Report to Congress:

Measuring “Need” for HUD’s McKinney-Vento Homeless Competitive Grants

U.S. Department of Housing and Urban Development Office of Community Planning and Development 451 Seventh Street, SW Washington, DC 20410

Table of Contents

Section Page

1. Introduction Background ...... 3 Congressional Direction ...... 3 Conference on Determining Need...... 4 About This Report ...... 4

2. Determining Need in HUD’s Competitive Programs Statutory Authority...... 5 Homeless Competitions: 1987-1993 ...... 5 Staff Rating of Project Need from Application Narratives ...... 5 Continuum of Care Competitions: 1993 – Present ...... 6 Ranking Project Need within a Community-Based System...... 6 The ESG/CDBG Hybrid Formula ...... 6 75/25 Percent Split ...... 7 CoC Formula Demographic Factors ...... 7 Current Approach for Rating Need by Formula and Selecting Projects Competitively ...... 8 Rating Need ...... 8 Total Project Score and Final Project Selection ...... 8

3. Alternative Approaches To Determining Homeless Need Direct Measures of Homeless Need ...... 9 Census Group Quarter Data ...... 9 National Survey Estimates ...... 10 Local Administrative Data on Homeless Needs ...... 10 Indirect Measures of Homeless Need ...... 11 Alternative HUD Formulas ...... 11 Conference Consideration of Alternative Formula Factors of Homelessness ...... 11

4. Findings Findings……………………………………………………………………………..13

Appendix A: Conference Participants…………………………………………15 Appendix B: Conference Agenda………………………………………………20 Appendix C: Conference Summary…………………………………………….23 Appendix D: CDBG, ESG, Home Formula Descriptions……………………34 Appendix E: Four Alternative Allocations Under Current Formulas…….54

2

Section I: Introduction

Background

Since 1987, the McKinney-Vento Homeless Assistance Act has been a major source of Federal assistance to states, local governments, and nonprofit organizations for meeting the needs of homeless individuals and families. The largest portion of McKinney-Vento assistance has been administered by the U.S. Department of Housing and Urban Development (HUD). HUD’s homeless assistance programs include the Emergency Shelter Grants (ESG), Supportive Housing (SHP), Shelter Plus Care (S+C) and Section 8 Moderate Rehabilitation Single Room Occupancy (SRO) programs. . Congressional Direction

In its report (S.R.106-410) on the FY 2001 HUD Appropriations Act, the Senate Committee on Appropriations expressed concern about whether HUD’s use of a variation of an existing program formula as the basis for scoring project “need” was consistent with a competitive program approach. In addition, the Committee raised the question as to whether the factors used in the formula program were a good measure of the extent of homelessness in a jurisdiction.

“The Committee continues to be very concerned over HUD’s administration of the McKinney homeless assistance programs through formula funding to local continuums of care. With the exception of the Emergency Shelter Grants program, the legislation for the Supportive Housing program, Shelter Plus Care and the Section 8 Moderate Rehabilitation SRO program requires a national competition by grantees, not a formula allocation. This is especially troubling since HUD uses a modified allocation formula pursuant to the Community Development Block Grants (CDBG) program to award funding to local continuums of care. The CDBG formula has no real nexus to homeless needs and the use of the CDBG formula also means that local continuums of care are assured of receiving a minimum amount of funds where a grant application meets certain minimum requirements regardless of the actual homeless assistance needs of the jurisdiction….”

The Senate Committee on Appropriations report directed the Department to:

“…convene a group of experts including academics, practitioners, national organization representatives, local and State government officials, and Federal officials to make recommendations to Congress by June 1, 2001 on alternatives to the formula by which “pro rata shares” are determined for local and State jurisdictions in its homeless assistance grants program.…”

3

Conference on Determining Need

In response to the Committee’s direction, HUD held a one-day conference in Washington, DC on April 19, 2001, that focused on alternative approaches to determining need in HUD’s McKinney-Vento programs. A broadly diverse group of forty- five experts attended the conference representing eight national organizations, two local and state governments, six practitioner or provider agencies, four academic institutions, and five Federal government agencies. In addition, four individuals from other organizations attended. Federal government agencies represented included the Department of Commerce, U.S. Census Bureau (Census); Office of Management and Budget (OMB); Federal Emergency Management Agency (FEMA); U.S. Department of Veterans Affairs (VA); and HUD. (Appendix A lists the conference participants by the Congressionally-mandated categories of organizations.)

The Conference reviewed: 1) the statutory provisions concerning factors for competitively rating McKinney-Vento projects; 2) the subjective scoring of need in project rating in competitions from 1987 to 1993; 3) the use of a formula for scoring project need from 1994 to the present; 4) the elements of the formula used for scoring project need in the competition; 5) alternative existing formulas that could be used for scoring project need in the competition; 6) alternative formulas or changes to existing formula factors for determining need; and 7) findings concerning changes to scoring need in the homeless competition. (Appendix B provides the conference agenda; Appendix C provides a conference summary.)

About This Report

This report is divided into four sections:

First, the report reviews how HUD has administratively implemented the statutorily- based need selection criteria for competitively rating projects seeking McKinney-Vento funding under the SHP, S+C and SRO programs. The components of and rationale for using a variation of the CDBG/ESG formula to rank individual project need in the Continuum of Care (CoC) competitive process are described. The process for arriving at a community’s need or “pro rata share” amount and scoring each project’s need is then described.

Second, the report contrasts how using alternative existing HUD formulas (ESG, CDBG and HOME) would affect the scoring of project need in a select number of jurisdictions.

Third, the report outlines alternative direct and indirect measures of homeless need that might be used instead of the current CoC formula or to supplement that formula.

Last, based on the conference proceedings, this report presents findings concerning potential changes to the current use of the CoC formula for scoring project need in a national competition.

4

Section 2: Determining Need in HUD’s Competitive Programs

Statutory Authority

The McKinney-Vento Act establishes need as a selection criteria in each of HUD’s three competitive programs. In the Supportive Housing Program, need is one of six specific selection criteria to be used. In the Shelter Plus Care program need is one of eight selection criteria, and in the SRO program need is one of two selection criteria provided in the statute.

Supportive Housing Program: Section 426 (b) (3)

“the need for the type of project proposed by the applicant in the area to be served;”

Shelter Plus Care Program: Section 455 (a) (3)

“the need for a program for providing housing assistance and supportive services for eligible persons in the area to be served;”

Section 8 SRO: Section 441 (c)

“The amounts made available under this section shall be allocated by the Secretary...on the basis of a national competition to the applicants that best demonstrate a need for the assistance...”

The Secretary is also given broad discretion in both the SHP and S+C programs to use other factors deemed appropriate to carry out the respective program in an “effective and efficient manner.”

The Department has used two approaches for competitively scoring project need in its national competitions. The first approach, operational between 1987 and 1993, used staff to qualitatively assess the need for a project and to give various need points for a project. The second, and current approach, applied since 1994, uses a variation of the CDBG/ESG formula to competitively award project “need points.” Under both approaches, “need points” are allocated to individual projects and these points are added to other competitive selection criteria points for a total project score. Once each project receives need, community plan and leveraging points, it is then rank-ordered in a national competitive system against all other projects.

Homeless Competitions: 1987- 1993

Staff Rating of Project Need from Project Narratives

HUD used a qualitative approach to assign points for the need criterion. An applicant was asked to describe the need for the project for which it was requesting funding. Included in each applicant’s description was an indication of the number of homeless people to be served in the proposed project and their characteristics and needs for housing and supportive services.

5

Upon receipt of an organization’s application, HUD assigned a staff person(s) to review each application and assign a score based on the selection factors published in the NOFA and application. Of the total possible points for a project, need was assigned a score along with other selection criteria. For instance, in 1993 up to 100 points were assigned for need, which represented approximately 13 percent of the total available points.

Reviewers compared application narratives to a set of narrative standards. The major difficulty encountered was in achieving objective application standards and assuring common application of those standards among reviewers. Since we then and still have only a limited understanding of the correlates of homeless need, even the reviewers had limited faith in the process of assigning scores for need.

Though the qualitative approach provided rich information on homeless sub-populations as provided by individual applicants in their narratives, it did not provide HUD sufficient information to understand the relative need of a project. For example, it was not possible for HUD to determine if a project would duplicate services proposed by another applicant in the same community for the very same persons. Moreover, HUD could not assess the relative need of the targeted homeless sub-population compared to other homeless sub-populations in the same community. HUD could not even assess need within the same homeless sub-population since no applicant’s need submission used the same approach. Some would portray need in terms of a need within the community or a neighborhood. Others would portray need in terms of a particular set of circumstances. Others would consider need in terms of the overall homeless population or within a given subgroup of homeless. Finally, and perhaps most significantly, because a pro rata approach was not used to rate homeless need, it was not possible to fairly rate need between communities.

Continuum of Care Competitions: 1994 - Present

Ranking Project Need within a Community-Based Planning and Prioritization System

HUD searched for an approach for scoring the need criterion that was consistent with community-based assessment of needs and planning. In the Continuum of Care competition, individual applicants partner with homeless providers and other interested organizations and individuals to determine the types of projects that are needed and the homeless populations to be served. The output from their joint planning efforts is, among other things, the submission to HUD of the community’s proposed projects for funding, rank ordered from the highest to the lowest priority project. The projects— submitted on a project priority chart in numeric order—represent the collective community’s need for homelessness funding, not an individual organization’s need.

Part of HUD’s challenge in scoring the need criterion in a community-based model, therefore, was to identify a data set that was relevant to measuring homelessness, accurate and timely, and readily available for every jurisdiction in the United States.

6

The ESG/CDBG Hybrid Formula

HUD looked to its existing program need formulas (ESG, CDBG and HOME) as a potential means of measuring need in a more objective community-based approach where a community’s need and the individual projects it submits are measured nationally, relative to others. The ESG formula and its unique universe of recipients was selected as the base formula for two reasons. First, Congress had specified the ESG program as a formula homeless program using the CDBG formula factors and base universe of recipients. Second, ESG’s approximately 320 entitlement recipients were thought to best reflect the predominantly urban focus of homelessness that had been found in established national analyses.

75 Percent ESG Entitlement / 25 Percent Non-Entitlement Split

As a result of the urban focus of homelessness, the CoC formula allocates 75 percent of the need allocation for the annual competition to the approximately 320 largest metropolitan cities and urban counties that are entitled under ESG. The remaining 25 percent is assigned to other metropolitan cities, urban counties and all other counties in the nation. HUD chose to adjust the assignment percentages used in the regular ESG program in order to target resources to the largest metropolitan cities and urban counties, areas experiencing high poverty, overcrowded housing, and other socio- economic conditions that correlate with homelessness. The following table shows the proportions of funds that are allocated to the various recipients under the CoC, ESG and CDBG formulas.

Comparisons of the Allocation of Funds to Recipients Under CoC, ESG and CDBG Formulas FY 2000

CoC ESG CDBG

Metro Cities (MC) Urban Counties (UC)

Number MC/UC 320 320 980

Percent of Allocation 75% 54% 70%

State Balance

State non-entitled/ 50 50 50 MC/UC 660* 660* N/A

Percent of Allocation 25% 46%+ 30%

* These are the CDBG eligible metropolitan cities and urban counties that do not make the .05 percent minimum grant ESG standard. + States have discretion to allocate their funds to MC/UC ESG entitlement areas

7

CoC Formula Demographic Need Factors

Since 1978, the CDBG funding of metropolitan cities, urban counties, and non-entitled areas has been based on a dual formula. Each area receives the greater of two formula allocation amounts (formula A and formula B). These formula factors and the application of the two formulas do not vary between the CoC, ESG, and CDBG formulas. (Appendix D contains a detailed description of the formulas.)

Formula A Formula B

Population (25 percent) Poverty (30 percent) Poverty (50 percent) Age of Housing (50 percent) Overcrowded Growth Lag (20 percent)* Housing (25 percent)

*States use population for growth lag.

Current Approach for Rating Need by Formula and Selecting Projects Competitively

Rating Need

To determine the need for each of the several thousand eligible homeless project applications, HUD establishes a proxy for initial national homeless need. The national need amount used is the level of funding available for the competition. In recent years this figure has been $850 million. This national need figure is then allocated on a pro rata share basis to every CDBG-entitled jurisdiction and county, using the factors of the CoC formula, as described above. If, for example, using the CoC formula factors, River City represented 1 percent of the national homeless need, then it would be assigned $8.5 million of the $850 million as its initial need figure.

