HomeMy WebLinkAboutdraft-home-together-2030-plan-appendices-adaDRAFT Home Together 2030 Appendices
Appendix A. Modeling Solutions to Homelessness in
Alameda County
Appendix B. Summary Description of the Home Together
2030 Racial Equity Analysis
Appendix C. Home Together 2030 Permanent Housing
Program Models
August 2026
1
Appendix A. Modeling Solutions to
Homelessness in Alameda County
Technical explainer for system modeling analysis in the
Home Together 2030 Plan
Home Together 2030 uses system modeling to estimate the scale and mix of key
homelessness response interventions needed to achieve the Plan’s five-year
goals: reducing overall homelessness by one-third and unsheltered
homelessness by 50 percent.
System modeling analysis is not predictive but rather a useful decision-support
tool that illustrates the relationship between key programmatic goals like
reducing new entries into homelessness, expediting exits to housing, and
reducing the share of people experiencing homelessness who are living
unsheltered. By estimating the new inventory (and associated investments)
required to realize homelessness reduction goals, the analysis points the way
toward a homelessness response system that is sufficiently scaled and
resourced to efficiently meet the needs of people experiencing or at risk of
homelessness in Alameda County.
Beginning in early 2025, All Home partnered with Alameda County Health’s
Housing and Homelessness Services (H&H) to lead system modeling analysis for
the Home Together 2030 strategic plan. The data inputs and driving assumptions
were selected through close collaboration with county staff and members of a
technical working group composed of service providers, people with lived
experience of homelessness, and staff from several cities within the county.
This report serves as a technical explainer of the methodology behind All Home’s
system modeling analysis, organized into the following sections:
● Opening sections (System Modeling and Its Uses, Model Design and
Scope, and Strategic Leverage Points) introduce the concept and key
2 MODELING SOLUTIONS TO HOMELESSNESS FOR THE HOME TOGETHER 2030 PLAN
applications of system modeling, describe the motivations behind All
Home’s approach to the analysis, and highlight the program investments
and performance improvements most likely to yield a more effective,
efficient homelessness response system.
● Analytical Approach explains the major calculation steps, data inputs, and
assumptions used to calculate inventory needs and associated funding
requirements;
● Investment and Inventory Targets reports the modeled resources needed
to sustain current capacity and expand the system in support of the Home
Together 2030 goals.
● The Limitations section acknowledges important boundaries of the
analysis, while Looking Ahead describes how the modeling estimates
should be revisited, revised, and used to inform monitoring and course
correction while implementing the Home Together 2030 plan in the years
ahead.
System Modeling and its Uses
In order for communities to most effectively and efficiently address
homelessness, they must develop a working understanding of the resources
required to achieve their strategic goals. This is a challenging task. It requires
using limited data to project into an uncertain future, while the homelessness
“system” consists of numerous and interacting program areas that don’t neatly fit
within well-defined boundaries. Still, a high-level roadmap is required to guide
strategy, investments, and coordination toward the most efficient, effective
solutions.
System modeling is an analytical tool that helps provide that roadmap. The
analysis generates estimates of the inventory and investments required to reach
strategic goals for reducing homelessness. All Home designed the system
modeling tool described below to be flexible to accommodate nuanced local
needs or priorities and to call out “leverage points” available to optimize system
performance and outcomes for households being assisted. All Home’s tool also
allows estimates to be easily revised to reflect updated data and evolving
contexts.
Findings from system modeling analysis provide a helpful guide for
implementing strategic plans by designating targets for the level of housing
assistance needed to make meaningful progress towards systemwide goals. But
the estimates produced through system modeling also serve an important role in
3 MODELING SOLUTIONS TO HOMELESSNESS FOR THE HOME TOGETHER 2030 PLAN
strategic communication: they help convey to the public and their elected
representatives the level of investment required to make timely and meaningful
progress in reducing homelessness. The estimates also help explain limited
progress in reducing homelessness over recent years, despite substantial–yet
insufficient–effort and investment. Further, data-driven analysis documenting
resource gaps and generating investment priorities can support case-making for
additional public and private sector funding.
Model Design and Scope
All Home’s system modeling tool begins with the County’s homelessness
reduction goals. It estimates how many people will experience homelessness
during each year, including those already homeless and those expected to enter
homelessness, and then calculates how many people must move into housing or
otherwise resolve their homelessness to reach the annual reduction goal. It also
estimates the number of interim housing units that the system would need to
achieve reduction goals for unsheltered homelessness. The analysis is informed
by a host of assumptions including the appropriate mix of housing interventions
needed to meet the varied needs of the county’s homeless population, the
pathways along which households will progress from program to program, and
the length of time they will receive the various forms of assistance included in
the system model.
The model is focused on three primary program areas1:
1. Targeted homelessness prevention- flexible and relatively modest
financial assistance with services intended to keep people from becoming
homeless in the first place;
2. Interim housing- temporary settings intended to expedite exits to
permanent housing and reduce the traumas associated with living
unsheltered;
3. Permanent housing solutions- a spectrum including both time-limited
and ongoing affordable housing assistance, accompanied by a range of
supportive services enabling exits from homelessness and sustained
tenancy in permanent housing.
All Home’s system modeling tool was designed to call to attention to several
consequential yet often-overlooked characteristics of homelessness and the
system designs to address it:
1 Problem solving assistance–sometimes referred to as diversion or rapid resolution
programs–is also included in the modeling for a portion of households resolving
without enrollment in permanent housing programs.
4 MODELING SOLUTIONS TO HOMELESSNESS FOR THE HOME TOGETHER 2030 PLAN
● The dynamic nature of homelessness- The number of people
experiencing homelessness at any given time reflects both how many
people enter homelessness and how quickly people exit. Lasting
reductions therefore require a balanced strategy that prevents avoidable
entries, expands housing exits, and reduces the length of time people
remain homeless.
● The interrelation of system program areas- the effectiveness of any part
of the system is shaped by the performance and resource availability in all
other parts of the system. Successful exits from interim housing depend
on the availability of permanent housing assistance, while the number of
interim housing units needed in a given year can be reduced when
prevention assistance reduces new entries. All Home’s model tries to call
out program interrelations and leverage points.
A variety of data sources–all specific to Alameda County–are used for data inputs
in the model and to inform assumptions regarding system needs and program
performance. Each of the following sources is limited and imperfect alone, but
together they provide a wealth of information to guide the system modeling
process:
● Point-in-time (PIT) count data generally inform estimates of baseline
population totals (both for the overall population experiencing
homelessness on any given night and the unsheltered homeless
population specifically).
● Administrative data from the county’s Homelessness Management
Information System (HMIS) provide information on numerous metrics
needed for the modeling, including estimates of the number of people
becoming newly homeless each year, the behavioral health needs of the
population, and the length of time households remain in various housing
assistance programs.
● Annual Housing Inventory Count (HIC) reports detail the system’s current
inventory of interim and permanent housing.
The scope of the model includes only assistance programs designed explicitly to
serve people experiencing or most imminently at risk of experiencing
homelessness. Focusing solely on these programs (which report into HMIS)
allows us to set manageable boundaries around the program areas included in
the modeling (and their interrelation), and provides a key data source for tracking
progress during implementation of strategies included in the Home Together
Plan. This means that there are important populations and programs not directly
included in the model even though they are undeniably impacted by housing
insecurity and efforts to reduce it. For example, households that are doubled-up,
in precarious housing, or who experience homelessness but never engage with
5 MODELING SOLUTIONS TO HOMELESSNESS FOR THE HOME TOGETHER 2030 PLAN
the response system are not included in assumption-setting, even though we do
account for the proportion of this population that does engage with
homelessness assistance programs in a given year. Upstream safety net
programs like Medi-Cal or CalFresh–both clearly impacting the ability of low-
income households to maintain stable housing–are not included in the model.
Nor are efforts to produce or maintain the broader stock of affordable housing
not designated to serve households who have experienced homelessness. These
programs and their anticipated impact on homelessness are incorporated into
strategic planning discussions, even if they are not directly incorporated into the
system modeling analysis.
Strategic Leverage Points
The system modeling project focuses attention on key investments areas and
performance improvements that provide strategic leverage points for
communities seeking the most effective, efficient strategies for reducing
homelessness:
➢ Reduce new entries and returns to homelessness.
○ Scale prevention assistance sufficiently to reduce the number of
people becoming homeless each year, thereby reducing the need
for more resource-intensive interim and permanent housing
inventory.
