SPECIAL REPORT

An Analysis of the Allocation of Federal Homeless Funding March, 2016

A grassroots collaboration of service providers sharing information and knowledge about homeless services

www.nhipdata.org Michael Ullman, Coordinator [email protected]

Background The largest amount of federal money allocated to homeless services is awarded by the U.S. Department of Housing and Urban Development (HUD) through the annual competition for McKinney-Vento homeless assistance funding. Communities request funds through a Continuum of Care, a planning body comprised of homeless service providers, local government departments, and local advocates or housing officials, that represents a geographically-unique area. Continuums typically encompass a large county or city, multiple, often adjacent small counties, or in the case of sparsely populated areas like Wyoming or Delaware, an entire state. There were more than 400 Continuums of Care identified by HUD eligible to receive a portion of the $1.9 billion available for CoC homeless assistance funding in FY2015 1. Applications were submitted in November 2015 with grant awards slated to be announced in late winter/early spring 2016. The purpose of this analysis is to understand the relationship between homeless prevalence and federal funding, and to identify current patterns that may be hindering or helping communities in their efforts to reduce homelessness. It is acknowledged that all federal funding allocations contain idiosyncrasies due to government bureaucracy, methodological imperfections, and political influence. Furthermore, the pattern of funding itself since the beginning of McKinney Vento in 1987 may be influencing current homeless counts. The evolution and change in homeless demographics over the past decades since the allocation formula was first implemented and the large reliance on HUD federal homeless dollars for many communities gives reason to review the current allocation on a Continuum basis to identify any large funding inequities. Comparing funding levels with current homeless bed inventories and estimated homeless counts can also be instructive to understand how the Continuum of Care’s strategies implemented over the past two decades of McKinney Vento funding may have impacted success in reducing homelessness. Data This report uses several sources of data: 1) Data on the HUD Annual Renewal Demand (ARD) and Planning funds allocated for availability to each of the 400 Continuums of Care during 2015 2, 2) Homeless Point-in-Time Count information culled from the Consolidated Application for the 2014 HUD Homeless Assistance application from each of the Continuum of Care 3, and 3) Housing Inventory Count (HIC) data submitted with the 2014 application, which reports the number of beds/units of each type of housing for homeless or formerly homeless persons in each Continuum 4. Actual funding amounts use the 85% threshold of ARD funds since HUD only guarantees the availability of these funds to each Continuum. Any additional funding must be won due to superior application and is not guaranteed. In addition, the “bonus funds” represent 15% of the ARD amount so inclusion would not change the basic relationship found between funding and homeless counts. A total of $1,617,390,013 is allocated for funding and planning under the 2015 homeless assistance funding application. Allocations per Continuum ranged from as high as $98 million for the New York City Continuum that reports total one-day homelessness exceeding 67,000 to many smaller, mostly rural areas receiving less than $250,000 in federal homeless assistance whose homeless one-day counts are often less than 200 persons.