A community’s initial pro rata share amount is considered to be its planning figure for the level of funding the community might be awarded. This initial pro rata need amount is adjusted upwards if the community has a particularly large Supportive Housing Program renewal need. An adjustment is made in the pro rata need for communities with very large renewal demands: the pro rata need is increased to reflect funding for at least one year of all renewal projects. Another need adjustment is made if the community submits an eligible, new permanent housing project as its number one priority. This latter adjustment is made to encourage communities to develop more permanent housing and to help achieve the requirement contained in the 2001 HUD Appropriations Act that at least 30 percent of the homeless program appropriation be used for permanent housing.

8

It is important to note that the pro rata need figures do not entitle a community to receive its need amount or any amount of funding in the competition. What the need figures do provide is a means for scoring the need of each project. A community’s top priority projects receive a full 40 points in need, whereas lower rated projects receive a lower (e.g., 15 or 10 points) score for need. The need amount determines which need score each project receives. In the River City example, the first $8.5 million in project requests on the local project priority list would receive the maximum 40 need points. The next $8.5 million in requests would receive 15 points for need, and any remaining projects would receive 10 points in need.

Total Project Score and Final Project Selection

Need is one of several scores that each eligible project receives in order to determine which projects will be awarded funding. The other scores are for coordination (maximum of 55 points) and leveraging of other resources (maximum 5 points). As with need, these are statutory rating factors. Up to 2 additional points are given to every project where there are significant linkages and coordination with the local Empowerment Zone/Enterprise Community (EZ/EC) planning process. Thus, the maximum score a project can receive is 102 points.

The need, coordination, leveraging, and EZ/EC scores are combined to produce the total score for each project. Each project is then arrayed nationally by score, from highest to lowest. HUD selects projects, in order by score, until all funds have been committed. HUD then determines if the projects selected satisfy the 30 percent permanent housing requirement. If not, HUD replaces otherwise selected new non-permanent housing projects with lower-rated permanent housing projects. (Appendix C contains a detailed explanation with examples of the how pro rata shares are used to rate an individual project’s need in the national competition.)

Section 3: Alternative Approaches To Determining Homeless Need

The Senate Report raised two primary concerns. First, the Committee expressed concern that the factors used in the CDBG program have no nexus to homeless need. Second, the Committee was disturbed that by using a formula, HUD assures funding to communities regardless of these communities having actual homeless need.

To address these concerns the Committee directed HUD to convene a group of experts and identify alternatives to the current CoC formula used by the Department. As part of this effort, it was agreed that two threshold standards must govern any consideration of alternatives:

• The data must be readily available for all potentially eligible communities; and

• The data must provide a generally accurate depiction of homeless need.

There are two basic categories of data that can be examined. The best measures of need for McKinney-Vento assistance are those that measure homelessness directly. This involves data from actual counts of homeless persons or estimates of homeless

9

persons from reliable surveys. The second category of data—where direct counts or estimates do not exist or are problematic—is based upon factors that indirectly measure the need for homeless assistance. The following describes direct and indirect measures of homelessness.

Direct Measures of Homeless Need

Census Data Collection

The most logical data set for measuring homelessness directly would include relevant Census data. The Census Bureau, however, does not produce counts of homeless people. In Census 2000, the Census Bureau used enumeration methods that targeted living quarters—either a housing unit or group quarters. The latter housing type encompasses homeless shelters, prisons, and other group settings. The Census Bureau also enumerated people at service-based operations including soup kitchens, regularly scheduled mobile food vans, and targeted non-sheltered outdoor locations. Enumeration occurred for one night at shelters, one day at soup kitchens and vans, and one day at targeted non-sheltered locations.

The Census Bureau has changed its data tabulations since the 1990 Census. In the 1990 Census, people enumerated at emergency shelters and at visible in-street locations were displayed separately. Census 2000, will not show the results of service- based enumeration separately. This change in tabulation methods was based on several factors:

First, the Census does not produce counts of homeless persons because of the limitations to the data collected in the census. The Census produces a one-time snapshot of “putting people in a location.” Because there are different definitions of homeless and different settings and housing environments, homeless will be included in the Census 2000, but not separately identified in the tabulations produced by Census. To produce an accurate count of homeless people, the Census Bureau or another organization would need to conduct a targeted study of people who are homeless and visit the same sites more than once. In addition, other methods would need to be used, such as general and sample surveys, administrative records, and other datasets.

Second, the Census wants to discourage people from taking the results from the service-based enumeration in an attempt to draw conclusions from the services available in a given community. Last, the Census does not want people to use the data as a measure of service usage.

The Census Bureau will release, starting in June 2001, Summary File 1 which presents counts and basic cross-tabulations collected from all people and housing units. Included in the “other non-institutional group quarters” category will be enumeration figures for emergency shelters and transitional shelters, soup kitchens, regularly scheduled mobile food vans, targeted non-sheltered outdoor locations, and other types of group quarters. This decision was made to reinforce the Bureau’s position that Census 2000 would not provide an account of the homeless population or of the service-using homeless population.

1 0

On a related note, the shelter count portion of the special 1990 Shelter-Night (S-Night) Census effort was generally considered to be quite accurate. The 1992 HUD report to Congress, entitled Allocating Homeless Assistance by Formula, recommended using the shelter portion of the 1990 S-Night count as the best direct measure of the incidence of homelessness in every jurisdiction for a formula allocation. However, the use of the 1990 shelter count found little support among Conference attendees for its use for allocating funds in any proposed formula program, except for the limited use of measuring other proposed indirect measures of homeless need.

A second survey to be conducted by the Census Bureau, the American Community Survey, was discussed. People living in shelters will be part of the sample, and the survey will include characteristics such as poverty status, but emergency shelters will not be identified separately (just as for Census 2000). The plan is that it will be fully implemented in every county, if funding is provided by Congress. This survey will provide demographic, social, economic and housing profiles that will be available for areas where 65,000 or more people reside. Starting in 2004, in 2008, and every year thereafter, the information will be available for areas where less than 20,000 people reside (including census tracts and rural areas). One can use the profiles in conjunction with summaries from local administrative records or other surveys matched to the same geographic areas (not to individual people). Profiles for every state and jurisdiction of 250,000 or more people will be available starting in July 2001 and will continue in 2002 and 2003.

National Homeless Survey Estimates

The National Survey of Homeless Assistance Providers and Clients was conducted in 1996. The survey was designed and funded by 12 federal agencies under the auspices of the Interagency Council on the Homeless. The survey was based upon a statistical sample of 76 metropolitan and non-metropolitan areas, including small cities and rural areas. While the survey was not designed to produce a national count of homeless persons, an independent national estimate of homeless persons was generated from the data by Martha Burt of the Urban Institute. However, no statistically reliable local estimates can be determined from this national survey.

Local Administrative Data on Homeless Needs Required by HUD

Consolidated Plan/CoC Gaps Analysis Data. HUD requires homeless service counts and estimates from every CDBG entitlement and non-entitlement jurisdiction as a part of the Consolidated Plan (ConPlan) process. In addition, current homeless service use and need estimates are required of every applicant community in applying for McKinney- Vento funding. While HUD sets minimum standards (it must be a point-in-time estimate and include shelter and street counts), there is variability in the quality of the data collected by communities.

Homeless Management Information Systems (HMIS) Data. A number of communities have implemented comprehensive homeless management information systems providing for client-intake and tracking across a system of homeless services. A HMIS has the capability to generate an unduplicated count of homeless people for a point in time and over a year or more. These systems also have the potential to calculate shelter turnover rates, develop client profile demographics and service use, and document project and program outcomes over time. A HMIS can be used as a vehicle

1 1

to streamline reporting to HUD and other funding agencies and facilitate agency- and program-level evaluation. Congress has directed HUD to assess the data for jurisdictions that have implemented comprehensive HMISs. In addition, Congress has directed jurisdictions to collect unduplicated client information by 2004.

Using ConPlan and Continuum of Care application data would be problematic at this time because of the potential variations in the quality of the estimates contained in the current plans. Any effort to collect data for funding allocation purposes would require agreements on the rigorous methods for conducting local counts and supervision by HUD. In addition, while the use of HMIS data may potentially be able to generate more rigorous counts, it will be at least several years before every jurisdiction will be systematically collecting unduplicated client-level data at an acceptable level of coverage.

Indirect Measures of Homeless Need

As described earlier, the Department selected an indirect measure of homeless need based upon a hybrid of the ESG/CDBG formula and recipient base for two reasons. First, the Department sought to follow Congress’ lead when Congress selected the CDBG formula (with a grant minimum) for distributing ESG funds. Second, the Department increased the targeting of the ESG/CDBG formula by splitting the funds 75/25 percent (ESG entitlements/Balance of recipients) to reflect the acknowledged urban focus of homelessness.

Alternative HUD Formulas

The differences in the existing CoC formula for assigning need points and the potential use of three other formulas (ESG, CDBG, and the HOME) was presented at the conference. (Attachment E presents information on alternative allocations.) A printout of “pro rata shares” for communities was generated under these alternatives and distributed to the Conference participants. No conclusions were reached about any of these alternative formulas. However, the significant variation in funding that would result in adopting another need formula prompted a number of comments that choosing any alternative would likely generate a divisive “formula fight” and result in no change from the present administrative process. In summary, the consensus of the conference participants about the current formula and other alternative formulae was: (1) there are A limited number of need-related factors for which we currently have comprehensive national data; (2) other formulae that have been developed have proved only marginally better than the current formula at capturing homeless need, at least so far as we are able to develop measurements of homeless need; (3) in the absence of any agreed upon surrogates for homeless need, we do not have a basis for unqualifiedly establishing an alternative formula; and (4) the current formula does have the major advantage of legitimacy or acceptance after several years use in determining pro rata need.

Conference Consideration of Alternative Factors of Homeless Need

During the Conference, HUD solicited from experts in attendance other factors that they thought might be better indirect measures of homelessness. These factors are shown below. There was no single factor that all attendees agreed upon. Most Conference attendees thought that these indices merited further analysis, especially when 2000

1 2

census data were available. However, a majority of attendees indicated that a factor addressing the affordability of local housing should be incorporated in the competition need formula. These comments tended to be made by representatives of places with long-standing and emerging housing affordability issues and who believed that this factor would allocate more resources to their jurisdiction. Support for this new affordability formula element was also expressed by several public and private housing experts.

Prospective Factors to Supplement the Factors in CoC Formula Factor Description: 1. Fair Market Rents (FMRs) versus the ability to pay rent, or income. 2. The cost of housing versus income. 3. Income inequality. 4. Percent of one-person households. 5. Immigration and emigration. 6. Percent of Supplemental Security Income (SSI) recipients. 7. Loss of housing. 8. Condition of housing stock, including whether the stock is severely substandard. 9. Affordability of housing, both the production and access. 10. Voucher (Section 8) utilization rates. 11. Number of people leaving institutions. 12. Unemployment. 13. Number of very low-income renters paying more than 30 percent of their income for rent. 14. Shortages of affordable and available number of units affordable to extremely low-income people per 100 renters.

Section 4: Findings

1. There was general support for the existing community-based planning concept and the rating of project need based upon a relative national need index using reliable and objective data that, in the absence of available, reliable and accepted direct measures, indirectly measures present or emerging homeless needs.

2. There were varying opinions about the best proxy measures of homeless need.

3. Shifting from the current CoC 75/25 percent modified ESG formula for determining relative national need to another HUD allocation formula, such as the regular ESG, CDBG, or HOME formulas, raised concerns over the significant changes to the current amount of funds awarded to communities.

4. Any drastic change in the formula for determining need, assuming level funding, would result in significant impacts to the existing structure of local homeless

1 3

programs, as some communities would see increases at the expense of other communities. This impact would be particularly acute since meeting funding demands for renewal projects would likely absorb nearly all available funds in future competitions.

5. There was overall agreement that the current approach used to assess community need for McKinney-Vento funding is working well and that there should be no significant disruption to existing programs and the community-based planning process generated by a major change in methodology.

6. However, a majority of attendees indicated that a factor addressing the affordability of local housing should be incorporated in the competition need formula.