○ Target prevention assistance efficiently to households most likely
to become homeless using research-based vulnerability criteria to
inform program eligibility and enrollment.
○ Expand access to problem-solving and rapid-resolution assistance
so more households can avoid entering homelessness or resolve a
housing crisis quickly.
○ Increase supports associated with time-limited subsidies so that
program assistance does not end before housing stability is
secured.
➢ Expand, expedite, and diversify exits to permanent housing.
○ Scale inventory of permanent housing solutions allowing more
households to move out of homelessness each year, while
reducing the average length of time households remain homeless.
○ Diversify the portfolio of permanent housing programs to reflect
the varied needs of people experiencing or at risk of
homelessness.
○ Refine program matching and prioritization to more rapidly provide
less intensive housing supports to populations with lower needs.
6 MODELING SOLUTIONS TO HOMELESSNESS FOR THE HOME TOGETHER 2030 PLAN
➢ Increase flow through the system to reduce the time people remain
homeless before connecting to permanent housing and interim housing .
○ Increase capacity to help people move directly from unsheltered
settings into permanent housing, when appropriate.
○ Reduce the average duration of stay in interim housing prior to
moving into permanent housing, while ensuring service needs are
met.
While these leverage points indicate opportunities for improving systemwide
outcomes and performance, strategic efforts must balance such efficiencies with
ensuring the needs of the population being served are met as fully as possible.
Analytical Approach
The outputs of the system modeling (primarily annual inventory targets and the
associated investments required) are the result of four broad analytical steps.
This section describes the calculations carried out in each of those steps, along
with the data inputs (e.g., existing unit counts or per-unit costs) and assumptions
incorporated into the analysis.
Step One: Goal Setting
Two key strategic goals serve as the primary drivers for all calculations in the
model:
● the five-year goal for reducing the overall population of people
experiencing homelessness at a given time, and
● The five-year goal for reducing the number of people experiencing
unsheltered homelessness at a given time.
Each goal is set as a percent reduction in the population experiencing
homelessness in the baseline year, based on the most recent PIT count
(conducted in January of 2026). Annual progress toward the five-year goal may
be distributed evenly among the five years modeled, or adjusted to model a
“ramp-up” strategy (where more progress is made in later years of the plan), or a
“surge” strategy (where greater progress is assigned to earlier years in the plan).
Distinct goals are set both for adult only households and for households
including both adults and children. This distinction follows those used by the U.S.
Department of Housing and Urban Development (HUD) and allows communities
to incorporate population-specific targets into the analysis. Distinguishing adult-
only households from households including adults and children is important from
7 MODELING SOLUTIONS TO HOMELESSNESS FOR THE HOME TOGETHER 2030 PLAN
an implementation standpoint because much of the existing housing inventory–
including both permanent supportive housing (PSH) units and emergency shelter
beds currently in operation–is designated to serve one population or the other.2,3
Additionally, the programs and housing units used to serve each type of
household typically differ by cost and design, distinctions that are helpful to
incorporate into the system modeling.
Table one reports the population reduction goals driving the estimates of
inventory and investment needs included in the Home Together 2030 plan.
Figure one, below, visualizes the modeled trajectory in historical context based
on point-in-time count trends in the county over the last decade.
Table 1. Population reduction targets included in system modeling
2 As of 2024, 21% (720 of 3,472 of Alameda County’s emergency shelter and
transitional housing beds were designated for adults with children, while 79% (2,740)
beds were designated for adult-only households. For permanent housing, about 36%
of the county’s inventory (2,333 of 6,491 total beds) were dedicated to adults with
children.
3 Adult-only households and households with children are modeled separately
because they generally require different unit sizes and cost assumptions. Differences
in needs (e.g., older adults) within each category are reflected through assumptions
about the appropriate mix of interventions and are further addressed through
program matching, prioritization, and implementation.
Overall population reduction target 33% Reduction
Cumulative Progress toward 5-yr goal Y1 Y2 Y3 Y4 Y5
20% 40% 60% 80% 100%
Unsheltered population reduction target 50% Reduction
Cumulative Progress toward 5-yr goal Y1 Y2 Y3 Y4 Y5
20% 40% 60% 80% 100%
8 MODELING SOLUTIONS TO HOMELESSNESS FOR THE HOME TOGETHER 2030 PLAN
Figure 1. Population reduction targets in historic context
Step Two: Calculating Population Totals & Assistance Needs
PROJECTED POPULATION TOTALS
To calculate the number of households requiring assistance to achieve these
strategic goals, we begin by estimating the total number of people expected to
experience homelessness over the course of a given year. Baseline estimates for
people experiencing homelessness at the start of the plan (both the overall
population and the unsheltered subset) are tied to the most recent point in time
(PIT) count data. To these baselines we add anticipated new entries into
homelessness. New entries include people expected to become homeless for
the first time as well as those returning to homelessness after an earlier exit to
permanent housing.
HMIS data are used to calculate anticipated annual “first time” homelessness as
defined by the California Inter-Agency Council on Homelessness (Cal ICH).4 This
metric includes all individuals enrolling in any homelessness assistance program
with no prior enrollment in such programs in the prior two years.5 We use the
4 For further details on how Cal-ICH defines and calculates system performance
measures, see https://www.hcd.ca.gov/funding/hhap/resources-ca-spms.
5 The traditional measure of first-time homelessness–as defined by HUD–includes
only people enrolling in “residential” programs (like shelter or rapid re-housing) but
excludes people only engaging in non-residential services or outreach programs.
This traditional definition can be artificially constrained by limited shelter availability,
9 MODELING SOLUTIONS TO HOMELESSNESS FOR THE HOME TOGETHER 2030 PLAN
average of the three most recent years of data for our input in the model. Absent
a clear trend line based on recent years data, we hold this input steady across all
five years included in the model.
Anticipated returns to homelessness in the first year of the plan are based on
HMIS data indicating the most recent rate of people reengaging with
homelessness assistance programs within 24 months of a prior exit to housing.
For the remaining years of the strategy, separate assumptions are made for the
return rates for each permanent housing solution and any other modeled
pathway out of homelessness.
Table 2. Population Data Inputs
Type Input Source
Baseline population 8,201 2026 Point-in-time count
Baseline unsheltered population 5,202 2026 Point-in-time count
Anticipated annual first-time
homelessness 8,200 HMIS
Baseline return rate 10% HMIS
PREVENTION
The system model incorporates strategic goals for reducing new entries into
homelessness each year. For first-time homelessness, the goal assigns a target
percent reduction in anticipated new entries. For returns to homelessness, the
goal assigns a target percent of all households exiting to positive housing
destinations that are assumed to return to a homelessness response system
within 12 months. (So, for returns, improvement entails achieving a lower percent
of returns than currently occur, based on most recent system performance data.)
Determining how many households must receive prevention assistance in order
to reach these prevention targets requires setting assumptions for the efficiency
with which prevention assistance will be targeted to households who would
become homeless without the intervention. In system modeling terms, this
and also misses a large subset of people experiencing homelessness in Alameda
County who never enroll in residential programs. In the HT 2030 modeling, we use
the broader Cal-ICH definition of first-time homelessness, which is a departure from
the way this metric has previously been calculated for the purposes of tracking and
monitoring of the 2026 Home Together plan. For a comparison of Cal-ICH and HUD
definitions used to calculate system performance measures, see
https://bcsh.ca.gov/calich/documents/crosswalk_of_system_performance_measur
es.pdf
10 MODELING SOLUTIONS TO HOMELESSNESS FOR THE HOME TOGETHER 2030 PLAN
requires setting a ratio for the number of households served for each reduced
new entry into homelessness. Targeted homelessness prevention is a relatively
new program model, and while robust research does exist evaluating the
program’s effectiveness and efficiency, evidence to guide efficiency assumptions
is limited.
To prevent first-time homelessness, we assume that ten households will need to
be assisted to prevent one from becoming homeless during a given year, which
equates to a 10 percent targeting efficiency. We assume that returns to
homelessness can be prevented with a somewhat higher level of efficiency,
based on the expectation that households who have previously moved out of
homelessness with the assistance of response system programs represent a
relatively smaller and better-known population. We assume three households
will need to be assisted to prevent one from returning to homelessness, equating
to 33% targeting efficiency.
