocation of $159.7 billion of federal funds, 22 states would
have received more Medicaid funding, 17 states would have received less,
while 11 states and the District of Columbia would have received the same.
The gaining states would have received an additional $208.5 million, and
the losing states would have received $368 million less in funding. Based
on our simulation of the formula funding for Medicaid---Nevada would have
gained 1.47 percent in grant funding and Wisconsin would have lost 1.46
percent. (Appendix IV contains tables showing the difference between using
estimated and actual population data from the 1990 and 2000 Censuses to
recalculate Medicaid allocations.)
Figure 1 shows the state-by-state result-gain or loss--of recalculated
Medicaid grant funding using the statistical population estimates. Most of
the estimated increases in state allocations would have tended to
congregate in the northwestern, southwestern, and southeastern regions of
the country and Hawaii and Alaska. Most of the estimated decreases in
state allocations would have tended to congregate in the northcentral
region of the country. The southeastern and northeastern regions would
have experienced both increases and decreases in funding and all
southeastern states except Florida would have experienced increases.
Figure 1: Estimated Change in Federal Medicaid Grant Funding Using
Statistical Population Estimates for States
Note: Percentage changes are based on a fiscal year 2004 federal Medicaid
allocation of $159.7 billion. These percentages are the result of a
simulation using alternative population estimates and are presented for
illustrative purposes only. Kentucky's percentage change in grant funding
would have been -0.0045 percent.
Figure 2 shows how much (as a percentage) and where Medicaid funding would
have shifted as a result of using statistical population estimates for
recalculating formula grant funding by state. We estimate that 20 states
would have received an increase in allocations from more than 0 to less
than 1 percent, while 2 states would have increased by more than 1
percent. Conversely, 7 states would have experienced a decrease in
allocations of greater than one to less than 1.5 percent; 10 states'
allocations would have decreased by more than 0 to less than 1 percent;
and 11 states and the District of Columbia would have experienced no
change because the shift would have fallen below the floor and above the
ceiling that are built into the FMAP formula.
Figure 2: Estimated Medicaid Percentage Change in Grant Funding Using
Statistical Population Estimates for States
Note: Percentage changes are based on a fiscal year 2004 federal Medicaid
allocation of $159.7 billion. These percentages are the result of a
simulation using alternative population estimates and are presented for
illustrative purposes only. Kentucky's percentage change in grant funding
would have been -0.0045 percent.
Statistical Population Estimates from the 2000 Census Would Have
Reallocated 0.25 Percent of SSBG Funds Among the States
Using statistical population estimates to recalculate federal SSBG
allocations, 0.25 percent of $1.7 billion in SSBG funds would have shifted
in fiscal year 2005 as a result of the simulation. The total $1.7 billion
SSBG allocation would not have changed, because SSBG receives a fixed
annual appropriation. In other words, those states receiving additional
funds would have reduced the funds of other states. In short, 27 states
and the District of Columbia would have gained $4.2 million and 23 states
would have lost a total of $4.2 million. Based on our simulation of the
formula funding for SSBG, Washington, D.C. would have gained 2.05 percent
in grant funding and Minnesota would have lost 1.17 percent. (Appendix IV
contains tables showing the difference between using estimated and actual
population data from the 1990 and 2000 Censuses for SSBG funding.)
Figure 3 shows the state-by-state result-gain or loss-of recalculated SSBG
grant funding using statistical population estimates. Because the
reallocations are based on the same census statistical population
estimates as the Medicaid estimated reallocations, most of the estimated
increases in state allocations would have tended to congregate in the
southeastern, southwestern, and northwestern regions of the country, as
they did in our Medicaid simulation. The estimated decreases would have
been grouped in the northcentral region and several states of the
northeastern region of the country. The northeastern region would also
have experienced both increases and decreases in funding.
Figure 3: Estimated Change in Social Services Block Grant Funding Using
Statistical Population Estimates for States
Note: Percentage changes are based on a fiscal year 2004 SSBG allocation
of $1.7 billion. These percentages are the result of a simulation using
alternative population estimates and are presented for illustrative
purposes only.