1 4

APPENDIX A:

Conference Participants

1 5

NATIONAL ORGANIZATIONS

Steven Roger Berg Andrew T. McMahon Mary Louise Stover National Alliance to End COSCDA Housing Assistance Council Homelessness 444 North Capitol Street, 1025 Vermont Avenue, NW 1518 K. Street, NW Suite 224 #606 Suite 206 Washington, DC 20001 Washington, DC 20005 Washington, DC 20001 Phone: 202–624–3631 Phone: 202–842–8600 Phone: 202–638–1526 Fax: 202–624–3639 Fax: 202–347–3441 Fax: 202–638–4664

Sheila Crowley Brad Paul Sue Watlov Phillips National Low Income National Coalition for the National Coalition for the Housing Coalition Homeless Homeless 1012 14th Street, NW 1012 14th Street, NW 1012 14th Street, NW Suite 610 Suite 600 Suite 600 Washington, DC 20005 Washington, DC 20005 Washington, DC 20005 Phone: 202–662–1530 Phone: 202–737–6444 Phone: 202–737–6444 Fax: 202–393–1973 Fax: 202–737–6445 Fax: 202–737–6445

Eugene Lowe Marcia Sigal Vickie E. Watson U.S. Conference of Mayors NAHRO NCDA 1620 Eye Street, NW 630 Eye Street, NW 522 21st Street, NW Suite 400 Washington, DC 20001 Suite 120 Washington, DC 20036 Phone: 202–289–3500 Washington, DC 20006 Phone: 202–861–6710 Fax: 202–289–4961 Phone: 202–887–5532 Fax: 202–293–2352 Fax: 202–887–5546

LOCAL OR STATE AGENCY OFFICIALS

Kelley A. Cronin Patrick Thomas Leary Emergency Shelter MN Department of Children, Commission Families, and Learning Boston City Hall, Room 716 1500 West Highway 36 Boston, MA 02201 Roseville, MN 55113 Phone: 617–635–4507 Phone: 651–582–8349 Fax: 671–635–4507 Fax: 651–582–5491

1 6

PRACTIONER/PROVIDER AGENCY STAFF

Martha A. Fleetwood Jonathan Harwitz Philip Mangano Home Base Corporation for Supportive MA Housing and Shelter 870 Market Street, Suite 1228 Housing Alliance San Francisco, CA 94102 50 Broadway, 17th Floor Five Park Street Phone: 415–788–7961 New York, NY 10004 Boston, MA 02108 Fax: 415–788–7965 Phone: 212–986–2966 Phone: 617–367–6447 Fax: 212–986–6552 Fax: 617–367–5709

James Forsberg Fred Karnas Sue A. Marshall NCHV Consultant Community Partnership for 526 North Ivy Street 20001 Columbia Pike, #706 the Prevention of Arlington, VA 22201 Arlington, VA 22204 Homelessness Phone: 703–525–6256 Phone: 703–920–5276 801 Pennsylvania Avenue,SE, Fax: 703–525–5132 Suite 360 Washington, DC 20003 Phone: 202–543–5298 Fax: 202–543–5663

ACADEMIC/PUBLIC POLICY

Martha R. Burt Stephen Metraux Urban Institute University of PA, CMHPSR 2100 M Street, NW 3600 Market Street Washington, DC 20037 Philadelphia, PA 19104 Phone: 202–463–5551 Phone: 215–349–8487 Fax: 202–463–8522

Donna Haig Friedman Cynthia Taeuber Center for Social University of Baltimore/U.S. Policy/McCormack Institute Census Bureau 100 Morrissey Boulevard 1420 North Charles Street Boston, MA 02125 Baltimore, MD 21201 Phone: 617–287–5565 Phone: 410–837–6551 Fax: 617–287–5544 Fax: 410–837–5814

1 7

FEDERAL AGENCY STAFF OR OFFICIALS

Donna Abbenante John Garrity Kevin Neary U.S. Department of HUD U.S. Department of HUD U.S. Department of HUD 451 Seventh Street, SW 451 Seventh Street, SW 451 Seventh Street, SW Washington, DC 20410 Washington, DC 20410 Room 8140 Phone: 202–269–2690 Phone: 202–708–4300 Washington, DC 20410 Fax: 202–708–3617 Phone: 202–708–3700 Fax: 202–708–5873

Lauren Bloomquist Yolanda Lynne Gaston Kathy Nelson Office of Management and FEMA U.S. Department of HUD Budget 500 C Street, SW 451 Seventh Street, SW 725 17th Street, NW Washington, DC 20410 Room 8120 Room 9226 Phone: 202–646–4543 Washington, DC 20410 Washington, DC 20503 Fax: 202–646–4371 Phone: 202–708–1520 Phone: 202–395–4610 Fax: 202–708–0573 Fax: 202–395–1307

Elaine Braverman Doris Hill F. Stevens Redburn U.S. Department of HUD U.S. Department of HUD Office of Management and 451 Seventh Street, SW 451 Seventh Street, SW Budget Washington, DC 20410 Washington, DC 20410 725 17th Street, NW Phone: 202–708–2140 Phone: 202–708–0614 x4398 Room 9226 Fax: 202–708–3617 Washington, DC 20503 Phone: 202–395–4610 Fax: 202–395–1307

Annetta Clark-Smith Mark Johnston Mike Roanhouse U.S. Bureau of the Census U.S. Department of HUD U.S. Department of HUD Building Three, Room 2384 451 Seventh Street, SW 451 Seventh Street, SW Washington, DC 20233 Room 7258 Room 7258 Phone: 301–457–2378 Washington, DC 20410 Washington, DC 20410 Fax: 301–457–6634 Phone: 202–708–4300 Phone: 202–708–4300 Fax: 202–708–3617 Fax: 202–708–3617

Carol B. Coleman Marge Martin Denise Smith FEMA U.S. Department of HUD U.S. Census Bureau 500 C Street, SW 451 Seventh Street, SW Building Three, Room 2384 Washington, DC 20472 Room 8110 Washington, DC 20233 Phone: 202–646–3107 Washington, DC 20410 Phone: 301–457–2378 Fax: 202–646–4317 Phone: 202–708–1520 Fax: 301–457–6634 Fax: 202–708–0309

1 8

Paul B. Dornan Robert E. Meehan Joe Ventrone U.S. Department of HUD U.S. Department of HUD U.S. Department of HUD 451 Seventh Street, SW 451 Seventh Street, SW 451 Seventh Street, SW Washington, DC 20410 Washington, DC 20410 Room 10218 Phone: 202–708–0574 Phone: 202–708–0790 Washington, DC 20410 Fax: 202–708–5873 Phone: 202–708–0636 Fax: 202–708–2689

Peter H. Dougherty John A. Nagoski U.S. Department of Veterans Consultant Affairs U.S. Department of HUD 810 Vermont Avenue, NW Mail 380 Aquia Creek Road Stop 075 Stafford, VA 22554 Washington, DC 20420 Phone: 540–657–7766 Phone: 202–273–5774 Fax: 202–273–9472

OTHER INDIVIDUALS

Ray Heer Tony Russo Alderson Reporting Consultant 1111 14th Street, NW Route One, Box 63 Fourth Floor Shenandoah Junction, WV Washington, DC 20005 25544 Phone: 202–289–2260 Phone: 703–346–6229 Fax: 202–289–2221 Fax: 304–876–1885

Cynthia Hernan Debby Shinnebarger Aspen Systems Corporation Aspen Systems Corporation 2277 Research Boulevard 2277 Research Boulevard Mail Stop 6G Mail Stop 6E Rockville, MD 20850 Rockville, MD 20850 Phone: 301–519–5302 Phone: 301–519–5574 Fax: 301–519–6655 Fax: 301–519–5161

1 9

APPENDIX B:

Conference Agenda

2 0

Determining the Need for Assistance in HUD’s McKinney-Vento Homeless Competitive Programs Conference

April 19, 2001 9:00 a.m. – 4:15 p.m.

U.S. Department of Housing and Urban Development Departmental Conference Room, Room 10233 451 Seventh Street, SW Washington, DC 20410

AGENDA

8:30 a.m. Registration

9:00 – 9:15 Welcome and Overview Donna M. Abbenante, Acting General Deputy Assistant Secretary

• Welcome, introductions • Meeting purpose • Meeting format, ground rules and agenda: • Tony Russo, conference facilitator

9:15 – 10:15 Assigning need in the Competition John D. Garrity, Director, Office of Special Needs Assistance Programs (SNAPs); Mark Johnston, Division Director, SNAPs

• History • Continuum of Care (CoC) and current method • Discussion, Q and A Break (15 min.)

10:30 – 11:45 Existing Need Formulas Bob Meehan, Senior Advisor, HUD; John Nagoski, Consultant, HUD

• CDBG, ESG, HOME and CoC • Discussion, Q and A

11:45 – 1:00 p.m. Lunch (On your own)

1:00 – 2:15 Measures of Homelessness • Census 2000 - Annetta Clark-Smith and

2 1

• Denise Smith, Population Division, Bureau of the Census, Department of Commerce • Administrative Data - Donna Haig Friedman, Center for Social Policy, McCormack Institute, University of MA-Boston; Steve Metraux, Center for Mental Health Policy and Social Research, University of PA • Discussion, Q and A

Break (15 min.)

2:30 – 3:00 Measures of Homelessness (continued) • Indirect Measures- Martha Burt, Urban Institute • Discussion, Q and A

3:00 – 4:15 Feedback and Recommendations • Facilitated discussion: How to improve the current method using a four-prong filter of objectivity, accuracy, accessibility and timeliness • Next steps, closing comments, HUD

2 2

APPENDIX C:

Conference Summary

2 3

Current Continuum of Care Approach

Prior to the advent of Continuum of Care planning, HUD used a qualitative approach to assign points for need for homeless grant funding. At that time need was also weighed less as a percentage of the overall points assigned in the annual competition. Though the qualitative approach resulted in rich information on homeless sub-populations as provided by project applicants in their applications for funding, the need for the project on a community level was unknown. In addition, because a pro rata approach was not used to match resources with intensity of needs, it was not possible to ensure communities received funding commensurate and appropriate to levels of poverty and other socio-economic factors that correlate with homelessness.

Starting in 1993 with the advent of Continuum of Care, HUD turned to the formula factors used in the Community Development Block Grant (CDBG) and Emergency Shelter Grants (ESG) Programs as the best proxies to determine a community’s need for homeless grant funding. Under the Continuum of Care approach, a planning community is assigned an initial, or preliminary, pro rata need figure based on the higher of two formulas used in the programs. The demographic factors for Formula A are population, poverty and overcrowded housing, weighted at .25, .5, and .25 respectively. The factors for Formula B are population growth lag from 1960, poverty and pre-1940 housing, weighted .3, .3, and .5 respectively.

Under the Continuum of Care approach, a community is assigned a numeric planning figure equal to the approximate amount of funding it could receive in the annual competition for SHP, S+C and SRO Program funding. These planning figures are published annually in conjunction with HUD’s publication of the Notice of Funding Availability (NOFA) and application. Experience has shown that providing communities with planning estimates during their Continuum planning and application preparation phase assists communities in making local decisions about the scope and level of projects to apply for, and priorities for funding.

Using the Continuum of Care method, 75 percent of the money available in the annual competition is assigned to metropolitan cities and urban counties that are entitled under ESG. The remaining 25 percent is assigned to other metropolitan cities, urban counties and states. Within these categories, HUD calculates a “relative need index” for each jurisdiction and applies the index against the total amount of funding available nationally in each year’s competition to determine the preliminary pro rata need for each geographic area. The goal of this approach is to target resources to jurisdictions that are experiencing high poverty, overcrowded housing, and other socio-economic conditions that correlate with homelessness.

In the FY 2001 competition, HUD will assign up to 102 points to each project requesting funding: sixty-two points for Continuum of Care and 40 points for need. For the Continuum of Care score, 30 of the available points will be awarded based on the existence of a coordinated and inclusive community process and structure that develops a comprehensive strategy to address gaps in the current delivery system. Twenty points will be awarded based on the community’s use of data to identify gaps in their current inventory of housing and supportive services and based on the identified gaps, the completion of a project priority setting process that rank orders the community’s

2 4

proposed projects. Ten points will be awarded based on the extent to which the application incorporates mainstream resources and leverages HUD’s project funding. Last, HUD can award up to two bonus points to the Continuum of Care score when at least one proposed project will be located within the boundaries and/or will principally serve the residents of a federal Empowerment Zone (EZ) or Enterprise Community (EC).

Following application submission, HUD reviews each application to determine which jurisdictions worked together during the planning cycle to develop a Continuum of Care planning system. HUD aggregates the preliminary pro rata need numbers for all jurisdictions under each Continuum of Care system. This process results in the assignment of one planning amount to each system. For example:

Example of Initial PRN: Syracuse: .0035 x $850M = $2.9M Onondaga Co.: .0005 x $850M = $.4M Oswego Co.: .0005 x $850M = $.4M

Total Syracuse CoC System = $3.7M

HUD then adjusts for renewals and adds a bonus amount of $500,000 to a Continuum’s pro rata amount for new permanent housing projects that are identified as a community’s top priority for funding.

