This manual describes the step-by-step procedures by which ImpactMatters analyzes emergency shelter nonprofits. We recommend reading the program analysis methodology first.
Nights of shelter (a.k.a. “shelter-nights”)¶
When the nonprofit has reported shelter-nights for a specific period on its Form 990: Shelter-nights may be found in Part III lines 4a–d, Additional Data and Schedule O of Form 990. If found, we always use the 990 data instead of calculating shelter-nights using the U.S. Department of Housing and Urban Development (HUD) Housing Inventory Count (HIC) method described below. We generally do not search other sources from the nonprofit (e.g., annual report, GuideStar profile) because shelter-nights are very infrequently reported on such sources.
The nonprofit may not report shelter-nights per year. If the nonprofit reported shelter-nights per month, and is likely open every week, we multiply reported shelter-nights by 12 months. If the nonprofit reported shelter-nights per week, and is likely open every week, we multiply by 52 weeks. If the nonprofit reported shelter-nights per day, and it is likely open all year, we multiply by 365 days. If the nonprofit is a seasonal shelter, we multiply by the estimated number of days it is open, e.g., there were 90 days in winter 2016–17.
However, use the HUD HIC method instead if:
The nonprofit has only reported bed capacity (without a corresponding capacity utilization rate).
The nonprofit has only reported the number of unique beneficiaries served (without a corresponding average length of stay per beneficiary).
The nonprofit serves families and has not clearly indicated that its shelter night count is per individual and not per family.
When the nonprofit has not reported shelter-nights for a specific period: In this case, the HUD HIC method can be used to calculate shelter-nights based on the point-in-time (PIT) count of sheltered individuals on a single night in January.1
When using HUD HIC data, we account for the fact that the PIT count is conducted on a single night in January, which may not be representative of the average night in a year. In areas with harsh winters, for instance, the count of sheltered individuals may be much higher in January than in most other months. Therefore, instead of multiplying shelter-specific PIT counts by 365, we multiply by 345. This is calculated based on quarterly PIT counts available at the national level, which show that there were, respectively, 92 percent, 93 percent and 93 percent as many homeless people in emergency shelters on a single night in October, April and July as there were on a single night in January.
In the HUD HIC dataset, if the number in column “Total Seasonal Beds” is 0, this means the nonprofit provided only year-round beds and the PIT count can be assumed to reflect the number of people it sheltered on an average night. We then multiply the PIT count by 345.
If the number in column “Total Seasonal Beds” is non-zero, then we calculate total shelter nights by the following formula: [(PIT Count - Total Seasonal Beds ) * 345] + [Total Seasonal Beds * DAYS(Availability End Date, Availability Start Date)]. The start and end dates correspond to the period when seasonal beds were available (usually during winter). This effectively assumes 100 percent capacity utilization of seasonal beds, which we think is reasonable given these beds were likely made available to meet peaks in demand.
Starting figure for program costs¶
We first search the nonprofit’s Form 990 (Part III lines 4a–d) for emergency shelter-specific costs. If they cannot be found, we search for audited financials on the nonprofit’s website, annual report and the Federal Audit Clearinghouse.
Extra services for shelter residents¶
We note whether an emergency shelter provides extra services besides a bed and associated basic services (e.g., showers, clothing, other personal hygiene and grooming). We first review the nonprofit’s Form 990 for written descriptions of any of the following extra services. Following this, we search the nonprofit’s website. Note that presence of certain line items and noncash contributions on the nonprofit’s 990 or financial statements are generally not considered strong enough evidence of the existence of a corresponding programmatic activity. For instance, presence of food donations are not considered evidence that the nonprofit serves meals to its emergency shelter residents. The nonprofit must describe the presence of extra services in writing. Exceptions may be made if the line item itself unambiguously indicates the existence of an extra service (e.g., “Evening meals for emergency shelter guests”), particularly if the nonprofit only runs an emergency shelter (i.e., the line item could not possibly be related to other programs run by the nonprofit).
The list below identifies and describes these extra services2:
Food: Providing meals and/or groceries to shelter beneficiaries.
Case management: “Assessing, arranging, coordinating, and monitoring the delivery of individualized services to meet the needs” of shelter beneficiaries.
Child care: Caring for the children of shelter beneficiaries, “including providing meals and snacks, and comprehensive and coordinated sets of appropriate developmental activities.”
Education services: “Improving knowledge and basic educational skills” of shelter beneficiaries.
Employment assistance and job training: Helping shelter beneficiaries gain employment by training in job-related skills.
Outpatient health services: Providing shelter beneficiaries with “direct outpatient treatment of medical conditions.”
Legal services: Providing shelter beneficiaries with “legal advice and representation by attorneys.”
Life skills training: “Teaching critical life management skills” that allow the shelter beneficiary “to function independently in the community.”
