Report to the Legislature — Permanent Supportive Housing Need Forecast (December 1, 2023)
Summary
This December 1, 2023 report to the Legislature, the Permanent Supportive Housing Need Forecast, was produced by the Department of Social and Health Services' Research and Data Analysis Division in collaboration with the Department of Commerce. It responds to engrossed Substitute Senate Bill 5187 (Chapter 475, Laws of 2023), which directed the division to prepare an annual report on the projected demand for permanent supportive housing. The report states that the number of individuals potentially eligible for PSH increased 6 percent from 70,373 in January 2018 to 74,697 in December 2022. If recently observed trends continue, the projected number is expected to be 82,238 by December 2025.
Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used
Cite this document
“Report to the Legislature — Permanent Supportive Housing Need Forecast (December 1, 2023).”
Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/2023-permanent-supportive-housing-need-forecast-final-f8d8ca57-2302-41-9a715ecd42/
, SHA-256 e4c3d18c8f07b894c3afa984a62b2882736d078f4c6b27e838efc0cc863338e1.Full text
REPORT TO THE LEGISLATURE
Permanent Supportive Housing Need Forecast
This report was produced in collaboration with the Department of Commerce and funded by the
Washington State Legislature under engrossed Substitute Senate Bill passed by the 67th Washington
State Legislature on April 23, 2023
December 1, 2023
Facilities, Finance, and Analytics Administration
Research and Data Analysis Division
PO Box 45204
Olympia, WA 98504-5204
(360) 902-0707
http://www.dshs.wa.gov/rda
CONTENTS
PAGE
SCOPE AND PURPOSE ............................................................................................................ 1
KEY FINDINGS .......................................................................................................................... 1
FIGURE 1. Forecasted Permanent Supportive Housing Need, January 2023 to December
2025 .................................................................................................................................................... 1
PERMANENT SUPPORTIVE HOUSING FORECAST ................................................................ 2
FIGURE 2. Historical and Forecasted Need for Permanent Supportive Housing January
2018 to December 2025 ..................................................................................................................... 2
TABLE 1. Forecasted Need for Permanent Supportive Housing Detail, January 2028 to
December 2025 .................................................................................................................................. 3
PERMANENT SUPPORTIVE HOUSING CASELOAD ............................................................... 4
FIGURE 3. Permanent Supportive Housing Caseload January 2028 to December 2022 ................ 4
LIMITATIONS AND ADDITIONAL CONSIDERATIONS ............................................................. 4
TECHNICAL NOTES .................................................................................................................. 6
APPENDIX – Changes from Previous Forecast ...................................................................... 7
FIGURE 4. Comparison of Current Forecast with Previous Forecast................................................ 7
REFERENCES ........................................................................................................................... 7
Taylor Danielson, PhD
Anita Rocha, PhD
David Mancuso, PhD
RDA Report Number 11.268
https://www.dshs.wa.gov/rda
Permanent Supportive Housing Need Forecast
December 1, 2023 0
Scope and Purpose
Washington’s engrossed Substitute Senate Bill 5187 (Chapter 475, Laws of 2023, §1208(9))
directed the Department of Social and Health Services’ Research and Data Analysis Division
(RDA) to “prepare an annual report in consultation with the Department of Commerce on the
projected demand for permanent supportive housing.” The following brief summarizes the
results of the assigned analysis. Estimates of permanent supportive housing (PSH) need
were calculated using data from RDA’s Integrated Client Databases, which integrate
administrative data from the state’s Medicaid claims database, public assistance eligibility
determination system, and Homelessness Management Information System, among others
(Mancuso & Huber, 2021). We constructed unduplicated monthly counts of adults ages 18
and above with state or federally funded medical coverage who were potentially eligible for
PSH from January 2018 to December 2022. These counts were then used to forecast the
number of individuals potentially in need of PSH over a 36-month forecast period ending
December 2025.
Key Findings
Based on a time-series analysis of administrative data used to identify individuals who fit the
profile of a PSH client, we found the following:
• The number of individuals potentially eligible for PSH increased 6 percent from 70,373
in January 2018 to 74,697 in December 2022.
• If recently observed trends continue, the projected number of individuals potentially
eligible for PSH is expected to be 82,238 by December 2025. This is nearly equivalent
to prior forecasts despite changes to inclusion criteria that increased the number of
PSH-eligible individuals. These similarities are attributable to a slower projected rate of
growth for this population.