Figure 4 shows how much (as a percentage) and where SSBG funding would
have shifted as a result of using statistical population estimates for
recalculating formula grant funding by state. By recalculating SSBG state
allocations using the statistical population estimates for states based on
2003 Census population numbers, we estimate that 27 states would have
experienced an increase from more than 0 to less than 1 percent; the
District of Columbia would have increased by more than 2 percent; 2
states' allocations would have decreased by more than one percent; and 21
states' allocations would have decreased by more than 0 to less than 1
percent.
Figure 4: Estimated Social Services Block Grant Percentage Change in Grant
Funding Using Statistical Population Estimates for States
Note: Percentage changes are based on a fiscal year 2004 SSBG allocation
of $1.7 billion. These percentages are the result of a simulation using
alternative population estimates and are presented for illustrative
purposes only.
Statistical Population Estimates from the 2000 Census Would Have Shifted a
Smaller Percentage of Funding Compared to Those from the 1990 Census
For the Medicaid program, recalculating state allocations using
statistical population estimates based on the 2000 Census would have
changed the funding for 39 states in fiscal year 2004. In particular, 22
states would have increased their allocations by $208.5 million, 17 states
would have decreased them by $368.0 million, and 11 states and the
District of Columbia would have had no change. By contrast, recalculating
state allocations using statistical population estimates based on the 1990
Census, the number of changing states would have remained the same but the
amounts shifting among the states would have changed in fiscal year 1997.
Table 2 presents the comparative information from the two analyses. The
allocations for the gaining states would have decreased by almost 50
percent, from $402.4 million for the 1990 Census to $208.5 million for the
2000 Census, while the allocations for the losing states would have
increased by 7 percent, from $344.6 million to $368.0 million. While total
allocations under the Medicaid program increased by over 75 percent from
fiscal year 1997 to fiscal year 2004, the relative or percentage change in
state funding would have decreased in our simulation of recalculations of
state allocations using statistical population estimates.
Table 2: Comparison of the Changes in Medicaid and SSBG Allocations Using
Statistical Population Estimates Based on the 1990 and 2000 Censuses
Dollars in
thousands
Gaining Losing
states states
Census year Total Number Amount Number Amount Percentage
statistical allocations shift in
population funding
estimates for
programa
Medicaid
Program
2000 Census $159,703,910 22 $208,486 17 $368,035 -0.23
1990 Census 90,002,703 22 402,434 17 344,593 0.45
SSBG
Program
2000 Census 1,690,512 28 4,158 23 4,158 0.25
1990 Census 2,286,151 28 8,514 23 8,514 0.37
Source: GAO analysis of HHS and Commerce data.
a The percentage shift in Medicaid funding was calculated by dividing the
larger of the gaining or losing amounts by total Medicaid allocations in
their respective years. The percentage in SSBG funding was calculated by
dividing the gaining amount by total SSBG allocations. GAO/HEHS-99-69
shows a percentage shift in funding of 0.43 percent, because it divided
the total amount for gaining states by the total Medicaid allotment,
whereas in this report we divided the total amount for gaining states by
the recalculated Medicaid allocation.
We have a similar finding for the SSBG program. Our recalculation of state
allocations would have resulted in a smaller change in allocations when we
compare the results of our recalculation using statistical population
estimates based on the 2000 Census to the results based on the 1990
Census. The change in funding would have been reduced by half using the
statistical population estimates based on the 2000 Census. Total SSBG
state allocations decreased by 26 percent between fiscal year 1998 and
fiscal year 2005, and the percentage shift in funding would also have been
reduced, from 0.37 percent to 0.25, using the statistical population
estimates based on the 2000 Census.
In summary, using the statistical population estimates based on the 2000
Census to recalculate Medicaid and SSBG allocations would have resulted in
a smaller shift in program funding than using the statistical population
estimates based on the 1990 Census. This is because the difference between
the actual and estimated population counts was smaller for the 2000 Census
compared to the 1990 Census. As mentioned earlier, the recalculated
allocations are the result of simulations using statistical population
estimates and were done for the purpose of illustrating the sensitivity of
these two formula grant programs to alternative population estimates.