Table 3. Prevention Goals and Efficiency Assumptions
Description Input
First-time Homelessness- Reduction Goal
5% in Y1
10% by Y5
First-time Homelessness- Targeting Efficiency
Assumption
10% in Y1
15% by Y5
Returns- Target Rate 8%
Returns- Targeting Efficiency Assumption 33%
Prevention goals are notably aspirational, and highlight the systemwide
efficiencies created as a result of significantly reducing the number of people
becoming homeless each year. By reducing new entries into homelessness, the
permanent and interim housing inventories (each a more time- and resource-
intensive intervention) needed to reach strategic reduction goals are substantially
decreased. Other important benefits of expanding prevention assistance–
including the reduced trauma of households living unsheltered or increased
reliance on emergency medical services–while not included in the system
modeling, should be considered in setting strategic investment priorities.
Targeting efficiency assumptions are similarly aspirational and emphasize the
importance of relying on research-based vulnerability criteria to inform program
eligibility requirements and distribution of limited assistance resources.
11 MODELING SOLUTIONS TO HOMELESSNESS FOR THE HOME TOGETHER 2030 PLAN
RESOLUTIONS & SYSTEM EXITS
The total number of people projected to experience homelessness in a given
year sums the population experiencing homelessness at the start of the year with
the number of people projected to enter into homelessness over the course of
the year, after accounting for the impact of prevention assistance. The difference
between this total annual population and the target population for a given year
equals the number of people who must move out of homelessness in the year to
achieve strategic reduction goals.
In the system model, households are assumed to move out of homelessness
and/or leave the local response system via two general pathways:
● with Problem Solving Assistance or through Self-Resolution- these
households only lightly engage response system programs, and resolve
homelessness (or leave the County’s response system) without receiving
permanent housing assistance.
● with Permanent Housing Assistance- these households resolve their
episode of homelessness through enrollment in one of the subsidized
permanent housing programs included in the system model. Some or all
may be assumed to stay in interim housing prior to their enrollment in a
permanent housing program.
Households following either pathway may temporarily stay in interim housing
prior to their exit, but interim housing is not in itself considered an exit destination.
HMIS data provide useful but incomplete information to guide assumptions for
the share of the homeless population expected to follow each of these general
pathways. They indicate, for example, the share of known households exiting
homelessness who receive subsidized housing assistance (and the share exiting
without such assistance).6 HMIS data also indicate the share of people who
ultimately enroll in subsidized permanent housing programs who stay in interim
housing prior to their permanent housing exit, although this datapoint is
6 Those exiting homelessness to a “positive destination” without subsidized housing
assistance includes people moving in with family or friends on a long-term basis, or
those moving into rental units without receiving housing subsidies. System data
indicate that more than half of adult-only households recorded as exiting to
permanent housing do so without subsidized housing assistance. More than 75
percent of households with both adults and children who exit to permanent housing
do so without such assistance.
12 MODELING SOLUTIONS TO HOMELESSNESS FOR THE HOME TOGETHER 2030 PLAN
constrained by the insufficient inventory of interim housing currently available.7
However, large percentages of people enrolled in response system programs in
a given year “exit” to unknown destinations.8 Some people who exit the
homelessness response system to an unknown destination may return to
homeless settings, while others leave the system or resolve their episode of
homelessness, but it is difficult to accurately predict the balance. Setting
assumptions for pathways out of homelessness is best informed by a diversity of
perspectives on the balance of interventions that will meet the housing needs of
the local population and result in more optimal system performance.
The model allows us to distinguish pathways for the existing homeless
population–those homeless at the beginning of a given year–and the newly
homeless population–those entering into homelessness over the course of a
given year. Since individuals remaining homeless for a longer period are
generally expected to have a higher level of need, this distinction allows us to
designate interventions more likely to result in housing stability for the respective
populations.
Table 4. Distribution of Resolutions/System Exits, by pathway
Description % following pathway
Problem Solving and/or self-resolve 25% existing/returns
45% newly homeless
Permanent Housing Assistance 75% existing/returns
55% newly homeless
A variety of permanent housing program types can be included in the model to
meet the needs and priorities of a particular community. The system modeling
for the Home Together 2030 plan includes four permanent housing program
types: time-limited subsidies akin to rapid rehousing, dedicated affordable
7 The landmark CASPEH study conducted by UCSF’s Benioff Homelessness and
Housing Initiative found that 45 percent of people living outside wanted but were not
able to access interim housing. See Unsheltered Homelessness: Findings from the
California Statewide Study of People Experiencing Homelessness.
8 In recent years, roughly 4,500 people exited to “unknown” destinations annually,
with more than 60% of this group never subsequently re-engaging with
homelessness response system programs.
13 MODELING SOLUTIONS TO HOMELESSNESS FOR THE HOME TOGETHER 2030 PLAN
housing, permanent supportive housing (PSH), and PSH for medically frail
populations.”9
While there is no definitive method of determining the right balance of program
types to meet the needs of those being served, local HMIS data provide some
instructive guidance. We considered the portion of the population served in
recent years identified as experiencing chronic homelessness, or recorded as
having multiple behavioral health needs, as proxies for the population best
matched with PSH. We considered the portion of the population indicating
current or recent experiences having earned income and being able to afford
rent as proxies for the population suitably matched with time-limited assistance.
We also relied heavily upon expert opinion from people with lived experience of
homelessness, service providers, and administrators to inform the modeled
assumptions regarding the appropriate distribution of permanent housing
assistance.10
Different distributions of permanent housing assistance were designated for
adult-only households and households including adults and children, and for
both the existing and newly-homeless populations within each group.
Table 5. Distribution of Permanent Housing Assistance, by population
Population Rapid
Rehousing
Dedicated
Affordable
Housing
Permanent
Supportive
Housing
(PSH)
PSH for
medically
frail
populations
Adult Only
Households
Existing/Return 30% 35% 25% 10%
Newly Homeless 35% 35% 20% 10%
Adults w/ children
Households
Existing/Return 45% 30% 20% 5%
Newly Homeless 55% 30% 10% 5%
9 See the Home Together 2030 Program Model Descriptions for additional detail on
each type of permanent housing assistance and the populations they are best suited
to serve.
10 Recent program enrollment data does not neatly translate to the best balance of
interventions to meet needs since prioritization practices have historically matched
households with programs that are helpful but insufficient to ensure housing stability.
Existing inventory is also more likely to be informed by the eligible uses of particular
funding sources than the distinct needs of local homeless households.
14 MODELING SOLUTIONS TO HOMELESSNESS FOR THE HOME TOGETHER 2030 PLAN
INTERIM HOUSING ASSISTANCE
To determine the total number of people expected to stay in interim housing in a
given year, the system model sums (a) the population staying in interim housing
prior to an exit to permanent housing in that year with (b) the population requiring
interim housing in order to reach the designated goal for reducing unsheltered
homelessness. While historical data provide a starting point for these
calculations, we know that interim housing use in recent years has been
significantly constrained by an insufficient supply of shelter beds and units.
Therefore, historic utilization should not be interpreted as a measure of the full
need for interim housing. The modeled target reflects the estimated additional
capacity that is needed to achieve the Plan’s goal for reducing unsheltered
homelessness, while recognizing practical limits on how quickly interim housing
can be expanded and the need to maintain balanced investment in other parts of
the homelessness response system. It should therefore be understood as a
strategic five-year investment target—not an estimate of how many people
would benefit from immediate access to a safe alternative to living unsheltered.
Using HMIS data from recent years, we can determine the percent of people
moving into rapid rehousing or PSH programs who stayed in emergency shelter
or transitional housing sites prior to moving into permanent housing. Based on
data from fiscal year 2023-24, we calculated that approximately 45 percent of
adult-only households stayed in shelter or transitional housing prior to their
permanent housing placement. In the modeling, we assume a slightly broader 50
percent of single adults will stay in an interim housing site prior to exiting to
permanent housing. For households including adults and children, the rate
calculated using HMIS data was just under 60 percent. In the modeling, we
assume 70 percent of these households will stay in interim housing prior to
moving to permanent housing.
After accounting for all the projected system exits and resolutions in a given year,
the system model calculates the number of households remaining homeless
who will stay in interim housing at some point during that year. This calculation is
informed directly by the designated goal for reducing unsheltered homelessness
by 50 percent over five years. The difference between the target unsheltered
population for a given year and the projected population remaining homeless
(i.e., not exiting the system or moving into housing) equals the number of
additional households projected to stay in interim housing during the year.