HUD takes each community’s final pro rata need amount and applies it against the requested amount (as adjusted where necessary) of each project on the community’s Project Priority chart. Starting with project priority #1 and proceeding down the chart, skipping individual projects rejected during the threshold review, projects whose requested amounts fall fully within the applicant’s CoC pro rata need amount, as adjusted (“first tier”), or those where more than one-half the requested amount falls within this “first tier,” receive the full 40 points available for Need. Continuing down the list, those projects whose requested amounts fall fully within the “second tier” (two times the pro rata need amount, as adjusted), or those where more than one-half the requested amount falls within the “second tier,” receive 15 points. Any remaining projects on the priority list each receive 10 points. Example for the Syracuse Continuum of Care:

Priority Request Need Points #1 $1,850,000 40 #2 $1,850,000 40 #3 $1,850,000 15 #4 $1,850,000 15 #5 $1,850,000 10

There are several pros and cons of the pro rata need and Continuum of Care approach. The benefits include that it:

• Allows a community to understand its potential allocation if successful in the competition; • Reflects community-wide need for homeless assistance, not just for a project; • Reflects a community’s determination of which projects have greatest need and rewards those; and

2 5

• Is objectively determined ensuring uniformity of application.

The cons to the approach include that it:

• Applies a formula that is not reflective of true homeless need; • Conveys a sense of entitlement to funding amount by applicant communities; and • Is difficult to explain/understand, especially by new applicants.

Existing Need Formulas

In weighing decisions to use a formula as a measure of need or as a distribution vehicle for homelessness funding, it is important to use criteria and define what one is trying to achieve and measure.

Community Development Block Grant (CDBG) Program

The Community Development Block Grant program provides annual allocations to eligible cities and counties and to states for areas that are not entitled to receive funds directly. As specified in sections 102 and 106 of the Housing and Community Development Act of 1974, the program allocates funds based on demographic data provided by the Census Bureau.

After setting aside funds for special purposes such as technical assistance, the annual appropriation for CDBG formula funding is split so that 70 percent is allocated among eligible metropolitan cities and counties and 30 percent among the states.

Although entitlement communities and states now receive funds under a dual formula system, the CDBG program first allocated funds in FY 1975 under a single formula. The block grant phased in this formula for eligible communities to replace funding for a group of special purpose, “categorical” competitive programs. For FY 1978, the program added a second formula that was intended to assist the older declining communities that would have had funding phased down under the initial formula. Communities and states receive funds under whichever formula provides the higher amount. Since FY 1980 when the dual formula was fully phased in, the allocation formula has been fundamentally unchanged.

The demographic factors for the initial formula (“formula A”) are population, poverty and overcrowded housing, weighted at .25, .5, and .25 respectively. The factors for the second formula (“formula B”) are population growth lag from 1960, poverty and pre-1940 housing, weighted .3, .3, and .5 respectively. For states, population replaces population growth lag as a factor in the second formula.

The data used to compute the formula are proxies for various indicators of community need. In 1980 and 1990 HUD published studies regarding how well the decennial data targets funds to needy communities. In both studies HUD determined that the data continued to target the funds to the most needy communities and recommended continuing the dual formula as specified in statute.

2 6

In general, the formula A factors of population, poverty and over crowded housing help target funds to communities that are experiencing the strains of population growth. Formula B factors of population growth lag, poverty and pre-1940 housing targets funds to the older and declining communities.

Section 106 of the Housing and Community Development Act requires that HUD allocations use a “share” method that computes formula amounts based on the ratio of an entitlement community’s need to an overall need value. The “shares” for each demographic need factor are multiplied by weights to reflect relative importance and then summed for formula A and formula B. The sums are multiplied by the amount available for distribution to give the two formula amounts.

While communities and states receive funding based on the greater of the two formula amounts, there is a final adjustment to the greater formula amount before the final allocation is determined. Since the sum of the higher amounts typically exceeds the total available for funding, the higher amounts are pro rata reduced.

Emergency Shelter Grants (ESG) Program

The Emergency Shelter Grant program provides formula funds to entitlement communities and states for emergency shelters and essential services for the homeless. ESG is authorized under Subtitle B of title IV of the Stewart B. McKinney Act (42 U.S.C. 11371 et seq).

The program funds four types of facility and service support activities to help in sheltering the homeless as well as assisting in the transition to permanent housing. The types of allowable expenses for ESG are rehabilitation, operation of facilities, delivery of essential services and homeless prevention.

Appropriated funds for ESG have held constant at about $150 million from FY 1998 to FY 2001.

Congress selected the CDBG formula as the basis for allocating ESG funds. The CDBG formula is an established funding formula that targets funds to large communities, particularly older and declining communities, with significant needs for homeless assistance.

The statute requires that initial allocations for ESG be proportionate to funding among states and entitlement communities under CDBG formula for the prior year. The eligibility for ESG funding is limited to those CDBG entitlement communities that receive more than 0.05 percent of the amounts appropriated.

After setting aside 0.2 percent of the appropriation to territories and insular areas, the ESG formula distributes the initial funds so that 70 percent goes to entitlement communities and 30 percent goes to states. However, as the initial allocations for CDBG entitlement communities with below threshold funding are transferred to the state pot, actual split is shifted to no less than 70 percent for entitlement communities. For FY 2001, the actual percent allocated to entitlement communities is 54 percent. States have the discretion to allocate their ESG funds to metropolitan and urban county jurisdictions that are receiving ESG entitlement allocations.

2 7

HOME Investment Partnership Act (HOME) Program

The “HOME Investment Partnership Act” (HOME), authorized under Title II of the National Affordable Housing Act of 1990, allocates formula funds to participating jurisdictions to increase the number of families served with decent, safe, sanitary, and affordable housing and to expand the long-term supply of affordable housing.

Participating jurisdictions (PJs) that receive a share of formula funds must submit annual plans on how they expect to use the funds. PJs make their plans available for public review and comment.

HOME funds, which are made available for allocation, are split so that 60 percent are initially designated for metropolitan cities, urban counties and consortia that receive more than the minimum funding and 40 percent for states. HUD allocates a share of HOME funds to jurisdictions and states with a mathematical formula that measures the relative need for affordable housing. Demographic factors, which are derived primarily from the most recent decennial census, represent the relative need.

The Department developed the HOME formula based on criteria established in the HOME legislation. The criteria provide that the formula should be spread geographically to reflect different types of housing need and should not allocate funds excessively to any one community or state. The criteria also identify the following types of affordable housing needs:

• Relative inadequacy of housing supply • Supply of substandard rental housing • Number of low-income families in rental housing units likely to be in need of rehabilitation • Cost of producing housing • Incidence of poverty • Fiscal incapacity to carry out housing activities without Federal assistance

The Department first allocated funds under the formula for FY 1992. The formula factors and funding procedure in subsequent years have been basically unchanged. The appropriation has grown each year, as has the number of PJs.

The formula factors are derived from objective, standardized data primarily from the Census Bureau that measure the criteria as identified in legislation. Some formula factors measure more than one criterion, so the relationship to the criteria and the weights assigned in the formula are as follows:

• Low vacancy and poor renters. This factor is partially indicative of an inadequate housing supply. The value in the formula is formulated by taking the number of rental units occupied by a poor family and multiplying this number by a market tightness measure. The market tightness measure is the ratio of the national vacancy rate for renters divided by the jurisdiction’s vacancy rate for renters. Within the formula this factor has a weight of 0.1.

2 8

• Rental housing with one of four problems. This factor is a measure of the amount of substandard housing as well as inadequate housing supply. The four problematic conditions are over-crowding, incomplete kitchen facilities, incomplete plumbing, and high rent-to-income ratio. Within the formula this factor has a weight of 0.2.

• The number of rental units that were built before 1950 and that are occupied by poor families. This measure is based on the number of low- income families in housing units likely to be in need of rehabilitation. This factor has a weight of 0.2 in the formula.

• RS Means cost index and rental units with one of four problems. This factor is the number of occupied substandard rental units (identified in factor 2) multiplied by a figure that measures the cost of producing housing. This factor has a weight of 0.2 in the formula.

The cost figure in the formula is the ratio of the RS Means cost index for an individual jurisdiction divided by the Means cost index for the nation as a whole. The source for these indices is the “Means Square Foot Costs Annual Edition.”

• The number of families in poverty. This factor is used to measure poverty and also the relative fiscal incapacity to carry out housing activities.

• Low PCI population. This factor is used to partially measure the relative fiscal incapacity to carry out housing activities. The factor is computed by multiplying the population of a jurisdiction by its net per capita income (PCI) index. Within the formula this factor has a weight of 0.1.

Census Data Collection

In Census 2000, the Census Bureau used enumeration methods that targeted a housing location—either a housing unit or group quarters. The latter housing type encompasses homeless shelters, prisons, and other group settings. The Census Bureau also did enumeration at service-based operations including soup kitchens, mobile food vans, and non-shelter based outdoor locations. Enumeration occurred for one night.

The Census Bureau has changed its data tabulations since the 1990 Census. In the 1990 Census, people enumerated at emergency and transitional shelters and at visible street locations were displayed separately. For the data to be released for 2000, the figures will not be shown separately in table form. This change in tabulation methods was based on several factors.

First, the Census Bureau does not produce counts of homeless people, as there are many limitations to the data that is collected. The Census produces a one-time snapshot of “putting people in a location.” Because there are different definitions of homelessness, and different settings and housing environments, not all homeless people will be counted as being homeless based on the specific definition used. To produce an accurate count of homeless people, the Census Bureau or another organization would need to conduct a targeted study of people who are homeless and

2 9

visit the same sites more than once. In addition, other methods would need to be used such as general and sample surveys, administrative records, and other data sets.

Second, the Census Bureau wants to discourage people from taking the results from the service-based enumeration in an attempt to draw conclusions about the services available in a given community. Last, the Census Bureau does not want people to use the data as a measure of service usage.

The Census Bureau will release, starting in June 2001, Summary File 1 which presents counts and basic cross-tabulations of information collected from all people and housing units, such as age, sex, race, origin, household relationship and owner/renter. Included in the “other non-institutional group quarters” category will be the enumeration figures for emergency and transitional shelters, soup kitchens, etc. Summary File 3 will be released starting in June 2002 and includes data collected on a sample basis down to the block level.

Starting in the near future, the Census Bureau will begin to release data obtained through a new, general-purpose survey, the American Community Survey. The survey will provide annual information in several areas. Though the survey will produce fresh data on an on-going basis—not just every ten years—the survey will not produce complete results for a targeted population, such as homelessness.

Information to be provided by the survey includes age, gender, race/ethnicity, family status, country of birth, year of immigration, educational attainment, school enrollment, language, disability, types and income amounts, poverty, unemployment, and data on housing affordability.

The survey, therefore, provides new opportunities to study trends in an area’s population and migration patterns, inform strategic decision-making, assist in program evaluation, and improve population and poverty estimations.

Currently, the American Communities Survey is being tested in 31 demonstration sites. Release dates of information range from late summer/early fall 2001 for states, counties and cities with populations over 250,000 to 2008, to five-year averages for areas with populations less than 20,000.

Many Continuum of Care planning communities use the information that is provided by the Census Bureau as a reference point for understanding homelessness. Census data is then supplemented by other methods and surveys.

Homeless Management Information Systems (HMIS)

A HMIS is computer technology that helps jurisdiction-wide (city, region, state) efforts to carryout system-wide client intake and service processes. It has the capability to generate an unduplicated count of homeless people across the jurisdiction(s), calculate shelter turnover rates, develop client profile demographics and service use, and document project and program outcomes over time. A HMIS can be used as a vehicle to streamline reporting to HUD and other funders and facilitate agency- and program- level evaluation. In addition, it can enable electronic transmission of client records to improve service coordination and reduce duplicative service delivery. HMISs can also expediently connect homeless persons with resources they need.

3 0

To date, 30 to 35 jurisdictions are well on their way to a respectable level of coverage using a HMIS. The FY 2001 HUD Notice of Funding Availability (NOFA) and application collects information on communities’ use of HMIS. This information will help to develop a baseline on the status of HMIS nationally.

Last year, a variety of jurisdictions formed the National Human Services Data Consortium. Twenty jurisdictions are currently involved as follows: States—RI, WI, MA; Cities—Anchorage, AK; Austin, TX; Chattanooga, TN; Chicago, IL; Columbus, OH; Grand Rapids, MI; Jacksonville, FL; Nashville, TN; Oklahoma City, OK; San Diego, CA; San Juan, PR; Washington, DC; Counties or Regions—Cuyahoga County, OH; Lake County, IL; Seattle/King County, WA; Spartanburg County, SC; and Washtenaw County, MI.