The 82,238 projection is subject to uncertainty; with 95 percent confidence, the projected
number of PSH-eligible individuals will be between 63,577 and 100,900 in December 2025.
FIGURE 1.
Forecasted Permanent Supportive Housing Need
January 2023 to December 2025
120,000
100,900
100,000
Upper confidence interval
82,238
80,000 74,494
60,000 Lower confidence interval 63,577
40,000
20,000
0
36-Month Forecast
J F M A M J J A S O N D J F M A M J J A S O N D J F M A M J J A S O N D
2023 2024 2025
Permanent Supportive Housing Need Forecast
December 1, 2023 1
Permanent Supportive Housing Forecast
We used data from the DSHS-RDA Integrated Client Databases (ICDB) to identify
individuals who appear likely to need PSH.1 Individuals were counted towards each monthly
observation if they:
1. Were observed to be homeless in the month; AND
2. Were at least 18 years of age in that month; AND
3. Had at least one month of state or federally funded medical coverage2 in the past year
(inclusive of the month in question); AND
4. Had an indication of a mental health or substance use disorder in the past two years
(inclusive of the month in question) OR an indication of a significant chronic illness
based on Medicaid claims data.3
The resulting dataset was used to measure the total number of individuals potentially in need
of PSH in each month from January 2018 to December 2022. The statistical model chosen
to forecast PSH need is a variant of exponential smoothing. 4 As shown in Figure 2, the
results of these analyses indicate that the total number of PSH-eligible clients is expected to
increase from 74,494 in January 2023 to 82,238 in December 2025, which corresponds to a
10 percent increase over a three-year period. The forecasted number of persons needing
PSH is anticipated to increase by 221individuals per month, on average, through December
2025.
FIGURE 2.
Historical and Forecasted Need for Permanent Supportive Housing
January 2018 to December 2025
120,000
100,000
100,900
Upper confidence interval
82,238
80,000 74,494
70,373
Lower confidence interval
60,000 63,577
40,000
20,000
Estimated Need (Actual) 36-Month Forecast
0
J F MA M J J A S O N D J F MA M J J A S O N D J F MA M J J A S O N D J F MAM J J A S O N D J F MA M J J A S O N D J F M AM J J A S O N D J F M AM J J A S O N D J F M AM J J A S O N D
2018 2019 2020 2021 2022 2023 2024 2025
1 The ICDB integrates administrative data from several state data systems including the state’s Automated Client
Eligibility System (ACES), ProviderOne MMIS data system that contains Medicaid claims and encounter data, and
the Homeless Management Information System (Mancuso & Huber, 2021).
2 Including individuals dually enrolled in Medicaid and Medicare.
3 Chronic illness is identified using a claims-based medical risk score that measures an individual’s anticipated
medical costs relative to the average disabled Medicaid client. Individuals were considered to have a chronic
illness if that score is greater than or equal to 1.
4 All forecasts were calculated using PROC ESM in SAS® 9.4. We compared several ESM approaches to identify a
method that best fit the historical data. Based on these comparisons, we selected the Holt-Winters’ additive
method. This method prioritizes more recent data over earlier segments of the time series, assumes that there is
an underlying linear trend in the data, and adjusts the forecasts for 12-month seasonality.
Permanent Supportive Housing Need Forecast
December 1, 2023 2
To acknowledge the inherent uncertainty of any forecast, we present this forecast with a 95
percent confidence interval, which ranges from a low of 63,577 to a high of 100,900 by
December 2025. The monthly forecasted counts generated using this approach — and the
upper and lower confidence limits — are provided in Table 1 below.
TABLE 1.