15 MODELING SOLUTIONS TO HOMELESSNESS FOR THE HOME TOGETHER 2030 PLAN
Step Three: Calculating Inventory Needs
PREVENTION
To calculate the number of prevention interventions required in a given year, we
begin by identifying the targeted reduction we aim to achieve in first-time and
returns to homelessness. We then apply the targeting-efficiency assumptions
described above to estimate how many households must receive assistance to
achieve the reduction goal. For first-time homelessness, a 10 percent targeting-
efficiency assumption means that approximately ten households must be served
to prevent one household from becoming homeless. For returns, a 33 percent
targeting-efficiency assumption means that serving approximately three
households is needed to prevent one household from returning to homelessness.
PERMANENT HOUSING SOLUTIONS
The model estimates the number of households that will need each type of
permanent housing intervention by applying assumptions about the share of
households whose homelessness can be resolved through each intervention. It
then compares that need with the existing inventory for each program type,
based on the most recent Housing Inventory Count and information provided by
County staff. An assumed turnover rate is used to estimate how many existing
units or program slots will become available to serve new households each year.
Any remaining gap between the estimated need and available inventory
represents the additional capacity that must be added in a given year.
Table 6. Permanent Housing Starting Inventory and Turnover Rates
Type Existing Units Turnover Rate
for Adult-only Households
Rapid Rehousing 422 80%
Dedicated Affordable Housing 0 15%
Permanent Supportive Housing 3722 10%
PSH for medically frail populations 60 10%
for Adult-with-Children Households
Rapid Rehousing 295 50%
Dedicated Affordable Housing 0 15%
16 MODELING SOLUTIONS TO HOMELESSNESS FOR THE HOME TOGETHER 2030 PLAN
Permanent Supportive Housing 456 10%
PSH for medically frail populations 0 10%
For some permanent housing models, added inventory includes both newly
constructed units (which involve capital costs plus ongoing operating and service
costs) and new scattered site units (requiring only additional operating and service
costs). The appropriate mix of newly constructed versus scattered site units
should be guided by local priorities and pipeline expectations. Production data
reported on the state’s Housing Element Implementation and APR Data
Dashboard provides useful guidance regarding the number of newly constructed
units that can feasibly be completed each year of the modeled strategy.
New permanent housing units/slots in the HT 2030 model are added in the
following proportions:
● Rapid Rehousing: 100% scattered site
● Dedicated Affordable Housing: approximately 75% scattered site, 25% new
construction.
● Permanent Supportive Housing: 50% scattered site, 50% new construction
● PSH for Medically Frail Populations: 100% new construction
INTERIM HOUSING
Each interim housing unit or bed can potentially be used by multiple individuals
or households over the course of a single year. We calculate the total inventory
needed in a given year by multiplying the total households projected to stay in
interim housing by their assumed average length of stay (in months). Length of
stay assumptions are based on HMIS data indicating recent system performance
and can be modified to reflect strategic aspirations to improve system flow and
expedite movement from interim settings into permanent housing.
As with permanent housing solutions, existing interim housing inventory is based
on most recent HIC data and confirmed by county and city staff. New inventory
added equals the gap between units/beds available in a given year and those
required to meet needs associated with achieving strategic goals.
Starting interim housing inventory (including emergency shelter, transitional
housing, and safe haven projects) includes:
● 2,740 beds/units for adult-only households.
● 240 units (including 720 beds) for households including adults with
children.
17 MODELING SOLUTIONS TO HOMELESSNESS FOR THE HOME TOGETHER 2030 PLAN
Step Four: Cost Calculations
Per-unit cost inputs in the model represent the anticipated average cost
associated with either operating or constructing a single unit/slot in a given
program area. In practice, actual program costs will vary according to a
number of factors. Newly added PSH units, for example, may involve new
construction or acquisition and rehab, resulting in substantially different per-
unit capital costs. Operating and service costs vary according to the needs of
the population being served, program design (including site-based vs.
scattered site), and economies of scale. Cost inputs were selected in
consultation with county staff and were informed by review of actual program
costs and aspiration toward investment in best-practice models.
Note that the capital costs included in the model for new permanent housing
units do not capture the entire per-unit development costs, but only the subsidy
provided through state and local programs after accounting for tax financing
and hard debt (a mortgage loan). This subsidy portion included in the system
modeling represents approximately 35% of total development costs.11
All starting costs increase over the course of the modeled timeline by an
inflation rate of 3% per year. New start up or capital costs are associated with
the year in which a new unit comes on line. Operations/services costs are
based on the cumulative inventory for a given program area, and include
units added in prior years in the model.
Table 7. Per-Unit Costs, by program area
Program Capital Costs* Operating/Services
Costs
Targeted Prevention Assistance - $7,000
Rapid Rehousing - $24,000-$30,000
Dedicated Affordable Housing $250,000-$300,000 $20,000-$28,000
Permanent Supportive Housing $250,000-$300,000 $25,000-$30,000
PSH for medically frail populations $250,000-$300,000 $28,000-$35,000
11For more on this, see The California Affordable Housing Production Pipeline,
Enterprise Community Partners, 2025.
18 MODELING SOLUTIONS TO HOMELESSNESS FOR THE HOME TOGETHER 2030 PLAN
Interim Housing $180,000-$230,000 $35,000-$40,000
*Note- capital costs include only local subsidy, approximately 35% of total
development costs.
Investment & Inventory Targets
The table below reports estimates for the required inventory and associated
costs for two inventory categories:
● Preservation- maintains program inventory at approximately current
levels.
● Expansion- system capacity required to achieve the 33 percent
reduction in overall homelessness and 50 percent reduction in
unsheltered homelessness
Both inventory and cost estimates have been rounded to avoid implying
greater precision than system modeling is able to produce. Estimates for both
preservation and expansion inventory are based on average, per unit costs.
Estimates for current inventory are based on most recent data at the time of
analysis, but fluctuate regularly and may have changed since that time.
Table 8. Inventory and Investment Target Summary
Preservation Expansion
Targeted Prevention
Households Assisted ~500 HHs/year 4,700 HHs/year
Avg. Annual Ops/Services Costs $3.7M $37.0M
Permanent Housing Solutions
Total HHs Exiting to Subsidized Housing 2,400 HHs/year 2,900 HHs/year
Time-limited Subsidies 717 slots 1,400 slots
Dedicated Affordable Housing - 3,500 units
Permanent Supportive Housing 4178 units 4,900 units
PSH for medically frail populations 50 units 2,000 units
Avg. Annual Ops/Services Costs $132.4M $337M
5-year total Capital Costs - $1.87B total
Interim Housing
19 MODELING SOLUTIONS TO HOMELESSNESS FOR THE HOME TOGETHER 2030 PLAN
Households Assisted 5,100+ HHs/year 8,000 HHs/year
Units in Operation 2,980 Units 3,400 Units
Approx. Program Costs $105.7M $136M
5-year total Capital Costs - $98M total
Total Annual Ops/Services Costs $242M $510M
Limitations
Like any attempt to model complex human systems, All Home’s system
modeling tool cannot fully represent real-world conditions or precisely predict
how the system will perform. Many broad social forces–future fluctuations in the
economy or policy decisions made at the state or federal level–are not directly
included in the analysis even though their impact on homelessness and response
systems is undeniable. Pursuing the high-level estimates generated by the
model requires further and more-nuanced implementation planning. System
modeling serves as a roadmap to set the direction toward a better-resourced
and better-balanced homelessness response system, but an extraordinary
amount of decisions, collaboration, and problem-solving will be needed to make
it a reality.
There are several more specific limitations of the tool and estimates worth
acknowledging, some with important implications for implementation of the
Home Together plan. First, while we believe that we are drawing on the best
available data to serve the purposes of the system modeling, data sources on
homelessness are notoriously imprecise. For example, available data for the
number of households both experiencing and resolving episodes of
homelessness in a given year are indicative, but almost surely undercounts.
Our high-level modeling estimates—which lump solutions into broad
categories—are not intended to imply that the needs of people experiencing
homelessness are homogenous. The programs and practices required to best
assist those vulnerable to or currently experiencing homelessness are more
nuanced and detailed than our estimates can convey.
Other limitations that are important to note and consider moving forward include:
● While the system modeling includes an expansive collection of
homelessness-related programs, upstream safety net programs and
much affordable housing serving low-income households are beyond the
scope of the analysis. These programs play a key role promoting housing
20 MODELING SOLUTIONS TO HOMELESSNESS FOR THE HOME TOGETHER 2030 PLAN
stability and preventing homelessness, and significantly shape the need
for programs included in the model, even if the broad array of social
services and affordable housing programs are not directly included in the
analysis.