As agreed with your offices, unless you publicly announce the contents of
this report earlier, we plan no further distribution until 30 days from
its issuance date. At that time we will send copies of the report to other
interested congressional committees, the Secretary of Commerce, the
Secretary of Health and Human Services, the Director of the U.S. Census
Bureau, and the Director of the Office of Management and Budget. We will
make copies available to others upon request. This report will also be
available at no charge on GAO's Web site at http://www.gao.gov .
If you or your staff have any questions concerning this report, please
contact me on (202) 512-6806 or by email at [email protected] . GAO staff
who made major contributions to this report are listed in appendix VI.
Contact points for our Offices of Congressional Relations and Public
Affairs may be found on the last page of this report.
Brenda S. Farrel Acting Director Strategic Issues
Appendix I
Scope and Methodology
As agreed with your offices, we identified (1) the top 20 formula grant
programs based on the amount of funds targeted by any means, and (2) how
much money would have been allocated using census data for certain formula
grant programs, and the prospective impact of using estimated population
counts from the 1990 and 2000 Censuses to recalculate state allocations
for these grant programs. We use the term "allocation" to include
Department of Health and Human Services (HHS) reimbursement to states of
Medicaid expenditures subject to the Federal Medical Assistance Percentage
(FMAP) formula and Social Services Block Grant (SSBG) state allotments. We
use the term "statistical population estimates" to refer to the results of
the coverage measurement programs that the Census Bureau (Bureau)
conducted following the 1990 and 2000 Censuses.
To identify the top 20 formula grant programs based on the amount of funds
targeted by any means, we used fiscal year 2004 grants expenditure and
obligations data from the Bureau's Consolidated Federal Funds Report
(CFFR), the most recent data available at the time of our review. While we
recently reported on inaccuracies in the CFFR,1 we determined that the
CFFR is adequate for purposes of identifying the top 20 federal formula
grant programs because it shows the overall magnitude of these programs.
Because the CFFR lists direct expenditures or obligations, the amount
shown for Medicaid in table 1 is different from the Medicaid allocations
shown in the rest of the report, where we use state expenditure data
subject to the FMAP formula, which exclude administrative costs.
Administrative costs for which Medicaid reimburses states include nine
broad tasks: (1) inform potentially eligible individuals and enroll those
who are eligible, (2) determine what benefits it will cover in what
settings, (3) determine how much it will pay for the benefits it covers
and from whom to buy those services, (4) set standards for providers and
managed care plans from which it will buy covered benefits and contract
with those who meet the standards, (5) process and make payments to
service providers, (6) monitor the quality of services to beneficiaries,
(7) ensure that state and federal health care funds are not spent
improperly or fraudulently, (8) have a process for resolving grievances,
and (9) collect and report information for effective administration and
program accountability.2
To determine how much money was allocated using census population counts
for Medicaid and SSBG, we obtained population and income data from the
Department of Commerce (Commerce). Additionally, we obtained Medicaid
expenditures, SSBG allocations, and certain other information from HHS.