NHIP: Analysis of the Allocation of Homeless Assistance Funding

2

The PIT homeless counts are divided into three types: 1) Emergency shelter counts, 2) Transitional housing counts, and 3) Unsheltered (street) counts. The total count reported from all Continuums during 2014 was 573,133 persons. While accuracy of the emergency and transitional facility counts are considered quite high since enumeration need only a physical count of persons staying at each homeless facility, unsheltered counts continue to vary in quality across communities. A general consensus about the unsheltered counts informs that: 1) They are the only counts or homeless estimates available, and 2) Unsheltered counts have improved over the past decade. Since this report’s analysis is only looking for large differences in allocation patterns, small or moderate inaccuracies in counts are not very troublesome since they will not alter basic patterns and conclusions. HIC data for each Continuum provides an inventory of both of emergency and transitional facility beds, in which people who reside in them are still defined as “homeless”, and permanent housing beds for formerly homeless persons who are no longer considered to be homeless. HIC data is considered to be highly accurate, though errors or exclusion of homeless housing resources do exist. HIC counts of Permanent Housing may suffer from larger unsystematic errors due to the often definitional problem of what is considered “homeless” permanent housing. As with the errors inherent in the homeless counts, small differences should not impact general associations found especially with large Continuums where exclusion of small beds would represent only a very small proportion of the true housing inventory. Analysis In order to compare funding levels across Continuums, this report developed three measures of funding allocations: 1) Allocations of annual federal funding per homeless person. The computation took the total amount of 85 percent of the ARD plus the allocated planning amount divided by the total homeless count as noted in the 2014 PIT Count (emergency, transitional and unsheltered counts). This “funding per capita” measure ranged as low of $100 per capita to over $18,000 per capita. The average funding per capita across the country was found to be $2,822 (dollars per homeless person). 2) Equalized level of funding. This computation took the average allocation of funding per capita from measure #1 and multiplied it by the actual total homeless count from each Continuum. This produced a funding level for each Continuum under an equal allocation scenario. 3) Change needed for equal funding. The change in funding represented the amount of funding increase or decrease needed to arrive at equalized funding based on homeless enumeration. Measure #2 is subtracted from the actual allocation to produce measure #3. In addition, a ratio of Permanent Housing Beds-to-Homeless beds was calculated for each of the Continuums. Bed counts are taken from the 2015 HIC. The ratio is employed as a way of controlling for the extent that HUD’s homeless funding supports permanent housing in communities versus beds that are filled by homeless persons. While HUD funding does not support all beds listing on the HIC, and often less than 50 percent of them, a large percentage of the permanent housing inventory is supported with HUD McKinney Vento funding. In addition, since emergency and transitional beds are “homeless beds”, inventory distribution will impact homeless counts, all things equal.

NHIP: Analysis of the Allocation of Homeless Assistance Funding

3

Results High and Low Allocation Communities Tables 1a and 1b presents the top ten Continuums that receive the most or least amount of funds per capita counted among medium-size or larger Continuums. The threshold for medium size was arbitrarily set at an HUD homeless assistance funding allocation of $1 million or more in 2015. Compared to the overall average of $2,822 per capita across the entire country, a review of the top ten allocations show funding received by some communities exceed 4 to 6 times the average level. The NY-518 Continuum which covers Utica, Rome/Oneida, and Madison counties received an average of $18,759 per capita. Additional review of the counts for the NY-518 Continuum in prior years also shows similar homeless counts of 200 in 2013 and 176 in 2012 and similar levels of funding 5.

Table 1a: Top Ten Highest Allocation per Homeless Census CoC Number and Name NY-518 - Utica, Rome/Oneida, Madison Counties CoC

Homeless Funding

Total Homeless

Funding Per Homeless

$2,981,170

159

$18,749

CT-503 - Bridgeport, Stamford, Norwalk/Fairfield County CoC

$8,024,575

440

$18,238

PA-503 - Wilkes-Barre, Hazleton, Luzerne County CoC

$3,377,335

192

$17,590

PA-603 - Beaver County CoC

$1,386,720

95

$14,597

OH-500 - Cincinnati/Hamilton County CoC

$14,093,437

1,043

$13,512

NY-512 - Troy/Rensselaer County CoC

$2,676,929

216

$12,393

NY-500 - Rochester, Irondequoit, Greece/Monroe County CoC

$10,263,075

838

$12,247

NY-507 - Schenectady City & County CoC

$2,929,024

243

$12,054

MA-517 - Somerville CoC

$1,847,719

154

$11,998

OH-502 - Cleveland/Cuyahoga County CoC

$22,310,639

2,103

$10,609

Table 1b: Top Ten Lowest Allocation per Homeless Census CoC Number and Name FL-509 - Fort Pierce/St. Lucie, Indian River, Martin Counties CoC CA-508 - Watsonville/Santa Cruz City & County CoC CA-504 - Santa Rosa, Petaluma/Sonoma County CoC CA-506 - Salinas/Monterey, San Benito Counties CoC OR-505 - Oregon Balance of State CoC CA-603 - Santa Maria/Santa Barbara County CoC HI-500 - Hawaii Balance of State CoC TX-607 - Texas Balance of State CoC MA-504 - Springfield/Hampden County CoC KY-502 - Lexington-Fayette County CoC