● Because the modeling is goal-based, it makes no assumptions about
whether sufficient funding will be available to meet the investment
targets. It also does not account for restrictions attached to specific
existing or future funding sources. Analysis indicating the volume of
resources required to make timely progress is useful, even if those
resources are not currently available. However, thoughtful leadership,
sustained effort, and expertise will be required to negotiate a complex
and shifting policy landscape to prioritize available funding for the most
pressing system needs.
● Estimates of current program spending are approximations based on
average per-unit costs, not detailed accounting of actual program costs.
Despite limitations that apply to nearly any practical tool for assessing resources
required to address homelessness, system modeling provides valuable
direction-setting when used in conjunction with additional strategic planning and
community engagement practices.
Looking Ahead
The system modeling estimates reported above will be regularly revisited and
updated during implementation of the Home Together 2030 plan. The inventory
and investment targets–along with the key assumptions incorporated into the
modeling–will need to be assessed and refined to allow for improvement and
course correction in light of new data (e.g., updated Point-In-Time Count results)
or an evolving policy and funding landscape.
System modeling brings into view an array of useful indicators to track and guide
system performance improvements. Among the most consequential metrics
(yielding actionable insights) are the topline indicators included in the Home
Together 2030 plan:
● The number of people experiencing first-time homelessness annually.
● The number of people returning to homelessness (or re-engaging with
the homelessness response system) after a previous exit to housing.
● The number of people experiencing chronic homelessness.
● The number of people resolving an episode of homelessness by moving
into permanent housing.
21 MODELING SOLUTIONS TO HOMELESSNESS FOR THE HOME TOGETHER 2030 PLAN
● Equity (or disparities) in each of the above in terms of race or ethnicity,
age, disability status, etc.
Moving forward, the System Modeling Technical Working Group will reconvene
as needed to assess these and other metrics of progress and provide guidance
on updates to the Plan’s inventory targets. All Home looks forward to providing
ongoing technical assistance to the County and supporting efforts to make the
goals of the Home Together 2030 plan a reality.
APPENDIX B. SUMMARY DESCRIPTION OF THE HOME
TOGETHER 2030 RACIAL EQUITY ANALYSIS
1
A. Introduction
Alameda County contracted with Focus Strategies to conduct a Racial Equity Analysis (REA) to
inform the development of the Home Together 2030 Plan, Alameda County’s strategic
roadmap for a more equitable and effective homelessness response system. This analysis
builds on work conducted during Home Together 2026 planning to identify disparities in
access and outcomes within the homelessness response system. Given the substantial and
persistent racial disparities documented through Point-in-Time Count (PIT) data, Homeless
Management Information System (HMIS) data, and prior County analyses, the County
undertook a deeper examination of patterns by race and ethnicity during the planning
process to better understand areas of progress and where continued focus may be needed
under Home Together 2030.
The overarching goal of the 2025 REA was to assess how people of different racial and ethnic
identities experience the Alameda County Homelessness Response System and to identify
where disparities occur across homelessness response system access, receipt of services and
housing and other outcomes. Focus Strategies used a quantitative approach to the REA,
informed by community voice.
To ensure that the REA analysis was translated into meaningful action in the Home Together
planning process, the County convened a Racial Equity Analysis Technical Working Group
(REA TWG). The REA TWG reviewed information from Focus Strategies and Alameda County’s
Housing and Homelessness Services Department (H&H), interpreted results of the REA,
surfaced community-rooted insights, and developed recommendations for addressing
documented disparities and improving system access and outcomes. The 10-member group
contributed local knowledge, system expertise, and lived experience of homelessness to
forming insights and recommendation development.
This document summarizes the REA scope, the type and range of analyses performed, the
critical role of the REA TWG, and the high-level findings that were most instrumental in
shaping the recommendations and actions of the Home Together 2030 Plan. Detailed results
of the analyses completed were shared with the REA TWG and the Home Together 2030 Task
2
Force, and a technical report was provided to H&H, outlining the specific methodology
utilized for the analysis.
B. Scope of Analysis
The REA used data from multiple sources to examine patterns of representation, access, and
outcomes across the Alameda County Homelessness Response System. Analyses using HMIS
data were conducted at the household level (which may represent a solo individual or the
head of a household of two or more), deduplicated, and examined separately for adult-only
and family households to account for differences in need, program eligibility, and service
pathways. Some analyses conducted included comparisons across multiple datasets including
the Homeless Management Information System (HMIS), the Point‑in‑Time (PIT) Count, US
Census data, and prevention program data to better understand representation, disparities,
and patterns across the system over time.
The scope of analysis included:
1. Understanding the Current State: Data was analyzed from HMIS, the PIT, and the Census
to identify:
• Differences between who experiences homelessness in the community and who
accesses the Homelessness Response System.
• Disparities in who becomes newly homeless.
• Differences in other demographics (e.g., gender, sexual orientation, age group) and
household characteristics across racial groups.
2. Barriers and Access to Services: Analyzed data from HMIS and prevention program
records to identify patterns and differences in:
• Access to crisis response services (shelter, street outreach, Housing Problem Solving).
• Enrollment in permanent housing interventions (Rapid Rehousing and Permanent
Supportive Housing).
• Access to homelessness prevention programs.
3. Service Delivery and Outcomes: Analyzed data from HMIS to identify patterns and
differences in:
• Rates of permanent housing placements by program type.
3
• Exits to permanent housing.
• Returns to homelessness within 24 months of a program exit.
Focus Strategies conducted statistical significance testing where feasible and appropriate. The
results were interpreted in close collaboration with the REA TWG to determine which
disparities were meaningful and actionable given system context and community insights.
C. Technical Working Group Process
The REA TWG consisted of 10 members, including five individuals with lived experience of
homelessness, along with service providers and city and county partners. The group met
monthly from February through August 2025.
The REA TWG contributed essential guidance throughout the process by helping shape the
analytic scope to ensure alignment with community priorities. Members drew on lived
experience, operational knowledge, and local context to interpret quantitative results and
make sense of emerging patterns. They also played a key role in identifying potential factors
contributing to disparities both within and beyond the Homelessness Response System, and
in elevating the experiences of overrepresented racial and ethnic groups—particularly Black or
African American households—to ensure the findings accurately reflected population realities.
The REA TWG’s input grounded the REA in community experience and strengthened the
accuracy, meaning, and usefulness of the findings. Interpretation of disparities also
considered structural factors such as housing availability, economic conditions, and
documented differences in access to public systems, which shape entry into and movement
through the homelessness response system.
D. Analytic Approach
The analytic approach for the REA was designed to provide a comprehensive and structured
assessment of racial and ethnic disparities across Alameda County’s homelessness response
system. The approach combined multiple data sources, a clear comparative framework, and
standardized quantitative methods to evaluate representation, access, flow through the
system, and outcomes for different household types. The subsections that follow describe the
data used, how the data was prepared for analysis, the analytical methods applied, the
comparative framework utilized, and the interpretation process.
4
Data Sources
The REA drew on multiple data sources to capture a comprehensive picture of household
interactions with Alameda County’s homelessness response system. The data sources are
summarized in the subsections below.
1. HMIS
The analysis relied primarily on de-identified client-level data from Alameda County’s HMIS.
HMIS data included enrollment and exit data covering three fiscal years from July 2021 to
June 2024 from the following project types:
• Coordinated Entry
• Emergency Shelter
• Safe Haven
• Street Outreach
• Supportive Services Only
• Transitional Housing
• Rapid Rehousing
• Permanent Supportive Housing
• Other Permanent Housing
2. PIT
H&H provided Focus Strategies with row-level de-identified 2024 PIT Count data. The row-
level format allowed for intersectional analysis between race and ethnicity and other
demographic and household characteristic variables. This data served as a reference point for
the demographics of people experiencing unsheltered homelessness, including those who do
not access services within the homelessness response system.
3. Provider-Level Prevention Data Tables
Data for homelessness prevention programs were not available in HMIS. Instead, prevention
providers submitted aggregate counts of the demographics of households they served to
Focus Strategies using a standardized Excel template.
4. Census
5
Focus Strategies downloaded 2020 Decennial Census microdata tables for Alameda County
from the Integrated Public Use Microdata Series (IPUMS).1 The data included household type
(e.g., family with children vs. household without children), sex, age, and race and ethnicity
indicators. This data was used to produce County-wide general population statistics to serve
as a comparison group and reference point for understanding who experiences homelessness
in the community.