Table 3 displays the census population counts for 1990 and 2000 and their
statistical estimates. We obtained state per capita income-the ratio of
personal income to population-for 2000, 2001, and 2002 from Commerce and
replicated the actual FMAP for 2005 using fiscal year 2004 state
expenditure data. For the SSBG state allocation formula, we obtained state
population estimates for 2003 and replicated the SSBG allocations for
2005. The official 1990 Census population counts and statistical
population estimates from the 1990 coverage measurement program known as
the Post-Enumeration Survey (PES) come from our earlier report.3
Table 3: Census 2000 and 1990 Population Counts and Statistical Population
Estimates
States 1990 Census 1990 PES 2000 Census 2000 A.C.E. statistical
official statistical official estimates
counts estimates counts
Alabama 4,040,587 4,113,810 4,447,100 4,432,192
Alaska 550,043 561,276 626,932 628,774
Arizona 3,665,228 3,754,666 5,130,632 5,114,152
Arkansas 2,350,725 2,392,596 2,673,400 2,670,915
California 29,760,021 30,597,578 33,871,648 33,915,728
Colorado 3,294,394 3,363,637 4,301,261 4,300,930
Connecticut 3,287,116 3,308,343 3,405,565 3,380,307
Delaware 666,168 678,385 783,600 781,132
District of 606,900 628,309 572,059 580,982
Columbia
Florida 12,937,926 13,197,755 15,982,378 15,880,398
Georgia 6,478,216 6,620,641 8,186,453 8,208,427
Hawaii 1,108,229 1,129,170 1,211,537 1,214,225
Idaho 1,006,749 1,029,283 1,293,953 1,288,683
Illinois 11,430,602 11,544,319 12,419,293 12,245,193
Indiana 5,544,159 5,572,057 6,080,485 5,981,091
Iowa 2,776,755 2,788,332 2,926,324 2,884,712
Kansas 2,477,574 2,495,014 2,688,418 2,654,471
Kentucky 3,685,296 3,746,044 4,041,769 4,022,526
Louisiana 4,219,973 4,314,085 4,468,976 4,465,176
Maine 1,227,928 1,237,130 1,274,923 1,259,856
Maryland 4,781,468 4,882,452 5,296,486 5,309,521
Massachusetts 6,016,425 6,045,224 6,349,097 6,285,987
Michigan 9,295,297 9,361,308 9,938,444 9,845,028
Minnesota 4,375,099 4,394,610 4,919,479 4,837,392
Mississippi 2,573,216 2,629,548 2,844,658 2,832,958
Missouri 5,117,073 5,148,974 5,595,211 5,520,892
Montana 799,065 818,348 902,195 906,595
Nebraska 1,578,385 1,588,712 1,711,263 1,697,592
Nevada 1,201,833 1,230,709 1,998,257 2,008,216
New Hampshire 1,109,252 1,118,632 1,235,786 1,221,757
New Jersey 7,730,188 7,774,461 8,414,350 8,370,927
New Mexico 1,515,069 1,563,579 1,819,046 1,819,367
New York 17,990,455 18,262,491 18,976,457 18,928,895
North Carolina 6,628,637 6,754,567 8,049,313 8,037,253
North Dakota 638,800 643,033 642,200 633,176
Ohio 10,847,115 10,921,741 11,353,140 11,210,356
Oklahoma 3,145,585 3,202,963 3,450,654 3,443,913
Oregon 2,842,321 2,896,472 3,421,399 3,409,373
Pennsylvania 11,881,643 11,916,783 12,281,054 12,169,921
Rhode Island 1,003,464 1,004,815 1,048,319 1,036,531
South Carolina 3,486,703 3,559,547 4,012,012 3,997,436
South Dakota 696,004 702,864 754,844 745,278
Tennessee 4,877,185 4,964,261 5,689,283 5,666,047
Texas 16,986,510 17,472,538 20,851,820 20,862,065
Utah 1,722,850 1,753,188 2,233,169 2,230,962
Vermont 562,758 569,100 608,827 602,066
Virginia 6,187,358 6,313,836 7,078,515 7,098,004
Washington 4,866,692 4,958,320 5,894,121 5,881,537
West Virginia 1,793,477 1,819,363 1,808,344 1,795,195
Wisconsin 4,891,769 4,921,871 5,363,675 5,284,299
Wyoming 453,588 463,629 493,782 491,841
United States 248,709,873 252,730,369 281,421,906 280,090,250
Source: Census Bureau.
