The Struggle to Count Bodies & Beds

 By Sarah Owens and Michael Livingston

Every January, hope springs anew in communities across the country that they may be successful in the struggle to give an accurate accounting of all their homeless beds and bodies.  

Donations are solicited and warehoused, volunteers recruited and trained, and, to a greater or lesser extent, the count has been publicized in print and social media.  "The day of the count" will, usually, get heavy media coverage, even though the counting will continue for up to two weeks. 

The results, however, have tended to get very little, if any, attention.  Certainly not in Marion and Polk counties, which have historically been very slow to publish results compared to other geographic areas, or doesn't publish them at all, which happened in 2020.  

The state agency, the Oregon Housing and Community Services Department, publishes only odd-year counts, by county, but buries them deep in the OHCS website under a tab for providers, in inconsistent formats that make detailed comparisons difficult.  

Polk (L) and Marion (R) counties 2015-2017 PITC Summaries

 

Polk (L) and Marion (R) counties 2019 PITC Totals

So, what happened to the report on Marion and Polk's 2020 count -- the very first count for the newly-formed Mid-Willamette Valley Homeless Alliance?  See "The Pointless Point-in-Time Count."  24 December 2019.  

After asking Alliance staff repeatedly for several months when the 2020 PITC report would be published, and finally being told (incorrectly) "it's on the website", we decided to compile our own report, using raw data and the format that MWVCAA has used in prior years.  Turns out, the 2020 count showed a deep (>50%) drop in sheltered homelessness from previous years, and a significant (~10%) drop in overall homelessness from 2019.     

 

Realizing those figures couldn't be right, we asked MWVCAA Executive Director Jimmy Jones by email "what happened" with the 2020 count.  Here's what he told us.

The 2019 PIT Count showed 121 homeless persons in Polk and 974 in Marion, for a total in both counties of 1,095. The transition from a ROCC to a local process produced some confusion as there were a lot of people who were new to their roles.  The 2020 count showed 998 [homeless] persons [total], which is about a 10 percent reduction from the prior year.  Nothing fundamentally changed in the homeless population in Marion and Polk between 2019 and 2020.  The 10 percent decline was a product of new people in new roles, the transition to a new COC, and the use of a new technology platform also played a role. The shelter count was especially inadequate. And then, we have to come to terms with the fact that [the Homeless Management Information System (HMIS)] in Oregon is completely unreliable. I have no confidence in the ultimate numbers that are produced by the state.  In recent years they’ve had to make adjustments to make the numbers they have in their system, and the numbers reported to HUD, match.   Until we have a competent statewide system the numbers are unreliable.

So why then are we doing a full count in 2021?  [HUD offered to grant CoCs waivers that would relieve them of the requirement of conducting a 2021 unsheltered count, but the Alliance chose not to seek a waiver.]

Because the community has to make a better effort at getting this right and we can’t wait til 2022 to figure out how to do it the right way.  [An unsheltered count is required only in odd-numbered years.]   Unsheltered homelessness will continue to get worse the next decade, and state and federal funding formulas will bend increasingly heavily toward point-in-time counts, especially if there are additional future increases in the Emergency Solutions Grant, which during COVID have been [awarded in proportion to community need based on communities'] unsheltered counts.  There are a lot of reasons not to do a count this year, and I would have opposed it but for the simple fact that the window HUD gives us is a full two weeks under these [pandemic] conditions.  That dramatically increases the odds of getting something closer to a[n] accurate count.   We also have next to no information on how much our homeless population has increased because of the wildfires. I think it’s greater than people believe and we need to be able to prove that now.

We asked Jones to say more about his assessment of Oregon's HMIS as completely unreliable.  (HMIS is a shared database that is used to track individuals entering and exiting the homeless services system, and is supposed to let communities know how well or poorly their homeless services system is functioning.)  Basically, his view is that the reports generated in HMIS should be viewed with skepticism.  Ask about the report's design (what data was included and what data was excluded) and the margin of error.  Ask about the report's history of reliability, and whether any program-level staff have signed off on this particular report, and if so, what, if anything, did they think the numbers signify.  Don't just accept a report at face value. 

With that caveat, we turn to a less known, but equally important, metric:  the Housing Inventory Count, or HIC, which also occurs in January, and is designed to inform communities about their homeless shelter and housing programs.  The HIC tells communities how many beds they have, of what type, and the extent to which providers and programs are participating in HMIS. 

Marion and Polk counties started 2020 far behind where we should have been in terms of HMIS "bed coverage."  See "HMIS and Bed Usage Rates", revised January 2019.  Last January's HIC demonstrated what was known already -- i.e, that the Alliance had a lot of HMIS "on-boarding" to do with providers before it could begin to compete as a CoC.  In this respect, the Alliance's greatest accomplishment in 2020 was adding Simonka Place, a program of the Union Gospel Mission of Salem that serves women and children, to the number of providers participating in HMIS.  

That said, because UGM's men's programs won't be added until later this year, when the new Men's Mission opens, the Alliance will be falling short of HUD's 85% bed-coverage benchmark in Emergency Shelter (44%) and Transitional Housing (61%) types this January.  This failure means the Alliance will fall short on the 2021 System Performance Metrics, or SPMs, that are based on ES and TH data (which is to say, most of them).  That's in addition to the reliability problems that Jones identified.  To find out more about plans for the 2021 PITC, visit the Alliance web page here.