Data Analysis
To support accurate comparisons and meaningful interpretation of results, all data was
cleaned and converted into a consistent format. This allowed all data to be analyzed similarly
and for results to be comparable across the various data sources. Focus Strategies defined
and re-coded data elements for the REA using the analytical software R; however, all methods
utilized in this analysis can be replicated using any tool that allows analysts to clean data,
apply consistent coding rules, and conduct proportion or rate comparisons.
Focus Strategies applied two complementary methods to assess racial and ethnic equity: (1) a
comparison of proportions used to identify differences in racial demographics across different
datasets or groups and (2) a rate comparison to identify differences in access or outcomes
from a single data set. Together, these analyses provided robust results for identifying
disparities. The subsections below summarize the methodology used for each type of
comparison.
1. Proportion Comparisons (Across Datasets)
This approach was used to evaluate representation and disproportionality and examined:
• How the racial/ethnic makeup of a system or program population (e.g., CES
enrollments) differs from a comparison group (e.g., unsheltered PIT Count, Alameda
County general population).
1 IPUMS USA. U.S. Census data for social, economic and health research. University of Minnesota. Retrieved from
https://usa.ipums.org/usa/
6
• Whether groups were overrepresented, proportionately represented, or
underrepresented across system access points (e.g., Coordinated Entry, housing
problem-solving street outreach, emergency shelter) and program outcomes.
To measure this, a disproportionality index was calculated by dividing the proportion of each
racial or ethnic group within the primary population by the proportion of each within the
comparison population. For example, if Black families comprised 40% of families experiencing
unsheltered homelessness and 20% of families enrolling in Coordinated Entry, the
disproportionality index would be 0.5 (0.2/0.4 = 0.5), showing that black families experiencing
homelessness were half as likely to access Coordinated Entry. A disproportionality index of 1.0
indicates proportional representation, values over 1.0 indicate overrepresentation, and values
less than 1.0 indicate underrepresentation.
The analysis used two-sample tests of proportions (z-tests) to test whether the observed
differences between proportions were statistically significant. The statistical threshold used
was p < 0.01, meaning there is less than a 1% likelihood that a difference in the results is due
to random variation.
To explore how overlapping identities influence outcomes, the analysis incorporated
intersectional comparisons (e.g., race and gender, race and veteran status). Proportionality
indices and z-tests were also conducted for intersections of race and other key variables (e.g.,
race and gender, or race and veteran status). The results of these analyses helped identify
disparities within subgroups whose experience may not be reflected in the results for the
racial or ethnic group overall. Findings from the intersectional analyses were reviewed
alongside the overall results for each racial or ethnic group to allow for deeper understanding
of how households experience the local homelessness response system.
2. Rate Comparisons (Within a Single Dataset)
This analysis was used to assess equity in outcomes within a single group and compared:
• Rates of exits to permanent housing, returns to homelessness, and other key outcomes
across demographic groups.
• Whether outcome gaps indicated meaningful disparities in how households
progressed through the system.
7
For example, if 250 Hispanic or Latine adult-only households exited from any program in HMIS
in the reporting period, and 50 of those exits were to homelessness, then the homelessness
exit rate was 0.2 (50/250 = 0.2), showing that Hispanic or Latine adult-only households exited
to homelessness at a 20% rate. This rate was then compared with other groups’ rates to
identify differences. Statistical testing was not applied because small sample sizes for groups
made results highly sensitive to small changes in counts. These results were instead
interpreted in the context with other findings and the input from the REA TWG to assess
whether observed differences were meaningful.
Comparative Framework
The analysis utilized a consistent set of comparisons to examine racial and ethnic equity across
Alameda County’s homelessness response system. The comparative framework was
developed based on the REA Analysis Plan and was adapted to reflect the key questions
identified with the REA Technical Working Group. This framework defines the comparison
group for each group of interest (i.e., who the demographics of the groups of interest will be
compared with to identify racial differences and disparities).
Given the goals of the REA, the analysis was grouped into four categories: (1) Overall
Homelessness, (2) System and Service Access, (3) Program Outcomes, and (4) Pathways into
Homelessness.
Interpretation of Results
The REA included a structured approach to interpreting results that combined statistical
evidence with practitioner knowledge and lived experience to determine which disparities
were meaningful and actionable. Focus Strategies first reviewed the results to identify which
were statistically significant and meaningful based on the goals of the REA TWG and Home
Together Task Force. Then, the results were presented and examined in partnership with the
REA TWG and H&H to figure out what they mean for how the homelessness system actually
operates. This process ensured that findings were not only technically sound but also
grounded in the realities of service delivery, population needs, and system equity goals.
E. Summary of Major Findings
8
Across Alameda County’s Homelessness Response System, the REA reveals a complex picture
in which certain areas show proportionate access and outcomes across racial and ethnic
groups, while other areas show persistent and significant inequities.
Several groups—including Black, Hispanic/Latine, Native Hawaiian and Other Pacific Islander,
and American Indian/Alaska Native households—face barriers at key points in the
homelessness response system and experience disproportionately high rates of
homelessness, service gaps, or long‑term housing instability. Meanwhile, white households
generally experience fewer barriers, higher service uptake, and more favorable outcomes.
The findings below summarize disparities identified across the system, as well as areas where
disparities were investigated but not found.
1. Understanding the Current State
• Black adult-only and family households are highly overrepresented among people
experiencing homelessness relative to their representation in the county population.
• Black, indigenous, and people of color (BIPOC) adult-only households are more likely
than white households to become newly homeless, demonstrating persistent
disparities in housing stability and access. This finding is similar to a key finding of the
2021 Centering Racial Equity in Homeless System Design report, signaling that barriers
to housing stability for BIPOC households are ongoing.
• Native Hawaiian and Other Pacific Islander, Hispanic/Latine adult-only households, and
Black family households report the highest rates of fleeing domestic violence, a
significant pathway into homelessness.
• Asian families remain homeless after leaving programs at a higher rate than other
families.
• Among adult-only households, rates of remaining homeless after leaving programs are
proportionate across racial groups.
2. Barriers and Access to Services
• Overall enrollment in programs in the system is disproportionately low for Asian and
Hispanic/Latine adult-only households.
• Enrollment in system programs overall is proportionate across racial and ethnic groups
for family households.
9
• Asian family households enrolled in shelter at a lower rate than families of other racial
and ethnic groups.
• Black family households enrolled in street outreach at a lower rate than families of
other racial and ethnic groups.
• Enrollment in shelter and street outreach is proportionate across racial and ethnic
groups for adult-only households.
• Enrollment in Coordinated Entry and Housing Problem Solving is proportionate across
racial and ethnic groups for both adult-only and family households.
• Black adult-only households are less likely to be eligible for Permanent Supportive
Housing (PSH) and make up a smaller share of those enrolled in PSH than those
enrolled in Rapid Rehousing (RRH). Black family households also make up a smaller
share of those enrolled in PSH than those enrolled in RRH.
• Hispanic/Latine adult-only and family households are less likely to enroll in both RRH
and PSH, despite being equally likely to meet eligibility criteria.
• Prevention programs effectively reach Black households, though overall availability of
prevention funds is limited, constraining potential impact.
3. Service Delivery and Outcomes
• Black family households, Hispanic/Latine adult-only households, and American
Indian/Alaska Native adult-only households experience higher rates of returns to
homelessness within 24 months of exiting the system to permanent housing. This
finding is similar to a key finding of the 2021 Centering Racial Equity in Homeless
System Design report, signaling that BIPOC families face continued challenges in
maintaining a stable living situation once they exit the Homelessness Response System.
• Hispanic/Latine adult-only and family households are less likely to exit shelter to
permanent housing, indicating barriers in shelter-based rehousing pathways.
• Street Outreach exits to permanent housing are proportionate across racial and ethnic
groups for both adult-only and family households.
• PSH retention is proportionate across racial and ethnic groups for both adult-only and
family households, suggesting that once housed with supports, outcomes stabilize
across demographics.