The Consortium was created to evaluate current software vendors and software. It created a collective approach to reviewing proposals from vendors and negotiating a contract with the vendor that was ultimately selected. The Consortium’s purchasing power resulted in large discounts, and it put in place a mechanism in the contract for services for a long-term development partnership to provide feedback to improve the product. It is web-based. The Consortium also convenes on a regular basis to provide implementation support.

In Massachusetts, interested parties are working to implement a statewide HMIS. The priority for coverage for the state is to start with the City of Boston, particularly providers of emergency shelter, and to collect a core set of data elements from all of their clients over time. The City of Boston’s priority is to generate an automated APR and to enable the APR reports to be more accurate to help the City paint a clearer picture of what’s happening in the City.

The challenge in using a HMIS is that it cannot facilitate or account for the intake of a large number of homeless people at one time. To gather this valuable data, organizations are working now to develop a sampling process at large shelters to gain information about individual shelter users.

The use of HMIS requires vision and leadership, commitment, clarification of objectives and core data elements, privacy protections, client codes, access to data protocols, system-level governing structures, coverage standards, and other requirements. Communities must plan for staged implementation over a two to four year period to reach a representative level of coverage.

HMISs have many advantages, including the generation of solid evidence for public policy decisions, efficiencies in reporting to HUD and other funders, streamlined case management and client access to resources they need to move out of homelessness, and outcome measurement and improved service provision.

The challenges in using HMIS administrative data for the HUD’s Needs formula includes the need by Continuum of Care communities for time and financial/technical resources to reach representative levels of coverage. In addition, attention must be given to the assumptions and limitations of data, the need for common data elements and common definitions of programs and services.

3 1

The University of Pennsylvania (PA) is working with cities, counties and states that are in the most advanced stage of data collection using HMIS. “Advanced” is defined as having coverage of at least 75 percent of services provided. The University of PA has been working with communities to aggregate data to look at the characteristics of homelessness and homeless services used in jurisdictions. In addition, the University is beginning to facilitate cross-jurisdictional comparisons. The challenge is to develop consistent measures across communities.

Jurisdictions who participated in Prevalence 1999 are Atlanta, GA; Boston, MA; Columbus, OH; Grand Rapids, MI; State of Iowa; Kansas City, MO; Montgomery County, MD; New York City, NY; Philadelphia, PA; State of Rhode Island; St. Louis County, MO; St. Paul, MN; San Diego, CA; Spokane, WA; and Washington, DC.

The data collected is: emergency shelter and transitional housing use; annual use for families and homeless individuals (or unaccompanied adults); report on annual prevalence (the number of unduplicated persons using either shelter or transitional housing over the course of a year); average nightly census (point-in-time average which gives more stable count); average length of stay (the number of days or the number of bed nights used per household or person over the course of year); and average annual rate of turnover (the number of homeless persons that would occupy a shelter bed over a course of a year).

Administrative data can provide the number of homeless people in a given jurisdiction over the course of a year. Also, HMIS data is dynamic. That is, taking a count of the homeless population over a year—not just in a point in time—results in data that better adjusts for, and factors in, seasonal fluctuations in service use. Also, HMIS data accounts for turnover. Point-in-time counts tend to represent chronic service users disproportionately. Annual HMIS data, however, can incorporate turnover and capture data on more of the homeless populations. It can show that homelessness affects more people over time than is apparent at any given point-in-time. Also, if address information is available through administrative records–such as the last known address–a community can begin to develop a greater understanding of where homeless people live within different neighborhoods and develop models based on characteristics of homelessness.

The challenge in using administrative data is the significant amount of resources it takes—both human and financial—to begin and sustain a process. It requires a high level of coordination and buy-in. Maintaining consistent data standards and coverage of services to facilitate cross-jurisdictional comparisons can also be challenging, as well as accounting for differences in definitions used for defining homelessness, and internal organizational policies.

It is important to expand data collection to include soup kitchens and outreach facilities and other nontraditional service settings to account for homelessness in more rural areas. HMIS is heavily dependent on the richness of the service system in place. That is, services must be in place in order to collect data. This may be problematic because not all communities are service-rich, which will directly affect their ability to collect data.

Rural homelessness can be accounted for by expanding the definition of services to include welfare offices, community action agencies, the mental health system, and other service vehicles, supplemented by counts of hard to reach homeless people.

3 2

There is a need for additional information on the policies and screening procedures used shelter by shelter. For example, shelters that are more accessible may appear to have a larger homeless population.

Other Measures of Homelessness

Martha Burt, Urban Institute, reviewed her research and work completed in support of her book, Over the Edge: the Growth of Homelessness in the 1980s.1 As part of her research, she analyzed the factors that predicted homelessness rates in large U.S. cities. That is, 182 cities with populations greater than 100,000. She used regression analysis to test the effect of 18 indicators on city differences in 1989 rates of shelter bed capacity among 147 primary cities. The four variables explaining the most variance in 1989 rates of shelter capacity are: (a) the 1980 percent of one-person households; (b) 1988 per capita state spending for alcohol and other drug abuse services; (c) the proportion of county employment in services in 1987; and (d) the 1980-1987 change in wholesale trade employment as a percentage of 1980 wholesale employment.

Data was provided for the research from a variety of sources including HUD. The Census Bureau’s introduction of the annual American Communities Survey will provide more timely data to use in any future research. Of the four variables, the single best predictor is the proportion or percentage of one-person households in a community. The challenge for HUD when trying to improve upon its existing formula for addressing need is to find a factor that is universally accessible and available and that can be updated regularly. Federal Emergency Management Agency (FEMA) uses unemployment rates because data is available monthly for every county in the country.

Another factor that is part of predicting homelessness is population change and dynamics in high growth and lower growth areas. Though the 1991 research focused just on urban areas, it is important to see if there are patterns in less urbanized—more rural—areas. For example, in slower growth areas, unemployment and loss of manufacturing jobs and the level of homelessness was higher.

Therefore, when examining ways to improve upon the existing formula for determining need for homeless assistance, HUD may wish to consider a way to give communities “credit” or “discredit” for a combination of factors, and play with how they are weighted.

The fundamental issue or long-term goal is to state with great confidence what a sound or good measure of the dependent measure of homelessness is, or could be. The other important issue is clarifying what one is trying to predict.

While the CDBG formula has been studied and researched extensively, there has been comparatively little work of a similar depth in the field of homelessness. One way of beginning work in this area is to rely more heavily—if feasible and if the data is sound— on the data submitted as part of the SuperNOFA (Notice of Funding Availability) competition. That is, capture each community’s service capacity (number of beds) from its application to HUD. HUD could also rely on “annuals” (e.g., Annual Progress Reports) that may include similar data sets.

1 New York and Washington, DC: Russell Sage Foundation and the Urban Institute Press, 1991.

3 3

Appendix D:

Descriptions of the CDBG, ESG, and Home Formulas

3 4

CDBG, ESG and HOME Formula

Introduction

Each year HUD’s Office of Community Planning and Development (CPD) allocates billions of dollars to state and local governments under four formula programs. The funds are a vital source of revenue for many local development projects and local services assisting low and moderate- income households. The annual allocations are provided on the web.

The formulas are computed based on data from the Census Bureau and on definitions of metropolitan areas from OMB. New data becomes available from Census 2000 and new metropolitan definitions are being made by OMB.

Community Development Block Grant - Formula Funding

General

The Community Development Block Grant program provides annual allocations to eligible cities and counties and to states for areas that are not entitled to receive funds directly. As specified in sections 102 and 106 of the Housing and Community Development Act of 1974, the program allocates funds based on demographic data provided by the Census Bureau.

After setting aside funds for special purposes such as technical assistance, the annual appropriation for CDBG formula funding is split so that 70 percent is allocated among eligible metropolitan cities and counties and 30 percent among the states. The communities and states must submit annual plans that show how they expect to use these funds and other CPD formula funds and report on their prior year accomplishments.

The following charts show the overall funding from FY 1996 to FY 2001 and the number of communities eligible to receive funds.

National Allocation (dollars in thousands)

FY 1996 FY 1997 FY 1998 FY 1999 FY 2000 FY 2001 Total CDBG $4,370,000 $4,310,400 $4,195,200 $4,225,700 $4,236,050 $4,339,300 Entitlement $3,059,000 $3,017,280 $2,936,640 $2,957,990 $2,965,235 $3,079,510 Metro cities $2,458,766 $2,432,680 $2,361,375 $2,376,349 $2.385,538 $2,463,897 Urban Counties $600,234 $584,600 $575,265 $581,641 $579,697 $615,613 States $1,311,000 $1,293,120 $1,258,560 $1,267,710 $1,270,815 $1,319,790

Number of eligible communities

FY 1996 FY 1997 FY 1998 FY 1999 FY 2000 FY 2001 Metro Cities 815 834 841 842 859 860 Urban Counties 139 141 145 147 149 153 Total 954 975 986 998 1,008 1,013

3 5

Eligible communities and States

Eligible communities must meet criteria established in section 102 of the Housing and Community Development Act. The statute makes the following areas eligible:

Metropolitan Cities • Central Cities of Metropolitan Areas (MAs), • Other cities with a current population of 50,000 or more that are also in MAs • Cities that retain metropolitan city status as a result of previously meeting the criteria for metropolitan cities

Urban Counties • Counties which are in MAs and which have a population of 200,000 or more after excluding metropolitan cities, small cities that do not participate with the county and eligible Indian tribes • Counties that retain qualification status as a result of previously meeting criteria for urban counties

States

The non-entitled portion of state receives funding based on the balance of demographic need characteristics that remain after subtracting data for metropolitan cities and urban counties. Data for eligible Indian tribes are also subtracted since they are eligible for funding under separate grant programs.

Note: The Office of Management and Budget defines Metropolitan Areas and designates central cities. They establish the criteria and update the metropolitan area list when decennial census data are issued and as the Census Bureau updates population estimates throughout the decade.

Qualification process

HUD designates metropolitan cities based on population estimates available from the Census Bureau and central cities designated from OMB. HUD uses the data that are available for all units of government ninety days prior to the start of the federal fiscal year.

HUD also identifies urban counties annually once the data show that a county could potentially be over 200,000 or meet other special legislative tests. The county includes local units of government where the county has authority to undertake community development activities. Urban counties go through a process of establishing legal agreements for participation by local governments when they are first qualified and every three years thereafter.

States are automatically entitled. They are funded based on the non-entitled portion in the state; that is, the balance of the state after excluding metropolitan cities, urban counties with their included units of government and all eligible Indian tribes. Only small cities, small towns and rural counties in the non-entitled area may apply for funding to the state. The Housing and Community Development Act defines the District of Columbia as a metropolitan city. It includes Puerto Rico as a state. Other territories, outlying areas and Indian tribes are excluded from the formula and funded under set-asides from the annual appropriation.

3 6

CDBG Formulas

Although entitlement communities and states now receive funds under a dual formula system, the CDBG program first allocated funds in FY 1975 under a single formula. The block grant phased in this formula for eligible communities to replace funding for a group of special purpose, “categorical” competitive programs. For FY 1978 the program added a second formula that was intended to assist the older declining communities that would have had funding phased down under the initial formula. Communities and states receive funds under whichever formula provides the higher amount. Since FY 1980 when the dual formula was fully phased in, the allocation formula has been fundamentally unchanged.

The demographic factors for initial formula (“formula A”) are population, poverty and overcrowded housing, weighted .25, .5 and .25 respectively. The factors for the second formula (“formula B”) are population growth lag from 1960, poverty and pre-1940 housing, weighted .2, .3 and .5 respectively. For states, population replaces population growth lag as a factor in the second formula.

To ensure objectivity and consistency, the decennial census is the primary source of the data in the CDBG formula. In years following release of the decennial data, the Census Bureau provides updated population estimates and identifies new incorporations, major boundary changes reported to the Bureau for use in the formula. As required by statute, HUD uses the latest consistent data available for all areas as of ninety days prior to the start of the fiscal year. Since HUD allocates funds to Indian tribes separately, HUD excludes data for Indian tribes from the formula data for all states and entitlement communities.

Formula Data

The data used to compute the formula amounts are proxies for a various indicators of community need. In 1980 and in 1990 HUD published studies of how well the decennial data targets funds to needy communities. In both studies HUD determined that the data continued to target the funds to the neediest communities and recommended continuing the dual formula as specified in statute.