To analyze the prospective impact of estimated population counts on the
money allocated to the states through these two grant programs, we
recalculated the state allocations using statistical estimates of the
population that were developed for the 1990 and 2000 Censuses in lieu of
the actual census numbers. We used the population estimates, which are
based on the 2000 Census counts, and then adjusted these population
estimates by the difference between the 2000 official population counts
and the statistical estimates of the population (A.C.E.). Our procedure to
simulate the formula allocations using adjusted counts was to (a) obtain
the population estimates used to calculate the Medicaid FMAP and SSBG
allocations, (b) subtract the A.C.E. population estimates from the
official 2000 Census population counts, and (c) add the difference from
(b) to the population estimates from (a). We included the 50 states and
the District of Columbia in our calculations, but did not include the
territories: American Samoa, Guam, the Northern Mariana Islands, Puerto
Rico, and the Virgin Islands, because their allocations use formulas that
are different from those used by the 50 states we analyzed.
To verify our approach, we spoke with Department of Commerce and
Department of Health and Human Services officials who administer these
grant programs about the procedures they use to calculate the formula
funding amounts. Importantly, our analyses of Medicaid and SSBG are
simulations and were conducted only to illustrate the sensitivity of these
two grant programs to alternative population estimates. Both the Census
Bureau and GAO deem the 1990 and 2000 statistical population estimates as
unreliable and they should not be used for any purposes that legally
require data from the decennial census.
Medicaid is an entitlement program. The federal share of total Medicaid
program costs is determined using the FMAP, a statutory formula that
calculates the portion of each state's Medicaid expenditures that the
federal government will pay. Our Medicaid simulation uses the fiscal year
2005 FMAP, which applies 2001 through 2002 personal income and population
data, and fiscal year 2004 expenditure data. The formula calculates the
federal matching rate for each state on the basis of its per capita income
(PCI) in relation to national PCI. States with a low PCI receive a higher
federal matching rate, and states with a high PCI receive a lower rate. If
applying the formula renders a state's reimbursement less than 50 percent
of its allowable expenditures, the state is still entitled to be
reimbursed for a minimum of 50 percent-or "floor"-of what it spent.
Conversely, a state cannot be reimbursed for more than 83 percent of
allowable expenditures-the "ceiling."4 Thus, if one used the A.C.E.
statistical estimates to recalculate state Medicaid allocations, states'
reimbursements for allowable expenditures would not be less than 50
percent, the "floor," or more than the "ceiling." Our calculations do not
include administrative costs, because they are not subject to the FMAP
formula. The Medicaid data we used in our calculations include the Indian
Health and the Family Planning programs, which are not subject to the
allocation formula. Agency officials told us that the expenditures for
these two programs are so small in relation to the total Medicaid
expenditures that they do not materially affect the calculations of state
allocations subject to the FMAP formula.
The SSBG federal grant is for a fixed amount determined in an annual
appropriation, and its formula is set up so that an increase in funding to
any state is offset by a decrease to others. To estimate the prospective
impact of using statistical population estimates to recalculate
allocations for SSBG, we used 2003 population data adjusted by the
difference between the 2000 Census and the A.C.E. estimates and fiscal
year 2005 allocations to the states for our analysis-the data HHS used in
its fiscal year 2005 grant allocations to the states. Unlike Medicaid,
SSBG includes administrative costs in its population-based formula to
calculate state allocations.
Appendix II
Medicaid
CFDA Number: 93.778
Program Objectives: To provide financial assistance to states for payment
of medical care on behalf of cash assistance recipients, children,
pregnant women, and the aged who meet income and resource requirements and
other categorically eligible groups.
Federal Agency: Department of Health and Human Services (HHS), Centers for
Medicare & Medicaid Services.
Fiscal Year 2004 Obligations: $183.2 billion. (Federal allocations
excluding administrative costs: $159.7 billion.)
Formula Calculation: Eligible medical expenses are reimbursed based on the
per capita income of the state. The federal reimbursement rate, known as
the Federal Medical Assistance Percentage (FMAP), ranges from a minimum of
50 percent to a maximum of 83 percent. Most administrative expenses are
reimbursed at a flat rate of 50 percent but may be as high as 100 percent
as is the case with immigration status verification.
Mathematical Structure:
Formula Constraints: No state may receive a matching percentage below 50
percent or in excess of 83 percent.