The 2025 REA identified continued disparities in several areas of the Homelessness Response
System, while also identifying areas in which access and outcomes were proportionate across
10
racial and ethnic groups. Findings from the REA was used by both the REA TWG and by the
Home Together 2030 Task Force to inform the Home Together 2030 Plan’s priorities,
measures of progress, and use of a targeted universalism framework: establishing shared
goals for all people experiencing or at risk of homelessness while recognizing that different
populations may face different barriers to achieving those goals. Continuing to examine
disaggregated data over time will help the County identify where disparities persist, whether
new disparities emerge, and inform strategies to improve access and outcomes across
populations.
F. Limitations
The findings of this analysis should be interpreted within the context of limitations related to
data quality, reporting practices, and the inherent constraints of statistical methods. These
limitations affect the level of precision possible in identifying disparities, the comparability of
results across datasets, and the degree to which quantitative findings alone can explain the
root causes of inequities within the homelessness response system. The subsections below
outline the key considerations that informed how results were interpreted and contextualized.
1. Data Completeness and Quality
Data completeness in HMIS varied across programs and providers. Some clients had missing
or incomplete demographic fields, which may have led to the underrepresentation of certain
groups in the analysis. Small sample sizes for specific racial or ethnic subgroups also limited
the ability to conduct statistical testing for some analysis, even when disparities appeared
large. Focus Strategies reviewed all output to contextualize the results based on sample size.
In instances in which data was missing or incomplete at a high rate for specific fields or
datasets, comparison results were excluded from consideration as key findings due to a lack
of reliability.
2. Prevention Data Reporting
Homelessness prevention program data were self-reported in an aggregate format by
providers using a standardized Excel template. As a result, Focus Strategies could not verify if
there were differences across providers in how demographic fields were categorized or
reported. Additionally, the aggregate data limited the variations of intersectional analysis that
11
could be conducted with prevention data. Because the prevention data was not directly
comparable with HMIS data, Focus Strategies limited the comparisons to race and ethnicity,
not disaggregated further by any other demographics or household characteristics.
3. Differences Across Data Sources
The U.S. Decennial Census and the 2024 PIT Count were collected using different
methodologies and population definitions than HMIS. These structural differences limited
direct comparisons when evaluating proportionality or representation across datasets. In
order to account for these differences, Focus Strategies recoded demographic and household
type variables from these data sources to align with the definitions used in the homelessness
response system, where possible.
4. Interpretation of Statistical Results
Statistical tests identified where differences between groups are unlikely to have occurred by
chance, but they do not explain why disparities exist. As emphasized by the REA TWG,
quantitative findings must be interpreted alongside qualitative insights such as operational
context and the perspectives of people with lived experience. Integrating these perspectives
helped ensure that the REA produced nuanced and actionable findings about racial equity
across Alameda County’s homelessness response system. As described above, Focus
Strategies engaged with the REA TWG to identify areas in which more qualitative information
would be useful to contextualize REA findings.
5. Rapidly Changing Landscape
The rigorous statistical methods utilized in this analysis required a large enough volume of
records to generate stable and interpretable results. Sample size was an especially limiting
factor when examining smaller demographic groups such as American Indian or Alaska Native
households or when conducting intersectional analyses. Focus Strategies utilized three years
of HMIS data for the REA, which provided adequate sample size for statistical testing.
However, the data from the early years included in the analysis may not reflect current system
practices, program designs, and population patterns, especially given that Alameda County’s
homelessness response system evolves rapidly in response to changes in funding, best
practices research, and the changing needs of people experiencing homelessness.
Appendix C. Home Together 2030 Permanent Housing Program Models
This appendix describes the primary permanent housing models used in Home Together 2030 and the distinct needs each is intended to address. Together, the models provide a range of housing assistance
and service intensity to help households exit homelessness and remain stably housed. The target populations described here are intended to guide system planning, investment decisions, and matching of
households to the types of resources most likely to meet their needs. They do not replace or redefine program eligibility requirements, which continue to be governed by specific funding sources and
Coordinated Entry policies.
Program Type Program Description Target Populationi Assistance/Essential Program Elements Core Performance
Measuresii Notes/Comments
Rapid Re-
Housing
Time-limited case
management and rental
assistance intended to
help households exit
homelessness and
stabilize in housing as
quickly as possible.
Individuals and families
experiencing
homelessness who have
income, employment
prospects, benefits, or
other anticipated
resources that make it
reasonable to expect the
household can maintain
housing independently by
the end of the available
period of assistance.
Rental Assistance
• Up to 24 months of tenant-based rental assistance, depending on
funding source and need. Under some funding sources, waivers
may be granted to extend time for households nearing the ability
to fully resume responsibility for rent payments.
o Under the RRH model, providers have flexibility to design
different tenant rent contribution methodologies as long as
methodologies: 1) include assessment of household
needs/circumstances, 2) are applied consistently/fairly across
households, and 3) comply with the CoC’s written standards.
• Households may also receive assistance for other housing-related
expenses, including security deposits, utility deposits, utility
assistance, rental arrears owed on past units, and
furniture/household supplies (again, depending on funding source
and need).
• Under RRH models, the household is leaseholder and may remain
in the unit permanently following exit from RRH (per the terms of
their lease agreement).
Case Management/Supportive Services
• Case management services will be provided throughout the
household’s tenure in the program (and may be continued for a
limited period afterwards, typically up to 6 months, depending on
the funding source and need).
• A wide variety of services are eligible under different funding
sources, but provision of employment/workforce assistance (either
Average Length of Stay
Percent of households
exiting to Permanent
Housing
Percent of households who
avoid returns to
homelessness at 12, 18, and
24 months
Percent of households that
increase or maintain income
through earned income,
benefits, or other sources.
• There are many different funding
sources (Federal, State, and local) for
RRH programs in Alameda County.
• Each may target different
subpopulations (e.g., families,
individuals, veterans, DV survivors).
• Each may have variations in rules (e.g.,
caps on assistance, eligible activities,
eligible expenses).
• However, key to the model is that
both funding and services are time-
limited, and accordingly, households
expected to have ongoing need for
support to remain stably housed are
not an appropriate match.
Program Type Program Description Target Populationi Assistance/Essential Program Elements Core Performance
Measuresii Notes/Comments
directly or through a partner) is essential.
• Other important stabilization supports include (but are not limited
to): budgeting/financial literacy, credit counseling, landlord
mediation support, and connection to mainstream benefits and
services.
Bridge
Subsidies
Time-limited case
management and rental
assistance intended to
help households exit
homelessness as quickly
as possible while awaiting
an ongoing housing
resource that has been
identified but is not yet
available.
Individuals and families
experiencing
homelessness who:
1) Are assessed to need
deeper/
ongoing assistance to
remain stably housed;
and
2) For which a specific,
ongoing housing
resource has been
identified for the
household and is
expected to become
available within 6
months.
Rental Assistance
• Up to 6 months of tenant-based rental assistance, depending on
the funding source and need (extensions may be allowed as
needed for a specific resource that is delayed).
o Under the Bridge Subsidy model, rent will be calculated using
the same methodology as voucher programs (with households
paying 30% of income towards rent) to facilitate a smooth
transition between programs.
• Households may also receive financial assistance for other
housing-related expenses, such as security deposits, utility
deposits, utility assistance, and rental arrears owed on past units
(again, depending on funding source and need).
• Under the Bridge model, the household is leaseholder.
o If the household is transitioning to a tenant-based voucher,
care should be used to ensure the unit meets the applicable
housing inspection standards and that the landlord will accept
the future funding source to allow the client to transition in
place.
o If the household is transitioning to a project-based unit,
thought should be given to the term of the lease to ensure the
household does not have to break their lease agreement when
it is time to transition.
Case Management/Supportive Services
• Case management services will be provided throughout the
Average Length of Stay
Percent of households
exiting to Permanent
Housing
Percent of households who
avoid returns to
homelessness at 12, 18, and
24 months
Percent of households that
increase or maintain income
through earned income,
benefits, or other sources.
While participants in all program models
may transition to other program types as
needs emerge/change, bridge housing is
intended for specific circumstances where
the needed resource is not available and
there is time sensitivity involved (e.g., a
program closure due to loss of funding; a
facility problem requiring emergency
transfers; an emergency encampment
closure).
Program Type Program Description Target Populationi Assistance/Essential Program Elements Core Performance
Measuresii Notes/Comments
participant’s tenure in the program.
• While services funded under bridge housing are time limited, they
should focus on helping participants stabilize in housing, address
healthcare needs, and prepare for transition to the permanent
resource. Ideally, the case manager would continue working with
the participant through the transition. If that is not possible, the
bridge provider should ensure a coordinated handoff to the
provider responsible for ongoing services.