In general, the formula A factors of population, poverty and over crowded housing help target funds to communities that are experiencing the strains of population growth. Formula B factors of population growth lag, poverty and pre 1940 housing targets funds to the older and declining communities.

The statute specifies the definitions for the formula demographic factors. The definitions are:

Population – total resident population. For FY 1999 and 2000, HUD used 1998 population estimates from the Census Bureau. For FY 2001, HUD used 1999 population estimates from the Census Bureau.

Poverty – persons below the poverty level. Since FY 1993, the Department used 1990 decennial census data.

Overcrowded housing – housing units with more than 1.01 persons per room. Since FY 1994, the Department used 1990 decennial data.

Age of housing – housing units built before 1940. Since FY 1994, the Department used 1990 decennial data.

Population growth lag – The shortfall in population that a city or county has experienced when comparing its current population to the population it would have had if it grew like

3 7

all metropolitan cities since 1960. Note the growth rate for FY 2001 was 32.5 percent. It is important to note that, while the latest population used to compute growth lag reflects recent boundary changes, HUD cannot make changes to the 1960 population for individual communities based on boundary changes that result from annexations since the 1960 data for are not available. HUD does make changes to the 1960 population data for communities that result from mergers, since the data are available.

Formula Amounts and Final Allocation

There are two general steps in computing allocations. First, find the two formula amounts for each community or state based on its “share” of an overall demographic need amounts. And then, second, find the final allocation by reducing the higher formula amount for each community or state on a pro rata basis so as not to exceed the total available for communities and states.

Formula “shares”

Section 106 of the Housing and Community Development Act requires that HUD allocations use a “share” method that computes formula amounts based on the ratio of an entitlement community’s need to an overall need value. (The “share” method is not a true share method for entitlement communities, because, as explained later, the sum of all the shares is less than 100 percent.) The “shares” for each demographic need factor are multiplied by weights to reflect relative importance and then summed for formula A and formula B. The sums are multiplied times the amount available for distribution to give the two formula amounts.

The two formulas for a metropolitan city are shown as follows:

Formula A = funding amount times the sum of

0.25 times population of the city divided by the population for all metropolitan areas,

0.5 times poverty of the city divided by the poverty of all metropolitan areas

0.25 times overcrowded housing of the city divided by the overcrowded housing of all metropolitan areas

Formula B = the funding amount times the sum of

0.2 times the growth lag of the city divided by the sum of growth lag for all metropolitan cities

0.3 times the poverty of the city divided by the sum of the poverty of all metropolitan areas

0.5 times the pre-1940 housing of the city divided by the pre-1940 housing of all metropolitan areas

Here are a few important points that are frequently misunderstood about the formula amounts:

Neither HUD nor a community can predict whether its formula amount will increase or decrease based on an increase or decrease in its demographic factors. This is because the formula computes amounts based on shares of a total need value rather than the demographic need for any given community. For example, even though a community’s population increases by 5 percent, it will not receive more funds on this factor if the denominator increases by 10 percent.

3 8

A higher population estimate for a community may result in lower funding. Communities that benefit from formula B use the population growth lag factor. If a formula B community has a population increase that is greater in percentage terms than metropolitan cities in general, then its population growth lag factor will decrease and funding will be less.

The formula appears to operate based on “shares,” but in fact the metropolitan area and growth lag denominators used to compute the shares for entitlement communities do not equal the sum of the numerators. While the initial formula did have denominators that equaled the sum of the numerators, this feature was lost as the formula was changed over time. To preserve the allocation weight of the initial formula established by Congress, the denominators for population, poverty, overcrowded housing and pre 1940 housing continue to be the sum of demographic values for metropolitan areas (MAs). For growth lag, the denominator for metropolitan cities is the sum of growth lag values for metropolitan cities, and for urban counties it is the sum of growth lag for metropolitan cities and urban counties. For states, however, the shares are computed based on the sum of data for all states.

Final Allocation

While communities and states receive funding based on the greater of the two formula amounts, there is a final adjustment to the greater formula amount before finding the final allocation. Since the sum of the higher amounts typically exceed the total available for funding, the higher amounts are pro-rata reduced.

Thus the final allocation for any community or state is computed as follows:

Final allocation = pro-rata reduction * (the greater of formula A or formula B)

Where pro-rata reduction = the percent necessary to reduce the sum of the higher formula amounts to equal the amount available for allocation

There are a couple of points worth noting about the influence of the pro-rata reduction on the final allocation.

Since the final pro-rata reduction is unknown until all the formula amounts are calculated, it is impossible to predict any individual allocation until all the allocations are computed.

Pro-rata reduction changes every year based on the changes in formula amounts for existing communities and states, and as a result of new entitlement communities moving from state funding to the entitlement pot. The identification of new entitlement communities reduces the pro-rata reduction for states and increases the pro-rata reduction for entitlement communities.

3 9

Examples of Formula Allocations

The following examples illustrate the computations involved in determining the formula “shares” and the final CDBG formula allocations to a metropolitan city and an urban county. In the examples below, fast growing Urban County A has substantial overcrowded housing, but relatively few houses built before 1940. Old town B is a declining town that saw its boom before 1940.

The demographic factors for the Urban County A and Old Town B are the numerators used to compute formula shares. To help explain the calculation of the allocations, all of the demographic values are reduced to unrealistically small numbers.

Factor Urban County A Old town B Formula A Population 333 222 Poverty 33 22 Overcrowded housing 33 22 Formula B Population Growth lag 0 22 Poverty 33 22 Pre-1940 housing 3 22

The denominators used to compute formula “shares” are demographic factors for all metropolitan areas (MAs) and, in the case of growth lag, for all metropolitan cities and urban counties. In this example they are chosen to help simplify the computation of formula “shares”.

Factor Denominator Population in MAs 100,000 Poverty in MAs 10,000 Overcrowded housing in MAs 1,000 Pre-1940 housing in MAs 1,000 Population Growth lag in Metro Cities 1,000 Population Growth lag in Metro Cities and 1,100 Urban Counties

The “shares” for a given year are computed by dividing the numerators, which are the values for the individual metropolitan cities or urban counties, by the denominators as described above.

Factor Urban County Denominators “Share” A Formula A Population 333 100,000 0.00333 Poverty 33 10,000 0.00330 Overcrowded 33 1,000 0.03300 housing Formula B Population Growth 0 1,100 0.00000 lag Poverty 33 10,000 0.00330 Pre-1940 housing 3 1,000 0.03000

Factor Old town B Denominators “Share”

4 0

Formula A Population 222 100,000 0.00222 Poverty 22 10,000 0.00220 Overcrowded 22 1,000 0.02200 housing Formula B Population Growth 22 1,000 0.02200 lag Poverty 22 10,000 0.00220 Pre-1940 housing 20 1,000 0.02200

The “shares” described in the table above are weighted by values specified in statute and the weighted values are summed. In this example the sum of the weighted “shares” is multiplied by $3 billion, the total available amount to distribute.

Factor Weight “Share” Urban Weighted County A “share” Formula A Population 0.25 0.00333 0.0008325 Poverty 0.5 0.00330 0.0016500 Overcrowded 0.25 0.03300 0.0082500 housing SUM 0.0107325 Formula A share of $32,198 $3 Billion thousand

Formula B Population Growth 0.2 0.0000 0.00000 lag Poverty 0.3 0.0033 0.00099 Pre-1940 housing 0.5 0.0030 0.00150 SUM 0.00249 Formula B share of $7,470 thousand $3 Billion

Factor Weight “Share” Weighted Old town B “share” Formula A Population 0.25 0.00222 0.000555 Poverty 0.5 0.00220 0.001100 Overcrowded 0.25 0.02200 0.005500 housing SUM 0.007155 Formula A share of $21,465 $3Billion thousand

Formula B Population Growth 0.2 0.0220 0.00440 lag Poverty 0.3 0.0022 0.00066

4 1

Pre-1940 housing 0.5 0.0220 0.01100 SUM 0.01606 Formula B share of $48,180 $3 Billion thousand

The higher of the two formula amounts for Urban County A and Old town B and the higher amounts for all other entitlement communities are totaled. That total is typically greater than the funds available to distribute. If in this example the total comes to $3.3 Billion, then the higher amounts for all communities must be pro rata reduced by 10 percent so that the overall amount does not exceed $3 Billion. Thus the higher formula amounts for the Urban County A and Old town B are pro rata reduced to compute the final amounts.

Formula amount Prorata reduction Final Allocation (thousands) (thousands) Urban County A $ 32,198 Minus 10% $28,978 formula A Urban County A 7,470 NA NA formula B Old town B formula A 21,465 NA NA Old town B formula B 48,180 Minus 10% 43,362

4 2

The HOME Formula

The "HOME Investment Partnership Act" (HOME), authorized under Title II of the National Affordable Housing Act of 1990, allocates formula funds to participating jurisdictions to increase the number of families served with decent, safe, sanitary, and affordable housing and to expand the long-term supply of affordable housing.

Participating jurisdictions (PJs) that receive a share of formula funds must submit annual plans on how they expect to use the funds. PJs make their plans available for public review and comment. HUD publishes program regulations that identify eligible activities to make rental housing and home-ownership more affordable for low income and very low-income families.

HOME funds, which are made available for allocation, are split so that 60 percent are initially designated for metropolitan cities, urban counties and consortia that receive more than the minimum funding and 40 percent for states. HUD allocates a share of HOME funds to jurisdictions and states with a mathematical formula that measures the relative need for affordable housing. Demographic factors, which are derived primarily from the most recent decennial census, represent the relative need.

The Department developed the HOME formula based on criteria established in the HOME legislation. The criteria provide that the formula should be spread geographically to reflect different types of housing need and should not allocate funds excessively to any one community or state. The criteria also identify the following types of affordable housing needs.

• Relative inadequacy of housing supply • Supply of substandard rental housing • Number of low-income families in rental housing units likely to be in need of rehabilitation • Cost of producing housing • Incidence of poverty • Fiscal incapacity to carry out housing activities without Federal assistance

The Department first allocated funds under the formula for FY 1992. The formula factors and funding procedure in subsequent years have been basically unchanged. The appropriation has grown each year, as has the number of PJs.

The following chart shows the funding and eligibility for FY 1998 through FY 2001. In general entitlement PJs receive funds under the 60 percent pot and state PJs receive funds under the 40 percent pot. However, as shown in the table, states with no participating jurisdictions also receive $500,000 in funds from the 60 percent pot. Also funds from entitlement PJs that do not apply for an allocation are reallocated to their state, thereby also shifting funds from the 60 percent pot to the 40 percent pot. For purpose of the HOME formula, the Puerto Rico and the District of Columbia are included with the states.

HOME Funding

FY 1998 FY 1999 FY 2000 FY 2001 Total Formula Funds $1,438,000 $1,550,300 $1,552,800 $1,733,577 (thousands) 60 percent pot $862,800 $930,180 $931,680 $1,040,146 Metro cities $621,390 $666,811 $666,555 $734,827 Urban counties $107,203 $115,475 $116,485 $129,078 Consortia $132,207 $145,894 $146,640 $174,241

4 3

States (with no PJ) $2,000 $2,000 $2,000 $2,000 40 percent pot $575,200 $620,120 $621,120 $693,431

Units of Government FY 1998 FY 1999 FY 2000 FY 2001 Receiving an Allocation Total local PJs 533 533 537 542 Metro cities 352 350 349 349 Urban counties 84 84 87 86 Consortia 97 99 101 107 States (with no PJ) 4 4 4 4 Total state PJs 52 52 52 52

Eligibility Requirements

States are automatically eligible to receive a HOME allocation. Metropolitan cities and urban counties may qualify for direct funding if they receive a minimum allocation of the $500,000, or $335,000 when Congress appropriates less than $1.5 billion. Metropolitan cities and urban counties are established under the CDBG program. Consortia, which are contiguous units of governments that join together to apply for the HOME program, may also qualify if they receive at least the minimum allocation. Once a PJ does qualify, it retains its status under the HOME program, even if its future allocation falls below the minimum.

Participation in the HOME program automatically includes all states and any other jurisdictions with funding of at least $750,000, or $500,000 when Congress appropriates less than $1.5 billion. Jurisdictions with formula allocations between the minimum $500,000 (or $335,000 with an appropriation less than $1.5 billion) and $750,000 may add matching funds to reach the $750,000 level that is necessary to actually receive the allocation.