Definitions:
FMAP = Federal Medical Assistance Percentage.
PCI = Per capita personal income.
PI = Personal income.
Pop = State population.
Data Sources: PI: Department of Commerce, Bureau of Economic Analysis.
Pop: Department of Commerce, Bureau of Economic Analysis, and Census
Bureau.
Amount Shifted: $368 million, or a 0.23 percent overall loss of the total
$159.7 billion allocated among the states as a result of the simulation.
Comments: Allotment amounts were calculated for fiscal year 2004, the
latest year for which data were available. Total federal allotment
includes some amounts for Family Planning and Indian Health Services that
are not subject to the FMAP. We use the term "allocation" to include HHS
reimbursement to states of Medicaid expenditures subject to the federal
FMAP formula (net of administrative costs).
Appendix III
Social Services Block Grant
CFDA Number: 93.667
Program Objectives: To enable states to provide social services directed
toward the following goals: (1) reducing dependency; (2) promoting
self-sufficiency; (3) preventing neglect, abuse, or exploitation of
children and adults; (4) preventing or reducing inappropriate
institutional care; and (5) securing admission or referral for
institutional care when other forms of care are not appropriate.
Federal Agency: Department of Health and Human Services, Administration
for Children and Families.
Fiscal Year 2004 Obligations: $1.7 billion.
Formula Calculation: State funding is allocated in proportion to each
state's share of the national population.
Mathematical Structure:
Formula Constraints: None.
Definitions:
Amt = Funds available for allocation to states.
Pop = A state's population count.
Data Sources:
Amt: Department of Health and Human Services, Administration for Children
and Families.
Pop: Department of Commerce, Census Bureau.
Amount Shifted: $4.2 million, or 0.25 percent of the total $1.7 billion
allocated. The Social Services Block Grant (SSBG) federal grant is for a
fixed amount determined in an annual appropriation; an increase in funding
to any state is offset by a decrease in others.
Comment: We use the term "allocation" to include SSBG state allotments.
SSBG state allotments are based on each state's population in proportion
to the total U.S. population.
Appendix IV
Estimated Reallocations Using Statistical Population Estimates Based on
the 1990 and 2000 Censuses
Table 4: Medicaid Allocations and Recalculated Allocations Using Revised
Population Estimates
The recalculated allocations are the result of a simulation using
statistical population estimates and are presented for illustrative
purposes only.
Source: GAO analysis of HHS and Commerce data.
aThe total amount gained recalculating allocations using statistical
population estimates based on the 1990 Census would have been $402.4
million, the total amount lost would have been $208.5 million, and the
percentage change would have been 0.43 percent.
bThe total amount gained recalculating allocations using statistical
population estimates based on the 2000 Census would have been $208.5
million, the total amount lost would have been $368 million, and the
percentage change would have been -0.23 percent.
Table 5: Social Services Block Grant Allocations and Recalculated
Allocations Using Revised Population Estimates
The recalculated allocations are the result of a simulation using
statistical population estimates and are presented for illustrative
purposes only.
Source: GAO analysis of HHS and Commerce data.
aBecause SSBG has a fixed allocation, the total amount gained and the
total amount lost recalculating allocations using statistical population
estimates based on the 1990 Census would have been the same: $8.5 million,
and the percentage change or shift in funding would have been 0.37
percent.
bBecause SSBG has a fixed allocation, the total amount gained and the
total amount lost recalculating allocations using statistical population
estimates based on the 2000 Census would have been the same: $4.2 million,
and the percentage change or shift in funding would have been 0.25
percent.
cColumn total adds to greater than zero due to rounding.