Mental health services: Providing shelter beneficiaries with ”direct outpatient treatment [for] […] mental health conditions.” This involves “the application of therapeutic processes […] in order to bring about positive resolution of the problem or improved individual or family functioning or circumstances.”
Substance abuse treatment services: Services that help shelter beneficiaries “prevent, reduce, eliminate, or deter relapse of substance abuse or addictive behaviors.”
Transportation: Services that help shelter beneficiaries “travel to and from medical care, employment, child care, or other service facilities.”
Services for special populations: Specialized services for shelter beneficiaries who fall under the category of: “homeless youth[; victims of domestic violence, dating violence, sexual assault or stalking; or] people living with H.I.V./AIDS.” These services are distinct from general case management.
Recall that these services need to be provided to shelter residents. If these services are provided to non-residents, then we disqualify that nonprofit and do not complete an analysis. This holds when a nonprofit delivers a service to both shelter residents and non-residents unless we believe that the portion of services delivered to non-residents is insubstantial. We do not disqualify nonprofits that operate hotlines and referral services to the shelter because we believe these costs are insubstantial and because these services function as a targeting mechanism for the shelter.
If the nonprofit has itemized its costs, we simply subtract the costs of the above extra services from the starting figure for program costs.
If the nonprofit has not itemized its costs, we apply a 5 percent discount to program costs for each extra service provided. E.g., if a family shelter provides both case management and child care, we discount program costs by a total of 10 percent.
We make an exception for nonprofits that have been designated a “Victims Services Provider.” Because of the substantial associated costs, if a shelter has such a designation by HUD and we find written evidence that such services are provided, we subtract 25 rather than 5 percent.
If the nonprofit has itemized costs for some services but not others, we subtract the available line items and apply the 5 percent discount for missing services.
If the nonprofit has itemized some costs for a service but we do not believe that the itemized costs fully account for the entire cost of that service, then we do not subtract the itemized cost and instead apply the 5 percent discount.
We assign a cost of $0 to noncash contributions such as food, clothing, drugs and medical supplies on the assumption that these items, if not donated, would otherwise have gone to waste.
Under Generally Accepted Accounting Principles, nonprofits are required to report the value of noncash contributions (a.k.a. in-kind donations).3 Noncash contributions are reported in the statement of functional expenses if they were expended during the fiscal year (e.g., donated food served to beneficiaries that year). As such, we assume that by subtracting the costs of extra services for shelter residents (described above), we are also subtracting any noncash contributions expended as part of those extra services. For instance, the cost of donated food distributed is already subtracted when we subtract the cost of the shelter’s meal services and the cost of donated medical supplies are already subtracted when we subtract the cost of the shelter’s outpatient medical care. Further subtraction of noncash contributions is therefore unnecessary. The same rule applies even if we are using the standard 5 percent cost adjustment in the absence of a better figure from the nonprofit’s own financials.
Other noncash contributions that go directly toward shelter operations rather than extra services for shelter residents (e.g., clothing and household goods) are assumed to be small compared to total program cost and need not be subtracted.
Other shelter and housing programs¶
Some nonprofits run both an emergency shelter as well as other shelter or housing programs, such as transitional housing or permanent supportive housing. Wherever possible, we exclude from our calculation the costs associated with these other shelter and housing programs, as reported by the nonprofit.
If the nonprofit has not separated out programmatic costs in this way, we apply a standard cost adjustment based on HUD HIC data. First, we calculate the fraction of the nonprofit’s total PIT count that comes from emergency shelter programs only. Then, we multiply that fraction by the starting figure for program costs.4
We assume that government costs are reflected entirely in the nonprofit’s financials.
We assume no partner organizations incur costs to provide a night of shelter. While emergency shelters commonly partner with other nonprofits like food banks and soup kitchens to provide meals to shelter residents, these partners are helping to provide ancillary services beyond those that generate the outcome of interest, a night of shelter.
We use HUD’s list of program components eligible for funding under the Emergency Solutions Grant. Quotations in the list are taken from this document. See HUD: Homeless Emergency Assistance and Rapid Transition to Housing: Emergency Solutions Grants Program and Consolidated Plan Conforming Amendments. We add to this list any food services a shelter provides.
This method effectively weights emergency shelter and other shelter or housing programs equally, making the simplifying assumption that each type of program costs about the same, on average, to shelter an individual for a night. In reality, HUD finds that emergency shelter for individuals tends to cost the least per day compared to transitional housing and permanent supportive housing, but emergency shelter for families tends to be equally or more expensive than transitional housing and permanent supportive housing. However, there is considerable variation in HUD’s findings, so we cannot be confident that ratios of relative cost calculated using HUD’s findings would be any more or less accurate than our simplifying assumption.