Since eligibility for new PJs is based on funding levels rather than population, it is not clear whether some potential PJs and new consortia will qualify until HUD receives its appropriation and computes the allocation. Any unit of government that does not qualify to receive a direct allocation may apply to its state for HOME program funds.

Formula Funding Allocations

HUD allocates funds to PJs with annual appropriations from Congress, after reductions for any special set asides such as funds for technical assistance and Insular Areas. Of this formula amount, 60 percent is allocated initially among metropolitan cities, urban counties, and approved consortia, while the other 40 percent is allocated among states.

HUD computes an initial HOME allocation amount based on each PJ’s share of need. The 60 percent entitlement funding is allocated based on the relative need among all potentially eligible metropolitan cities, urban counties and consortia. HUD allocates the 40 percent to states in two parts. Thirty-two percent is based on data for the balance of the state after excluding entitlement PJs that pass the minimum funding thresholds. The formula allocates the remaining 8 percent of the formula amount based on total state characteristics data.

After computing the initial allocation, there are adjustments to the initial amounts that ensure the wide-spread distribution of funds and that there is sufficient allocation to each individual PJ. In the case of the 60 percent pot designated for metropolitan cities, urban counties, and consortia, the funds from non-qualifying PJs with the lowest initial allocations are redistributed on a pro rata

4 4

basis to other PJs so as to maximize the number of qualifying PJs. In the case of the state, the initial allocations are adjusted so that no state receives less than $3 million. Also, states without qualifying PJs receive an additional $500,000 from the 60 percent pot. In the case of Puerto Rico and its metropolitan cities, the allocations are capped at no more than twice the per capita funding, and any excess funds are redistributed to other PJs competing in the same pot.

Formula factors

The formula has six measurable factors, each having a weight assigned to it. These formula factors were developed to meet the legislative criteria. The six legislative criteria are: 1) inadequate housing supply; 2) substandard housing; 3) low income families in housing units likely to be in need of rehabilitation; 4) costs of producing housing; 5) poverty; and 6) relative fiscal incapacity to carry out housing activities.

The formula factors are derived from objective, standardized data primarily from the Census Bureau that measure the criteria as identified in legislation. Some formula factors measure more than one criterion, so the weights reflect this overlapping. A detailed explanation of each formula factor including the relationship to the criteria and the weights assigned in the formula are as follows:

(1) Low vacancy and poor renters (LOW-VAC-POOR-RENTR) This factor is partially indicative of an inadequate housing supply. The value in the formula is formulated by taking the number of rental units occupied by a poor family and multiplying this number by a market tightness measure. The market tightness measure is the ratio of the national vacancy rate for renters divided by the jurisdiction's vacancy rate for renters. The national vacancy rate is 8.58 percent. Since FY 1994, the data source has been the 1990 census; the previous source was the 1980 census. Within the formula this factor has a weight of 0.1.

LOW-VAC-POOR-RENTR = 8.58 / (100 * VACANT / (VACANT + RENTOCC)) * TRHPOV

Where VACRENT = the number of vacant rental units RENTOCC = the number of occupied rental units TRHPOV = the number of rental units occupied by the poor.

(2) Rental housing with one of four problems (SUBSTANDARD - RENTALS) This factor is a measure of the amount of substandard housing as well as inadequate housing supply. The four problem conditions are overcrowding, incomplete kitchen facilities, incomplete plumbing and high rent to income ratio. Since FY 1994, the data source has been the 1990 census; the previous source was the 1980 census. Within the formula this factor has a weight of 0.2.

(3) The number of rental units that were built before 1950 and that are occupied by poor families (PRE50-RENTAL-OCC-POOR). This measure is based on the number of low-income families in housing units likely to be in need of rehabilitation. This factor has a weight of 0.2 in the formula. Since FY 1994, the data source has been the 1990 census; the previous source was the 1980 census.

(4) RS Means cost index and rental units with one of four problems (COST-SUBSTNDRD-RENTALS) This factor is the number of occupied substandard rental units (identified in factor 2) multiplied by a figure that measures the cost of producing housing. This factor has a weight of 0.2 in the formula.

The cost figure in the formula is the ratio of the RS Means cost index for an individual jurisdiction divided by the Means cost index for the nation as a whole. The source for these indices is the "Means Square Foot Costs Annual Edition."

4 5

(5) The number of families in poverty (FAM-POV) This factor is used to measure poverty and also the relative fiscal incapacity to carry out housing activities. Since FY 1994, the data source has been the 1990 census; the previous source was the 1980 census.

(6) Low PCI and population (POP-PCI) This factor is used to partially measure the relative fiscal incapacity to carry out housing activities. The factor is computed by multiplying the population of a jurisdiction by its net per capita income (PCI) index. This index is ratio of two factors: (1) the difference between national PCI of a 3 person family ($14,277) and the national PCI of a 3 person family below the poverty line ($3,295), and (2) the difference between the jurisdiction's PCI of a 3 person family and the $3,295 3-person poverty figure. Since FY 1994, the data source has been the 1990 census; the previous source was the 1980 census. Within the formula this factor has a weight of 0.1. POP-PCI = POP * (14,277 – 3,295) / (PCI – 3,295)

HOME Formula process – Initial allocation and adjustments

The same formula is used to compute allocations for all three funding pots-entitlement communities (60 percent), states based on non-entitlement communities (32 percent) and states based on total data (8 percent). There are two steps in the allocation process. First, compute an initial formula share based on the funds available for the pot, the assigned weights, the factors for the PJ, and the total factors for all PJs competing for the pot of funds. Second, make adjustments necessary to meet the legislative requirements that the funds be fairly distributed and sufficiently large for each community to carry out the HOME activities.

The formula for the initial allocation is:

ALLOCATION$ _ FUNDINGLEVEL * ((0.1 * LOW-VAC-POOR-RENTR/TOT-LOW-VAC-POOR-RENTR) + (0.2 * SUBSTNDRD-RENTALS/TOT-SUBSTUDRD-RENTALS) + (0.2 * PRE50-RENTAL-OCC-POOR/TOT-PRE50-RENTAL-OCC-POOR) + (0.2 * COST-SUBSTNDRD-RENTALS/TOT-COST-SUBSTNDRD-RENTALS) + (0.2 * FAM-POV/TOT-FAM-POV) + (0.1 * POP-PCI/TOT-POP-PCI))

Where

FUNDINGLEVEL = the funding level (either the 60 percent, 32 percent or 8 percent pot).

LOW-VAC-POOR-RENTR = factor 1. TOT-LOW-VAC-POOR-RENTR = the national total of factor 1 for all appropriate areas in the particular funding level.

SUBSTNDRD-RENTALS = factor 2. TOT-SUBSTNDRD-RENTALS = the national total of factor 2 for all appropriate areas in the particular funding level.

PRE50-RENTAL-OCC-POOR = factor 3. TOT-PRE50-RENTAL-OCC-POOR = the national total of factor 3 for all appropriate areas in the particular funding level.

COST-SUBSTNDRD-RENTALS = factor 4. TOT-COST-SUBSTNDRD-RENTALS = the national total of factor 4 for all appropriate areas in the particular funding level.

FAM-POV = factor 5. TOT-FAM-POV = the national total of factor 5 for all appropriate areas in the particular funding level.

4 6

POP-PCI = factor 6. TOT-POP-PCI = the national total of factor 6 for all appropriate areas in the particular funding level.

After entitlement PJs receive initial allocations under the 60 percent pot, there are several adjustments to ensure that the funds are spread to all communities and so that the funds are sufficiently large to carry out the activities. Initial allocations to Puerto Rican PJs are very high on a per capita basis due the extent of poverty and the weight given to poverty in the formula. The formula process limits funding for any such community so that it does not exceed twice the per capita funding for entitlement PJs.

Second, some potential PJs receive less than the minimum in the initial allocation. The lowest initial allocations are redistributed on a pro rata basis to all other PJs in such a way so as to add to the number of PJs above the minimum. To increase the fundable PJs, the funds from PJs at the lowest half of the range below minimum are successively added to other PJs. When at least 99 percent of the funds are distributed through this successive roll up, the remaining funds from any below minimum PJ is redistributed to other PJs.

There is also an adjustment to provide additional funding of $500,000 to states that have no PJs with minimum funding. In order to provide for this minimum, all potential PJs’ allocations are reduced on a pro-rata basis. If these reductions cause any potential PJ's funding amount to drop below the threshold level, then another pro-rata reduction is performed on all PJs above the threshold to collect an amount needed to bring all previously qualified potential PJs above the threshold.

The last adjustment for entitlement PJs corrects for rounding to the nearest $1,000. After rounding, if the total for all PJs is different from the amount appropriated for the 60 percent funding level, then the PJs with the highest allocation amounts are either reduced or increased by $1,000 until the total allocation equals the appropriation amount.

States receive funds out of a 40 percent funding pot. This overall state pot is split between two smaller funding amounts: a 32 percent pot and an 8 percent pot. The same formula is used for these two pots; however, there is a different basis for computing the state formula factor for each pot. The 32 percent pot is allocated based on balance of state factors. To compute these balance data, the demographics for all potential participating jurisdictions that received funding from the 60 percent pot are subtracted from the total state data. The resulting demographics are used in the HOME formula to compute the funding allocation each state will receive under this 32 percent funding pot. The 8 percent pot is based on total state data.

Within this 40 percent funding level, there are a few adjustments that take place after the initial allocations. First the Puerto Rico state amount may not exceed twice the average funding for all states on a per rental unit basis. Second, any state PJs may receive no less than $3,000,000. The 32 percent and 8 percent allocations are added together to determine the total 40 percent state allocation. If a state falls short of $3,000,000, then a pro-rata reduction is applied to the other states so that each state receives at least $3,000,000.

Last, as with entitlement PJs, there is a final rounding adjustment for states. After rounding state allocations to the nearest $1,000, if the total for all state PJs differs from the amount appropriated for the 40 percent funding level, then the state PJs with the highest allocation amounts are either reduced or increased by $1,000 until the total allocation equals the appropriation amount.

4 7

Examples

Entitlement funding under the 60 percent pot

An example of a consortium helps to illustrate the initial allocation and adjustment process for the 60 percent pots of funds. Suppose that there is a large consortium that combined an urban county, a metropolitan city and several small cities with legally binding agreements. All the jurisdictions are contiguous. The jurisdictions decided to join together, because they expected it would give them sufficient demographic characteristics to receive a minimum allocation under the formula. While it was impossible to say how they might be funded in a given year, they understood from HUD that they would have been funded in the previous fiscal year.

Compute an initial allocation for the consortium.

The initial allocation is based on the 60 percent of the funds available times the weighted sum of the shares the consortium has for the six factors. The allocation formula is:

ALLOCATION$ _ FUNDINGLEVEL * ((0.1 * LOW-VAC-POOR-RENTR/TOT-LOW-VAC-POOR-RENTR) + (0.2 * SUBSTNDRD-RENTALS/TOT-SUBSTUDRD-RENTALS) + (0.2 * PRE50-RENTAL-OCC-POOR/TOT-PRE50-RENTAL-OCC-POOR) + (0.2 * COST-SUBSTNDRD-RENTALS/TOT-COST-SUBSTNDRD-RENTALS) + (0.2 * FAM-POV/TOT-FAM-POV) + (0.1 * POP-PCI/TOT-POP-PCI))

Where

FUNDINGLEVEL = the funding level for the 60 percent pot.

LOW-VAC-POOR-RENTR = factor 1. TOT-LOW-VAC-POOR-RENTR = the national total of factor 1 for all appropriate areas in the particular funding level.

SUBSTNDRD-RENTALS = factor 2. TOT-SUBSTNDRD-RENTALS = the national total of factor 2 for all appropriate areas in the particular funding level.

PRE50-RENTAL-OCC-POOR = factor 3. TOT-PRE50-RENTAL-OCC-POOR = the national total of factor 3 for all appropriate areas in the particular funding level.

COST-SUBSTNDRD-RENTALS = factor 4. TOT-COST-SUBSTNDRD-RENTALS = the national total of factor 4 for all appropriate areas in the particular funding level.

FAM-POV = factor 5. TOT-FAM-POV = the national total of factor 5 for all appropriate areas in the particular funding level.

POP-PCI = factor 6. TOT-POP-PCI = the national total of factor 6 for all appropriate areas in the particular funding level.