NHIP: Analysis of the Allocation of Homeless Assistance Funding

Homeless Funding

Total Homeless

Funding Per Homeless

$1,275,051

2,591

$492

$1,981,430

3,529

$561

$2,584,247

4,266

$606

$1,906,658

2,962

$644

$2,750,224

4,122

$667

$1,307,179

1,832

$714

$1,700,258

2,206

$771

$7,479,825

8,903

$840

$2,277,409

2,690

$847

$1,340,942

1,544

$868

4

The top ten highest allocation includes three (3) other New York Continuums, two (2) Pennsylvania Continuums, one (1) Massachusetts, one (1) Connecticut and two (2) Continuums from Ohio. The top ten areas receiving the most generous allocation per current homeless person are all either northeastern areas or older industrial towns (Cleveland, Cincinnati). A review of the 10 Continuums with the lowest allocation per homeless person includes primarily west coast or Sunbelt areas with the exception of the Springfield, Massachusetts and the Lexington, Kentucky areas. The Fort Pierce Continuum received only $492 per capita, close to 40 times less than what Utica received. All of the ten low funded Continuums received more than 10 times less than even the lowest of the high allocation communities, and all received less than one-third the national average of $2,822. The lowest ten communities featured both rural areas including three Balance of State Continuums in Texas, Oregon, and Hawaii as well as urban areas like Santa Cruz and Springfield. Appendix 1 contains the allocation for all 400 Continuums. Equalization Adjustments In an effort to understand the extent of differences in allocation among Continuums in real dollar terms, an equalization adjustment was performed that used the average national allocation of $2,822 per homeless person as the funding formula. Tables 2a and 2b present the same set of communities with adjusted funding levels and modification amounts needed to attain this equalization. As expected, high allocation communities would have to see very large decreases in funding ranging between $1.1 million to $16.4 million to reach an equalized level of funding. In contrast, communities currently receiving low levels of support would need increases ranging from $3.0 million to $17.6 million to reach equality. The total change needed (moving funds from high funded Continuums to low funded Continuums) for equalized funding is $459 million out of the total of $1.6 billion, or 28 percent of the total allocation. A total of 198 Continuums would need a decrease in funding with 202 receiving an increase. It must be remembered that funding levels were determined based on formulas heavily influenced by population, poverty and other demographic factors during a time when any local counts of homelessness were unavailable. In addition, HUD has provided many opportunities to increase funding through better performance or strategic selection of projects. All things equal, the funding of transitional housing services versus additional permanent housing units will lower the homeless funding per capita number – is this a correct statement, may need more explanation as it goes against the HI-500 data. While “over” or “under” funded labels are helpful to understand distribution, the analysis remains much more complex since the Continuum actively determines these outcomes.

NHIP: Analysis of the Allocation of Homeless Assistance Funding

5

Table 2a: Top Ten Highest Allocation - Equalized Funding Levels Homeless Funding

Equalized Funding Level Change Needed

CoC Number and Name NY-518 - Utica, Rome/Oneida, Madison Counties CoC

$2,981,170

$448,698

($2,532,472)

CT-503 - Bridgeport, Stamford, Norwalk/Fairfield County CoC

$8,024,575

$1,241,680

($6,782,895)

PA-503 - Wilkes-Barre, Hazleton, Luzerne County CoC

$3,377,335

$541,824

($2,835,511)

PA-603 - Beaver County CoC

$1,386,720

$268,090

($1,118,630)

OH-500 - Cincinnati/Hamilton County CoC

$14,093,437

$2,943,346

($11,150,091)

NY-512 - Troy/Rensselaer County CoC

$2,676,929

$609,552

($2,067,377)

NY-500 - Rochester, Irondequoit, Greece/Monroe County CoC

$10,263,075

$2,364,836

($7,898,239)

NY-507 - Schenectady City & County CoC

$2,929,024

$685,746

($2,243,278)

MA-517 - Somerville CoC

$1,847,719

$434,588

($1,413,131)

OH-502 - Cleveland/Cuyahoga County CoC

$22,310,639

$5,934,666

($16,375,973)

Table 2b: Top Ten Lowest Allocation - Equalized Funding Levels CoC Number and Name FL-509 - Fort Pierce/St. Lucie, Indian River, Martin Counties CoC CA-508 - Watsonville/Santa Cruz City & County CoC CA-504 - Santa Rosa, Petaluma/Sonoma County CoC CA-506 - Salinas/Monterey, San Benito Counties CoC OR-505 - Oregon Balance of State CoC CA-603 - Santa Maria/Santa Barbara County CoC HI-500 - Hawaii Balance of State CoC TX-607 - Texas Balance of State CoC MA-504 - Springfield/Hampden County CoC KY-502 - Lexington-Fayette County CoC