Appendix V
Estimated Percentage Changes in State Funding Using Statistical Population
Estimates
Table 6: Percentage Difference between 1990 and 2000 Medicaid Allocations
and Recalculated Allocations Using Statistical Population Estimates
States Percentage difference between Percentage difference
Medicaid allocations and between Medicaid
recalculated allocations based on allocations and
the 1990 Census recalculated allocations
based on the 2000 Census
Alabama 0.16 0.11
Alaska 0.00 1.11
Arizona 0.75 0.15
Arkansas 0.11 0.25
California 2.33 0.00
Colorado 0.81 0.00
Connecticut 0.00 0.00
Delaware 0.00 0.31
District of 0.00 0.00
Columbia
Florida 0.58 -0.22
Georgia 0.67 0.95
Hawaii 0.00 0.97
Idaho 0.50 0.06
Illinois 0.00 0.00
Indiana -1.34 -1.40
Iowa -1.34 -1.11
Kansas -1.23 -1.03
Kentucky -0.02 0.00a
Louisiana 0.45 0.31
Maine -0.96 -0.78
Maryland 0.00 0.00
Massachusetts 0.00 0.00
Michigan -1.35 -0.73
Minnesota -1.93 0.00
Mississippi 0.27 0.04
Missouri -1.27 -1.10
Montana 0.61 0.73
Nebraska -1.28 -0.44
Nevada 0.00 1.47
New Hampshire 0.00 0.00
New Jersey 0.00 0.00
New Mexico 1.06 0.33
New York 0.00 0.00
North Carolina 0.23 0.37
North Dakota -0.86 -0.94
Ohio -1.25 -1.08
Oklahoma 0.15 0.23
Oregon 0.34 0.15
Pennsylvania -2.31 -0.76
Rhode Island -2.46 -1.05
South Carolina 0.37 0.09
South Dakota -0.66 -0.83
Tennessee 0.17 0.07
Texas 1.36 0.66
Utah 0.09 0.29
Vermont -0.60 -0.86
Virginia 0.69 0.85
Washington 0.46 0.00
West Virginia -0.14 -0.18
Wisconsin -1.34 -1.46
Wyoming 0.74 0.11
All states 0.06 -0.10
Source: GAO analysis of HHS and Commerce data.
a Kentucky's percentage change in funding would have been -0.0045 percent.
Table 7: Percentage Difference between 1990 and 2000 Social Services Block
Grant Allocations and Recalculated Allocations Using Statistical
Population Estimates
States Percentage difference between Percentage difference
Social Services Block Grant between Social Services
allocations and recalculated Block Grant allocations
allocations based on the 1990 and recalculated
Census allocations based on the
2000 Census
Alabama 0.18 0.13
Alaska 0.40 0.75
Arizona 0.79 0.16
Arkansas 0.15 0.37
California 1.17 0.58
Colorado 0.46 0.45
Connecticut -0.97 -0.27
Delaware 0.20 0.16
District of 1.86 2.05
Columbia
Florida 0.37 -0.14
Georgia 0.56 0.71
Hawaii 0.25 0.67
Idaho 0.60 0.07
Illinois -0.63 -0.92
Indiana -1.11 -1.15
Iowa -1.20 -0.96
Kansas -0.91 -0.79
Kentucky -0.03 -0.01
Louisiana 0.58 0.38
Maine -0.87 -0.70
Maryland 0.49 0.70
Massachusetts -1.14 -0.53
Michigan -0.91 -0.47
Minnesota -1.18 -1.17
Mississippi 0.46 0.05
Missouri -0.99 -0.85
Montana 0.77 0.94
Nebraska -0.96 -0.33
Nevada 0.77 0.91
New Hampshire -0.77 -0.63
New Jersey -1.04 -0.04
New Mexico 1.54 0.48
New York -0.12 0.21
North Carolina 0.21 0.32
North Dakota -0.95 -0.97
Ohio -0.93 -0.79
Oklahoma 0.19 0.27
Oregon 0.27 0.12
Pennsylvania -1.33 -0.44
Rhode Island -1.47 -0.64
South Carolina 0.46 0.11
South Dakota -0.64 -0.80
Tennessee 0.15 0.06
Texas 1.21 0.51
Utah 0.13 0.37
Vermont -0.50 -0.64
Virginia 0.38 0.73
Washington 0.25 0.25
West Virginia -0.19 -0.27
Wisconsin -1.00 -1.00
Wyoming 0.57 0.07
Total 0.00 0.00
Source: GAO analysis of HHS and Commerce data.