Shallow
Subsidies
Financial assistance
designed to help low-
income households
remain stably housed by
providing assistance that
bridges the difference
between monthly income
and rent costs.
Within Home Together
2030, shallow subsidies
are primarily intended to
prevent returns to
homelessness among
households already
housed through the
homeless response
system who need
additional affordability
support to remain stable.
Shallow subsidies may
also be used as a
Individual and families
who are in housing
(leaseholders) but are at
imminent risk of
homelessness. Target
households:
• Have low incomes, but
are not extremely low-
income (e.g., earning
between 30%–60% of
Area Median Income).
• Struggle with housing
affordability but can
generally maintain
housing stability with
limited assistance.
• Don’t require intensive
support services.
• May be transitioning
from a supportive
Rental Assistance
• Shallow subsidy programs may be structured in different ways but
are designed to help households bridge the gap between their
earnings and their rent. The program may provide a fixed dollar
amount, or assistance may be tiered based on income. The length
of the assistance may also vary depending on funding source,
though assistance will be medium to long-term.
• Households must have a lease in their own name to be eligible for
assistance. Depending on program design, assistance may be paid
to landlords or directly to program participants.
Case Management/Supportive Services
• Shallow subsidies programs generally do not include support
services, though some models may include budgeting, financial
literacy, credit counseling, and other types of financial services.
Percent of households who
avoid returns to
homelessness at 12, 18, and
24 months
Percent of households that
increase or maintain income
through earned income,
benefits, or other sources.
Shallow subsidies should not be treated as
an automatic next step for households
exiting RRH. RRH should continue to be
targeted to households reasonably
expected to maintain housing when time-
limited assistance ends. Referral to a
shallow subsidy should be based on
demonstrated need to prevent an
immediate return to homelessness.
Because shallow subsidies may be used
both to prevent first-time homelessness
and to prevent returns to homelessness,
providers serving both populations should
enroll clients in separate programs within
HMIS to allow for more accurate outcome
tracking.
Program Type Program Description Target Populationi Assistance/Essential Program Elements Core Performance
Measuresii Notes/Comments
prevention intervention
for households that are
currently housed but at
imminent risk of
homelessness.
housing program and
need a “step-down” to
remain stably housed.
Permanent
Supportive
Housing
(PSH)
Permanent rental subsidy
and wrap-around support
services that help
individuals and families
with intensive supportive
service needs to obtain
and maintain housing
stability.
Individuals and families
experiencing chronic
homelessness. This
means the household
meets the following
conditions:
• Head of household or
child in family
household has a
disabling condition;
and
• Continuously
homeless for one year
or more, or four
episodes of
homelessness within
the past three years.
Rental Assistance
• Rental assistance may be provided via the following models:
o Tenant-based (voucher is assigned to the household and moves
with the household);
o Project-based (voucher is assigned to a building/property and
stays with the building as households move in/out); or
o Sponsor-based (voucher is assigned to the provider and stays
with the provider as households enter/exit their program).
• With regard to location and concentration of units, assistance may
be:
o Scattered-site: Households served by the program have leases or
occupancy agreements in units dispersed across multiple
properties or locations in the community, rather than
concentrated in a single building or site.
o Site-based (single site): Households served by the program are
housed in a single building.
o Site-based (clustered/multiple sites): Households served by the
program are housed across multiple buildings or locations, but
each facility houses more than one program household.
• Regardless of the model, each household has a lease or sublease
agreement in their name, and each household contributes 30% of
Percent of households who
avoid returns to
homelessness at 12, 18, and
24 months
Percent of households that
increase or maintain income
through earned income,
benefits, or other sources.
Different PSH models help the CoC meet
different needs. For example, tenant-
based/scattered-site approaches can be
scaled more quickly and allow for greater
housing choice, while site-based
approaches allow for more intensive, 24/7
supports. Ensuring the CoC has a range of
models is important to meeting different
needs.
Given that there are many more people
eligible for PSH than there are available
units, the CoC will review/update
prioritization criteria on a regular basis to
ensure resources are being used to help
meet CoC objectives.
Because needs change over time, a key
aspect of implementation will be
designing a flexible service approach that
allows participants to step up to PSH+ or
Program Type Program Description Target Populationi Assistance/Essential Program Elements Core Performance
Measuresii Notes/Comments
their income toward rent, with the program covering the
remainder.
Case Management/Supportive Services
• Services should be flexible and participant-centered to better meet
participant needs. Primary focus of services is on tenancy supports
that help people access and remain in housing.
• An additional focus of services is to connect participants to, or
directly provide, supportive services, including mental health
services, substance use disorder services, physical health services,
benefits assistance, employment assistance, etc.
• Housing is dependent only on the participant’s compliance with
their lease agreement – not participation in case management or
treatment services.
• If a participant declines services, case managers must continue
efforts to engage the participant and provide supports needed to
help the household maintain housing stability.
step down to DAH as appropriate, without
unnecessarily requiring them to lose their
housing or start over in a new program.
PSH for
Medically
Frail (PSH+)
Permanent rental subsidy
and intensive case
management supports
and healthcare services
for individuals
experiencing
homelessness with
complex medical needs
(often a combination of
behavioral health
conditions, chronic health
conditions, and geriatric-
Individuals experiencing
homelessness who:
1) Are frequent utilizers
of emergency room
services and/or have
unmet need for
healthcare services;
2) Have functional
limitations/inability to
complete ADLs;
3) Have complex medical
needs/chronic
conditions (i.e.,
advanced disease that
Rental Assistance
• Rental assistance may be tenant-based, project-based, or sponsor-
based (see definitions in PSH line above), though to facilitate the
provision of more intensive services, project-based settings (single-
site or cluster) are optimal.
• Regardless of the model, each participant has a lease or sublease
agreement in their name, and each participant contributes 30% of
their income toward rent, with the program covering the remainder.
• If acuity level decreases for a participant that is in a site-based
program, they will keep their lease/unit, but nursing and caregiver
services will end. They may be offered the option to transfer to
another subsidized permanent housing option.
Percent of households who
avoid returns to
homelessness at 12, 18, and
24 months
Percent of households that
increase or maintain income
through earned income,
benefits, or other sources.
Program Type Program Description Target Populationi Assistance/Essential Program Elements Core Performance
Measuresii Notes/Comments
related conditions).
would worsen
significantly without
ongoing treatment
and attention); and
4) Are elderly, frail and in
danger of being
unable to live
independently
without a higher level
of support.
Case Management/Supportive Services
• PSH+ offers the highest level of service supports of all program
models, including intensive case management, onsite health/skilled
nursing supports, behavioral health supports, and support with
ADLs.
Dedicated
Affordable
Housing
(DAH)
Deeply affordable
housing
with light-touch or no
ongoing supportive
services. Units or
subsidies are dedicated
for people exiting the
homeless response
system (coordinated
entry referrals).
Individuals and families
experiencing
homelessness who:
1) Are extremely low
income;
2) Have long or multiple
episodes of
homelessness;
3) Have significant
barriers to increasing
income; and
4) Do not require
supportive services to
remain stably housed.
Rental Assistance
• Rental assistance may be tenant-based/scattered-site or project-
based/site-based (see PSH definitions above), though tenant-based
assistance can provide greater flexibility and participant choice.
• Each household has a lease agreement in their name, and each
household contributes 30% of their income toward rent, with the
program covering the remainder.
Case Management/Supportive Services
• DAH programs will generally not include support services, though
some models may include transitional/light-touch support to
facilitate a household’s exit from homelessness and lease-up in
their new unit, as well as connection to mainstream supports and
support around annual recertification periods.
Percent of households who
avoid returns to
homelessness at 12, 18, and
24 months
Percent of households that
increase or maintain income
through earned income,
benefits, or other sources.
i Program eligibility is established by each funding source and typically allows a different (often broader) population to be served. In contrast, defining a target population is the County’s attempt to be more strategic about how
different program models are used in the system to meet different needs. The target populations outlined in this document were used to help inform system modeling assumptions about how much of a given program type is
needed in Alameda County, and they will also help guide CES referrals during Home Together 2030 implementation. That said, identification of target populations is not a redefinition of eligible population. Eligible households that
fall outside of the described target population but are still determined to be a good fit for a particular program may be referred to and served by the program type in accordance with CES protocols.
ii Core performance measures are listed to provide a sense of the most important dimensions of performance of a given model. Additional measures may be established per a program’s funding source.