Forecasted Need for Permanent Supportive Housing Detail
January 2018 to December 2025
Lower 95% Upper 95%
Date Confidence Limit Forecasted PSH Need Confidence Limit
2023 JAN 73,641 74,494 75,347
FEB 73,137 74,363 75,589
MAR 72,747 74,329 75,911
APR 72,329 74,267 76,204
MAY 72,379 74,678 76,978
JUN 72,351 75,021 77,692
JUL 72,527 75,578 78,630
AUG 72,888 76,331 79,775
SEP 72,764 76,611 80,458
OCT 72,926 77,187 81,449
NOV 72,611 77,299 81,988
DEC 72,085 77,211 82,337
2024 JAN 71,397 77,008 82,619
FEB 70,808 76,877 82,945
MAR 70,306 76,842 83,379
APR 69,764 76,780 83,797
MAY 69,685 77,192 84,699
JUN 69,528 77,535 85,543
JUL 69,573 78,092 86,611
AUG 69,805 78,845 87,885
SEP 69,553 79,125 88,696
OCT 69,588 79,701 89,814
NOV 69,149 79,813 90,477
DEC 68,500 79,725 90,949
2025 JAN 67,699 79,522 91,344
FEB 66,990 79,391 91,791
MAR 66,369 79,356 92,344
APR 65,711 79,294 92,878
MAY 65,518 79,706 93,894
JUN 65,247 80,049 94,851
JUL 65,182 80,606 96,030
AUG 65,304 81,359 97,414
SEP 64,944 81,638 98,333
OCT 64,873 82,215 99,557
NOV 64,330 82,327 100,325
DEC 63,577 82,238 100,900
Permanent Supportive Housing Need Forecast
December 1, 2023 3
Permanent Supportive Housing Caseload
While the preceding analysis describes the potential demand for PSH, Figure 3 additionally
describes the number of individuals 18 years of age and above who actually received PSH in
Washington State.5 According to data entered into each county’s Homeless Management
Information System (HMIS), the number of individuals receiving PSH has grown from 7,518
in January 2018 to 8,548 in December 2022, a 14 percent increase overall, or on average, a
2.7 percent increase per year. PSH caseloads represented an average of about 11 percent
of the potentially eligible individuals in any given month.
FIGURE 3.
Permanent Supportive Housing Caseload versus Estimated Need
January 2018 to December 2022
120,000
100,900
100,000
Upper confidence interval
Monthly potentially PSH-eligible individuals 82,238
80,000 5-year increase = 6% overall (1.2% per year average) 74,494
70,373
Lower confidence interval
60,000 63,577
40,000
Monthly PSH caseload represents about 11%
20,000 of potentially eligible individuals
5-year increase = 14% overall (2.7% per year average)
7,518 8,548
36-Month Forecast
0
J F MAM J J A S O N D J F MAM J J A S O N D J F MA M J J A S O N D J F MAM J J A S O N D J F MA M J J A S O N D J F M AM J J A S O N D J F M AM J J A S O N D J F M AM J J A S O N D
2018 2019 2020 2021 2022 2023 2024 2025
Limitations and Additional Considerations
This forecast is subject to several limitations. First, because this is an update of the
estimates provided earlier in 2023, this forecast relies solely on recent historical trends and
does not incorporate information on external factors that may influence demand for PSH.
Second, this forecast is based in part on data generated during the COVID-19 pandemic. It
is unclear to what degree the pandemic affected the number of individuals experiencing
homelessness or the service utilization data used to identify those in need of PSH. Because
state data systems are intended to support agencies’ day-to-day business operations, they
only include information on individuals who came in contact with state-funded health and
human services systems. If an individual did not interact with these systems, we were not
able to determine if they were potentially eligible for PSH. This was particularly problematic
during the beginning of the COVID pandemic, as many key access points for public
assistance and housing services (e.g., community service offices and emergency shelters)
were closed. Additionally, recent analyses from RDA suggest that homeless individuals may
have faced greater barriers than others to establishing or maintaining connections to public
5 PSH recipients are restricted to individuals enrolled in PSH programs in the state’s HMIS system and excludes
those individuals who: 1) received PSH services/housing that were funded solely through local or philanthropic
dollars; 2) were housed in units operated by, or that received vouchers from local housing authorities; or 3) were
housed in a site-based PSH project funded through the Housing Trust Fund and administered by the Department
of Commerce’s Multi-Family Housing Unit.
Permanent Supportive Housing Need Forecast
December 1, 2023 4
assistance during the COVID pandemic (Patton et al. 2023). Consequently, we may have
underestimated the number of clients who needed PSH and thus underestimated the
forecasted demand.
To explore the sensitivity of our forecast to the use of data collected during the pandemic, we
also estimated an exponential smoothing model applied to data from January 2018 through
December 2019. This model, based on data from the years immediately prior to the
pandemic, produced estimates almost 7,700 lower (approximately 74,538 individuals in need
of PSH services) for December 2025 than those estimated using data spanning the full five
years from January 2018 through December 2022. This difference may be due in part to the
return of DSHS Community Service Offices to full in-person services on February 28th,
20226, which improved both client access to services and our ability to identify PSH-eligible
clients as services returned to normal. Estimates suggest that from March through
September 2022 there was a rapid increase in potentially PSH-eligible individuals compared
to October through December 20227. The surge in the number of individuals identified as
potentially PSH-eligible may have influenced the forecasted growth rate for those estimates
produced using all five years of data since the Holt-Winters’ additive method places heavier
weights on more recent observations. Still, the sensitivity forecast is within the bounds of the
95 percent confidence limits of the main forecast.