4 8

For this example, assume that the appropriation provides $1 billion for formula funding (with the funding level for the 60 percent pot at $600,000,000) and the shares are simple fractions 1/10 through 6/10 for each of the six factors. Then the overall formula for the initial allocation with funding level, shares and weights is:

ALLOCATION$= $600,000,000 * ((0.1 * 1/10) +(0.2 * 2/10) +(0.2* 3/10)+ (0.2 * 4/10) + (0.2 * 5/10) + (0.1* 6/10))

By simplifying the fractions,

ALLOCATION=

$600,000,000 * (0.01+0.04+0.06+0.08+0.10+0.06) = $600,000,000 * 0.35 = $210,000,000

The consortium has more than the minimum needed to receive funding. Since this year the funding level is less than $1.5 billion, the minimum is $335,000. The consortium qualifies as a PJ since it has more than this minimum.

Adjust the initial allocation.

There are several adjustments to the initial allocation. The first adjustment is to limit the funding to Puerto Rican PJs with high per capita funding. If we assume that there is a Puerto Rican PJ with 200,000 rental-housing units and with an initial funding of $20,000,000, then it has average funding per rental unit of $100. But if the average among potential entitlement PJs were $90.00, then the funding for this PJ would be limited to $18,000,000. The additional $2,000,000 in funding would be redistributed to the consortium in the example and all other entitlement PJs on a pro rata basis.

The second adjustment is a pro rata adjustment to redistribute funds from communities with low initial allocations. If we assume, there are 101 PJs that had never been funded and that 100 had initial allocations of $100,000 and the one another with $334,000, then there would be a rollup of the $1,000,000 to all other communities including the one with the initial allocation of $334,000. This would be enough to increase funding to the PJ with $334,000 over the $335,000 minimum and to add a fraction of a percent funding to the consortium in our example above with an initial funding amount of $210,000,000 and to all other PJs with above minimum funding.

The third adjustment is to add $500,000 to all states that have no PJs with minimum funding. Assume that there is one state without any PJ. Then $500,000 would be set aside for that state out of the 60 percent pot of funds and all PJs would need to be reduced on a pro rata basis reduced by a fraction of a percent of funding to make up for this amount.

The last adjustment is to make sure the amounts add up exactly to the amount available. Since the numbers are rounded to the nearest thousand dollars, it is possible that the amounts might be off by a few thousand when they are totaled. If the total in the example add to $600,001,000 and there is only $600,000,000 available, then the entitlement PJ with the largest allocation is reduced by $1,000. Similarly that PJ would have received a $1,000 increase if the sum had been $599,999,000.

State funding under the 40 percent pot

Similar to the funding procedure for entitlement PJs, HUD allocates funds to states first by developing initial funding amounts and then adjusting those amounts.

4 9

The initial allocation for each state equals the sum of the allocations under the 32 percent pot and the 8 percent pot. The first allocation for the state equals 32 percent of the funds available times the sum of the state’s share of the total state balance amount. The second allocation equals 8 percent of the funds available times the sum of the state’s share of the total state amount. The basic formula for both pots is the same as the formula used for the 60 percent pot.

ALLOCATION$ _ FUNDINGLEVEL * ((0.1 * LOW-VAC-POOR-RENTR/TOT-LOW-VAC-POOR-RENTR) + (0.2 * SUBSTNDRD-RENTALS/TOT-SUBSTUDRD-RENTALS) + (0.2 * PRE50-RENTAL-OCC-POOR/TOT-PRE50-RENTAL-OCC-POOR) + (0.2 * COST-SUBSTNDRD-RENTALS/TOT-COST-SUBSTNDRD-RENTALS) + (0.2 * FAM-POV/TOT-FAM-POV) + (0.1 * POP-PCI/TOT-POP-PCI))

Where

FUNDINGLEVEL = the funding level for 32 percent or 8 percent pot.

LOW-VAC-POOR-RENTR = factor 1. TOT-LOW-VAC-POOR-RENTR = the national total of factor 1 for all appropriate areas in the particular funding level.

SUBSTNDRD-RENTALS = factor 2. TOT-SUBSTNDRD-RENTALS = the national total of factor 2 for all appropriate areas in the particular funding level.

PRE50-RENTAL-OCC-POOR = factor 3. TOT-PRE50-RENTAL-OCC-POOR = the national total of factor 3 for all appropriate areas in the particular funding level.

COST-SUBSTNDRD-RENTALS = factor 4. TOT-COST-SUBSTNDRD-RENTALS = the national total of factor 4 for all appropriate areas in the particular funding level.

FAM-POV = factor 5. TOT-FAM-POV = the national total of factor 5 for all appropriate areas in the particular funding level.

POP-PCI = factor 6. TOT-POP-PCI = the national total of factor 6 for all appropriate areas in the particular funding level.

After computing the formula under the 32 percent and the 8 percent pots of funds, the amounts for the two allocations are summed. Assume there is a state with $7,000,000 allocated under the 32 percent pot and $3,000,000 under the 8 percent pot. It would receive the sum of these two amounts, $10,00,000 as its initial formula amount.

There are several adjustments for states. First, the state funding for Puerto Rico may not exceed twice the average funding for states per rental housing unit. So, if Puerto Rico received a combined initial allocation under the 40 percent funding of $30,000,000 and had 10,000,000 rental housing units, then it has an average funding of $3. But if the average funding for states is $2.50, then the funding for the state of Puerto Rico would be reduced to $25,000,000 and the $5,000,000 remainder would be spread to other states on a pro rata basis.

Second, since no state may receive less than $3,000,000, there is an adjustment to bring all states up to this amount. So, for example, if there is only one state with an amount below

5 0

$3,000,000 and that state has only $2,000,000, then it would receive an additional $1,000,000 and all other states would be reduced on a pro rata basis to provide this additional $1,000,000.

Third, after rounding all allocations to the nearest $1,000, there is a rounding adjustment to the largest states that adds or subtracts $1,000 from each state so as to ensure that the overall total equals the amount available from the 40 percent pot. So, for example, if the sum of the adjusted allocations was over the amount to be allocated by $2,000, then the two largest states would each receive a reduction of $1,000. The two largest states would each receive an increase of $1,000 if the total were less than the amount available under the 40 percent pot by $2,000.

Last, the state amounts are also adjusted so that they receive an additional $500,000 if there are no PJs that received funding under the 60 percent pot. This amount was set aside under an adjustment during the entitlement funding process. Thus, if there is no entitlement PJ in a state that received $4,000,000 under the 40 percent pot, that state would receive an additional $500,000 for a total of $4,500,000.

5 1

Emergency Shelter Grants – Formula allocation

General

The Emergency Shelter Grant program provides formula funds to entitlement communities and states for emergency shelters and essential services for the homeless. ESG is authorized under Subtitle B of title IV of the Stewart B. McKinney Act (42 U.S.C. 11371 et seq.).

The program funds four types of facility and service support activities to help in sheltering the homeless as well as assisting in the transition to permanent housing. The types of allowable expenses for ESG are rehabilitation, operation of facilitates, delivery of essential services and homeless prevention.

Congress funds ESG together with several other programs as Homeless Assistance Grants. These other programs include Supportive Housing; Section 8 Moderate Rehabilitation for Single Room Occupancy (SRO) Dwellings; and Shelter Plus Care. Of these programs, only ESG is a formula program; the other programs distribute funds based on competitions.

The ESG program is one of four of CPD’s formula programs. ESG funds can be used together with related CDBG, HOME and HOPWA formula programs to fund aspects of homeless activities that are eligible for these programs. The formula funding among states and among entitlement communities is based on the funding the community receives in CDBG. This paper describes funding in more detail and provides an example to illustrate formula funding.

Formula funding

The following table shows the funds that HUD allocated per year to all eligible recipients and the counts of the recipients by type. As shown by the table, appropriated funds for ESG have held constant at about $150 million from FY 1998 to FY 2001. Entitlement communities, which include eligible CDBG metropolitan cities and urban counties, have varied slightly from year to year with small changes to the initial allocation to communities that are close to the funding threshold. The fifty states, plus Puerto Rico and the territories have remained constant, except for the exclusion of Palau starting in FY 1999.

Fiscal year FY 1996 FY 1997 FY 1998 FY 1999 FY 2000 FY 2001 Allocated $113,736 $165,000 $150,000 $150,000 $150,000 $149,670 (thousands) Entitlement 311 313 314 313 311 312 areas States and 56 56 56 55 55 55 territories

ESG formula allocation

Congress selected the CDBG formula as the basis for allocating ESG funds. The CDBG formula is an established funding formula that targets funds to large communities, particularly older and declining communities, with significant needs for homeless assistance.

The statute requires that initial allocations for ESG be proportionate to funding among states and entitlement communities under the CDBG formula for the prior year. The eligibility for ESG funding is limited to those CDBG entitlement communities that receive more than 0.05 percent of the amounts appropriated. ESG legislation provides a single exception to this minimum threshold

5 2

for Woonsocket, Rhode Island. Funds from CDBG communities that are below threshold are redistributed to their states.

After setting aside 0.2 percent of the appropriation to territories and insular areas, the ESG formula distributes the initial funds so that 70 percent goes to entitlement communities and 30% to states. However, as the initial allocations for CDBG entitlement communities with below threshold funding are transferred to the state pot, actual split is shifted to less than 70 percent for entitlement communities. For FY 2001, the actual percent allocated to entitlement communities is 54 percent.

Example

Compute the ESG allocations for Big Town, Smallville, and their state. If $150,000,000 is appropriated for ESG, with set asides of $300,000,000, then the balance of $149,700,000 is initially allocated among all metropolitan cities, urban counties and states in proportion to the actual allocation for CDBG for the prior fiscal year. Based on this the initial allocation, assume Big Town receives $1 million, Smallville receives $40,000, and the state receives $45,000.

The initial allocations must be adjusted to limit allocations to entitlement communities so they all are above the minimum. The minimum allocation for entitlement communities is 0.05 percent of $150,000,000 or $75,000. Big Town with an initial allocation of $1 million is well above the $60,000 minimum and so it receives the entire amount that it has been allocated. There is no minimum allocation for states. Since Smallville has less than the $75,000 minimum, its allocation amount of $40,000 is added to the $45,000 of its state and the state receives a total of $85,000.

The example illustrates that the effect of applying the minimum to the initial allocations is to shift funds to the state.

Big Town Smallville State Initial allocation $1,000,000 $40,000 $45,000 Minimum $75,000 $75,000 NA Final after minimum $1,000,000 0 $85,000

5 3

Appendix E:

Four Alternative Allocations Under Current Formulas

5 4

Comparison of Alternative Pro Rata Need Allocation Methods

Introduction

The following table presents the pro rata need amounts under four different methods compared to the current Continuum of Care (CoC) method. This table was presented at the “Determining Need For Assistance in HUD’s McKinney–Vento Homeless Competitive Programs Conference” held April 19, 2001, at the Department of Housing and Urban Development in Washington DC. These allocation methods are applied to the CDBG universe of jurisdictions, and they do not represent what any particular existing Continuum of Care geography would actually get under each allocation method.

Alternative Allocation Methods Compared

Continuum of Care (CoC) Method: This is a hybrid of the regular ESG formula. It targets 75 percent of the McKinney-Vento competitive funds to the approximately 325 large metropolitan cities and urban counties eligible to receive ESG entitlement grants. The remainder (25 percent) is distributed to other metropolitan cities, urban counties and states. The distribution is done using the Community Development Block Grant (CDBG) formula.

Community Development Block Grant (CDBG) Method: This method allocates 70 percent of the funds to over 900 entitlement metropolitan cities and urban counties and 30 percent to states.

ESG Method: This method is based on the CDBG recipient base, formula factors, and initial 70/30 percent entitlement/state balance proportions. The key difference is that the funds are allocated based upon the prior years’ CDBG allocation to over 900 metropolitan cities, urban counties and states. Thus if a CDBG entity got 2 percent of the total CDBG allocation in 2000 it would get 2 percent of the total ESG allocation in 2001. The ESG program also has a minimum grant threshold for receiving a metropolitan city or urban county entitlement grant. If a metropolitan city/urban county receives less than the threshold (0.05 percent of total ESG allocation) that jurisdiction’s funds are transferred to the state’s share.

HOME Method: This method allocates 60 percent of the funds to CDBG metropolitan cities, urban counties and consortiums. To make this alternative comparable to the others, consortiums were removed from the geography base for allocation purposes. Metropolitan city and urban county members of consortia were treated as individual jurisdictions and other consortia members are include in the balance of state service areas. The HOME program also has a fixed dollar threshold ($500,000) for receiving an allocation. If the jurisdiction receives less than the $500,000 minimum allocation, then the jurisdiction’s share is redistributed to other jurisdictions.

Modified HOME Method: This is the HOME method without the $500,00 threshold.

5 5