Appendix VI
GAO Contact and Acknowledgments
Brenda S. Farrell, (202) 512-6806
In addition to the individual named above, Robert Goldenkoff, Assistant
Director, as well as Faisal Amin, Robert Dinkelmeyer, Carlos Diz, Gregory
Dybalski, Amy Friedlander, and Sonya Phillips made key contributions to
this report.
(450449)
www.gao.gov/cgi-bin/getrpt? GAO-06-567 .
To view the full product, including the scope
and methodology, click on the link above.
For more information, contact Brenda S. Farrell at (202) 512-6806 or
[email protected].
Highlights of GAO-06-567 , a report to congressional requesters
June 2006
FEDERAL ASSISTANCE
Illustrative Simulations of Using Statistical Population Estimates for
Reallocating Certain Federal Funding
Decennial census data need to be as accurate as possible because the
population counts are used for, among other purposes, allocating federal
grants to states and local governments. The U.S. Census Bureau (Bureau)
used statistical methods to estimate the accuracy of 1990 and 2000 Census
data. Because the Bureau considered the estimates unreliable due to
methodological uncertainties, they were not used to adjust the census
results. Still, a key question is how sensitive are federal formula grants
to alternative population estimates, such as those derived from
statistical methods?
GAO was asked to identify (1) the top 20 formula grant programs based on
the amount of funds targeted by any means, and (2) the amount of money
allocated for Medicaid and Social Services Block Grant (SSBG), and the
prospective impact of estimated population counts from the 1990 and 2000
Censuses on state allocations for these two programs. Importantly, as
agreed, GAO's analysis only simulates the formula grant reallocations. We
used fiscal year 2004 Medicaid state expenditure and 2005 SSBG state
allocation data, the most recent data available.
What GAO Recommends
GAO is not making any recommendations in this report.
In fiscal year 2004, the top 20 formula grant programs together had $308
billion in obligations, or 67 percent of the total $460.2 billion
obligated by the 1,172 federal grant programs. Medicaid was the largest
formula grant program, with obligations of $183.2 billion, or nearly 40
percent of all grant obligations. The federal government allocated $159.7
billion to states in Medicaid funds (not including administrative costs
such as processing and making payments to service providers) and $1.7
billion in SSBG funds. Recalculating these allocations using statistical
population estimates from the Accuracy and Coverage Evaluation and the
Post Enumeration Survey-independent sample surveys designed to estimate
the number of people that were over- and undercounted in the 2000 and 1990
Censuses-would have produced the following results:
o A total of 0.23 percent ($368 million) of federal Medicaid
funds would have been shifted overall among the states in fiscal
year 2004 and 0.25 percent ($4.2 million) of SSBG funds would have
shifted among the states in fiscal year 2005 as a result of the
simulations using statistical population estimates from the 2000
Census.
o With respect to Medicaid, 22 states would have received
additional funding, 17 states would have received less funding,
and 11 states and the District of Columbia would have received the
same amount of funding using statistical population estimates from
the 2000 Census. Based on a fiscal year 2004 federal Medicaid
allocation to the states of $159.7 billion, Nevada would have been
the largest percentage gainer, with an additional 1.47 percent in
funding, and Wisconsin would have lost the greatest
percentage-1.46 percent.
o With respect to SSBG, 27 states and the District of Columbia
would have gained funding, and 23 states would have lost funding
using statistical population estimates from the 2000 Census. Based
on a fiscal year 2005 SSBG allocation of $1.7 billion, Washington,
D.C. would have been the biggest percentage gainer, receiving an
additional 2.05 percent in funding, while Minnesota would have
lost the greatest percentage funding-1.17 percent.
o Statistical population estimates from the 2000 Census would
have shifted a smaller percentage of funding compared to those
using the 1990 Census because the difference between the actual
and estimated population counts was smaller in 2000 compared to
1990.
*** End of document. ***