Third, as noted earlier, the unit of analysis used in these forecasts is the number of
individuals 18 years and older potentially eligible for PSH in a month. However, PSH
service units are allocated at the household level. Other housing projections (e.g., those
previously produced under Engrossed Second Substitute House Bill 1220, Chapter 254,
Laws of 2021) center on the number of households that need PSH. The decision to focus
on individuals here is due to the difficulty in identifying and constructing households using
administrative data, where individuals are grouped together into household-like structures
based on program rules.
Fourth, our estimates rely primarily on Medicaid claims data to identify individuals with some
indication of a significant chronic illness or behavioral health disorder. Claims paid solely
through other funding mechanisms (e.g., third-party payers, Medicare) are not reflected in
these data at this time. Consequently, the forecasts presented here may systematically
undercount clients dually enrolled in Medicaid and Medicare and those with third-party
liability coverage, such as veterans, potentially eligible for PSH. We mitigate this risk through
the inclusion of individuals with a mental health or substance use disorder over the past two
years through Medicaid claims.
Finally, this forecast may not fully reflect the level of need for PSH in Washington. In contrast
to other states that limit eligibility for PSH to the chronically homeless, Washington has
adopted more expansive eligibility criteria. Specifically, individuals with a behavioral health
need or disabling condition who are currently housed but whose household earnings are less
than or equal to 80 percent of the Area Median Income (AMI) are eligible for PSH in
Washington State (Washington State Department of Commerce, 2023). Because these
forecasts focus on individuals who were homeless in a month, they omit housed individuals
potentially eligible for PSH.
6 DSHS Community Services Offices return to full in-person services: https://www.dshs.wa.gov/os/office-
communications/media-release/dshs-community-services-offices-return-full-person-services.
7 Specifically, the average increase in the number of potentially PSH-eligible individuals was 356 per month from
March to September 2022 compared to the average increase of 149 per month from October to December 2022.
Permanent Supportive Housing Need Forecast
December 1, 2023 5
TECHNICAL NOTES
Measures
Medical Coverage. Medicaid and other medical coverage information was obtained from
eligibility codes recorded in ProviderOne.
Behavioral Health Indicators. Data from the Health Care Authority’s (HCA’s) ProviderOne
Medicaid claims system and Behavioral Health Data System (BHDS), and Aging and Long-Term
Support Administration’s Comprehensive Assessment Reporting Evaluation (CARE) data
available in RDA’s Integrated Client Databases were used to identify indicators of substance use
disorders and/or mental health conditions based on diagnoses, prescriptions, and treatment
records. Drug- and alcohol-related arrest data maintained by the Washington State Patrol were
also used to identify probable substance use issues and were included in the definition of
treatment need for substance use disorders.
• Mental Health Condition. Mental health condition is indicated for any individual who: 1)
was diagnosed with a psychotic, mania/bipolar, depressive, anxiety, attention deficit and/or
hyperactive, disruptive/impulse control/conduct, or adjustment disorder; 2) had an
antipsychotic, antimanic, antidepressant, antianxiety, or ADHD prescription filled; 3) received
mental health services; or 4) received behavioral rehabilitation services from The
Department of Children, Youth, and Families.
• Substance Use Disorder. A substance use disorder is indicated for any individual who: 1)
was diagnosed with a substance use disorder; 2) had a prescription filled for medication for
opioid or alcohol use disorder treatment; 3) received substance use disorder treatment
services; or 4) was arrested for a substance-related offense.
• Chronic Illness Risk Score. An indicator of chronic illness was developed to identify
individuals with significant health problems. A risk score equal to one is the score for the
average Medicaid participant in Washington State meeting Supplemental Security Income
disability criteria. Chronic illness risk scores were calculated from health service diagnoses
and pharmacy claim information, with scoring weights based on a predictive model
associating health conditions with future medical costs (see Gilmer et al., 2001; Kronick et
al., 2000 for more information). Individuals were identified as having chronic illness if their
risk score was greater than or equal to one.