Homeless Funding

Equalized Funding Level Change Needed

$1,275,051

$7,311,802

$6,036,751

$1,981,430

$9,958,838

$7,977,408

$2,584,247

$12,038,652

$9,454,405

$1,906,658

$8,358,764

$6,452,106

$2,750,224

$11,632,284

$8,882,060

$1,307,179

$5,169,904

$3,862,725

$1,700,258

$6,225,332

$4,525,074

$7,479,825

$25,124,266

$17,644,441

$2,277,409

$7,591,180

$5,313,771

$1,340,942

$4,357,168

$3,016,226

Given these interplays, it is best to focus on Continuums where large variation from the average exists – variations that likely cannot be explained by factors controlled by the Continuum’s actions. Reviewing Continuums that receive more than double or less than half of their equalized value is a good start. This threshold would include 130 Continuums for more intensive review, or slightly less than one-quarter of Continuums. Qualitative review at the Continuum level would also be needed to augment homeless estimation and federal funding data, especially given the on-going difficulties of accurate homeless counts and local factors influencing homeless prevalence.

NHIP: Analysis of the Allocation of Homeless Assistance Funding

6

Comparison of Funding with Permanent Housing Units One of the limitations of the prior analysis comparing homeless counts and funding is that HUD homeless assistance funds only Transitional and Permanent Housing (Permanent Supportive Housing and Rapid Rehousing), not Emergency Beds 6. Furthermore, in the past five (5) years, HUD has restricted new project funding to only Permanent Housing projects. Therefore, the correlation of funding and homeless counts needs to also take into account the fact that between one-third to two-thirds of annual Homeless Assistance funding, depending on the Continuum mix, support these permanent units. By definition, all things equal, the greater the number of permanent housing units, the fewer the homeless persons. Suffice it to say, while correlations between levels of homeless funding and homeless counts should likely exist, the fact that funding supports permanent housing (non-housing) units clouds the previous analysis. The funding and success of permanent housing units, which move people out of homelessness, will actually works to increase the funding per capita since homelessness decreases while funding of permanent housing increases. This is one reason that funding discrepancies should likely focus on Continuums with funding levels per homeless person that are significantly different from average (factor of 2 times or greater). In order to factor in permanent housing into the funding analysis presented, an additional characteristic for each Continuum was calculated. Using data from each Continuum’s HIC, a ratio of Permanent Housing units compared to Homeless (Emergency and Transitional) units was computed. While far from perfect 7, this ratio can help inform the extent to which homeless dollars are more heavily focused on permanent housing solutions rather than the maintenance of emergency and transitional units. While emergency and transitional units house people, under the current federal definition the persons residing in this units are counted as homeless. In addition, given the various incentives that HUD has given to permanent housing projects over the last decade, additional permanent units would tend to be sign of communities who effectively “gamed” the HUD funding system to increase funding levels in the long run – would be interesting to point out the change in McKinney PH beds over a five-year span, this would illustrate the skill needed to bring PH units into the CoC. Table 3a and 3b depict the same high and low Continuums with the number of homeless units (emergency and transitional) compared to permanent units and the ratio between them. Looking again at the communities with the highest funding per capita, eight of the ten communities feature permanent-to-homeless bed ratio of 1.48 or higher. Three Continuums reported ratios in excess of 3.40, effectively meaning 10 permanent housing slots for every 3 homeless bed slots. Only two continuums had ratios under 1.0, both small-to-medium communities, with only Somerville, MA having a ratio slightly under the national average of 0.82, or approximately five homeless beds to every four permanent beds. Revisiting the lower funded communities shows a dramatic reversal of the ratio trend. Eight of these ten Continuums reported well below the national funding per capita average with only CA-504 and MA504 above 1.0. The bed inventory in OR-505 Balance of State Continuum contains 4 homeless beds for every 1 permanent bed. Four other Continuums had ratio of 0.53 or below. A simple correlation between this permanent-to-homeless bed ratio and funding per homeless person yielded a modest, but noteworthy and significant, Pearson R equaled to 0.47 (n=401, p