• Housing Indicator. Data from the Automated Client Eligibility System, Electronic Jobs
Automated System, ProviderOne, and the Homelessness Management Information System
were used to determine if individuals who received any form of cash, food, or medical
assistance and/or housing service experienced homelessness in the month. An individual
was considered unhoused if, for at least one day in the month: 1) they were identified as
homeless based on client living arrangement data entered into ACES; 2) they had an
address of “homeless” or an address format code that indicated the client was homeless; 3)
a medical or behavioral health provider submitted a claim to ProviderOne for that client with
a “Z590” or “Z591” diagnosis code; or 4) they were enrolled in any housing service that
requires someone to be literally homeless to be eligible for the service and did not have a
valid move-in date.
Permanent Supportive Housing. An individual was identified as having received permanent
supportive housing if they were enrolled in—but not necessarily housed through—a permanent
supportive housing program in HMIS in a given month. This measure excludes individuals who
received permanent supportive housing services through the Health Care Authority’s
Foundational Community Supports program and individuals housed through site-based PSH.
Permanent Supportive Housing Need Forecast
December 1, 2023 6
APPENDIX
Changes from Previous Forecast
There are several differences between the new forecast and the prior forecast. First, we relaxed
coverage-based exclusions to include individuals who had any form of medical coverage in at
least one month in the previous year. Prior forecasts were restricted to individuals with at least
one month of Title XIX Medicaid in the previous year. Second, current estimates include
individuals dually enrolled in Medicare and Medicaid. Prior forecasts excluded dual-eligible
clients, which artificially depressed the number of individuals 65 years old and older who met
PSH inclusion criteria. As shown in Figure 4, these two changes in our inclusion criteria resulted
in an increase in the absolute number of individuals identified as potentially eligible for PSH
benefits relative to previous estimates.
Third, the estimated rate of increase in the current forecast is lower than that previously
reported, with the 36-month PSH-eligible population forecasted to be 82,238 by December
2025, which is on par with the prior forecast of 83,971 in June 2025 despite the higher January
2023 starting point for the new forecast. One potential explanation for a comparatively lower
rate of increase in the new forecast is that we observed a sharp uptick in the number of
individuals potentially eligible for PSH during the last three months of the time series used in the
previous forecast. Because the forecasting method prioritizes more recent data, this uptick in
the count of individuals potentially eligible for PSH carried forward into the caseload forecast.
Updated information indicates that the rapid increase observed from March to July 2022 leveled
off over the following six months, dampening the forecasted trends in the current estimate
presented here.
FIGURE 4.
Comparison of Current Forecast with Previous Forecast
120,000
100,000
PRIOR
83,971 NEW
80,000
70,373 82,238
60,000
61,534 Despite changes to PSH eligibility criteria,
new 36-month forecast is on par with prior
40,000
20,000
Prior 36-Month Forecast
Estimated Need (Actual) 36-Month Forecast
0
J F MAM J J A S O N D J F MAM J J A S O N D J F MA M J J A S O N D J F MAM J J A S O N D J F MA M J J A S O N D J F M AM J J A S O N D J F M AM J J A S O N D J F M AM J J A S O N D
2018 2019 2020 2021 2022 2023 2024 2025
REFERENCES
Gilmer, T., Kronick,. R., Fishman, P., & Ganiats, T.G. (2001). The Medicaid Rx Model—Pharmacy-based risk
adjustment for public programs. Medical Care 39(11), 1188-1202.
Kronick, R., Gilmer, T., Dreygus, T., & Lee, L. (2000). Improving health-based payment for Medicaid beneficiaries:
CDPS. Health Care Financing Review 21(3), 29-64.
Mancuso, D., & Huber, A. (2021). Washington State Health and Human Services Integrated Client Databases.
Washington State Department of Social and Health Services, Research and Data Analysis Division, Report 11.205.
Patton, D., Kim, H., Liu, Q., & Felver, B. (2023). DSHS Economic Services Administration Service Trends Before and
During the COVID-19 Pandemic. Washington State Department of Social and Health Services, Research and Data
Analysis Division, Report 11.264.
Washington State Department of Commerce. (2023). Guidelines for the Consolidated Homeless Grant.
https://deptofcommerce.app.box.com/s/4d1ilui45uqljmhlseufez4flxqv1q6b.
Permanent Supportive Housing Need Forecast
December 1, 2023 7
File and source
- File
- 2023_permanent-supportive-housing-need-forecast-final-f8d8ca57-2302-41da-98.pdf
- Size
- 433,431 bytes
- SHA-256
- e4c3d18c8f07b894c3afa984a62b2882736d078f4c6b27e838efc0cc863338e1
- Original
- No public link identified.