Theme: Shelter and Housing COVID-19 & Evictions Agent-Based Modeling Results Key Takeaways • Our model predicts that once the moratorium is lifted, roughly 20% of renting households in Maricopa County will be affected by housing insecurity. • The model also projects significant household debt at the end of the moratorium — up to $1 billion in total rent and utility arrears. This debt will continue to increase as evicted households are rehoused but not recovered financially enough to pay full rents. • If renters do not have to save to pay rehousing costs and are quickly rehoused, we predict another jump in evictions as renters who can’t maintain the new rental payments are evicted again. • The model showed that even with agencies working at max capacity, all assistance funds will have been depleted by the end of the moratorium. Background and Methods The KER Eviction Model describes eviction dynamics within the context of the rental and utility assistance landscape for the greater Phoenix metropolitan area. This model is an agent-based model, in which each agent represents a renting household, and the Knowledge Exchange for Resilience population of these households derives from sampling a subset of the American Community Survey five-year data from 2015 to 2019. The sample captures income, rent and utility costs, as well as details about building type, household size, employment and income source. The model replicates the household budget decisions each month, where each household pays rent, utilities and subsistence costs. If a household experiences a shortfall during a month, it can request assistance from either a rental or utility assistance agency, and is granted that assistance according to the agency rules. If the household is unable to pay its rent over consecutive months, it will be evicted from the property and will seek rehousing once it has paid its debts and saved enough for rehousing deposits. The model was adapted to explore the possible consequences of the CDC moratorium on evictions starting in March 2020, expected to end with July 2021. The model was used to explore a series of questions relating to the degree of COVID-19-related housing insecurity, the timing of resuming evictions, and the effects of stimulus and debt reduction policies. The model was further modified to incorporate housing assistance rule changes in the ERA1 and ERA2 programs. Most notably, households can obtain assistance to pay 12 months of back rent, and can also obtain funds to cover rehousing expenses. We also updated the economic shock scenarios to match the monthly expected loss of income reported in the Census Bureau’s PULSE survey data from the Bureau of Labour up to May 2021, and August 2021 then gradually tapered off the shock through 2022. The baseline rate, just under 6% of renters, is the model’s eviction rate without an economic shock. We do not know precisely how many households are assisted each month by the various agencies, so we’ve run the model with a variety of assistance capacities, from 250 households per agency per month up to 10,000.1 Findings Figure 1 shows the number of households each month unhoused due to an eviction. As these households find homes, that number decreases. We see that without the requirement for rehousing costs, households are quickly rehoused, but they still can’t maintain rental payments so are evicted again, thus there is a bounce in the graph which eventually tapers off. Most importantly, the model predicts that once the moratorium is lifted, roughly 20% of renting households will face the threat of eviction. Table 1 shows the number of households under threat of eviction as the moratorium ends. Figure 2 shows the same information but with a first and last month’s rent rehousing cost that isn’t paid for through assistance, and the bounce disappears, but at the end of 2022 about the same number of households are without homes due to evictions. The remainder of the graphs will use the funded rehousing costs version since that is the current rule, but at the end of 2022 about the same number of households are without homes due to evictions. 2023 2022 2021 Figure 3 shows the predicted number of evictions per month. Again, we see the bounce due to the assistance in rehousing which generates a two month delay between eviction cycles. Figure 4 shows the same information as in Figure 1 but with the households broken into single individual households and multi-person households, and we observe that the majority of households facing eviction are multi-person households. 0.25 percent of renters evicted 0.20 capacity 0.15 0.10 250 500 1000 2500 5000 10000 0.05 Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan 0.00 month Figure 1: Percentage of rental population without housing due to an eviction and with rehousing costs funded by a program. Information from the City of Phoenix ERA dashboard, https://opengov.civicdashboards.com/embed/371959, suggests the actual capacity may be around 500 households a month. 1 August 2021 Capacity Eviction Threat SinglePerson MultiPerson Percentage of Renters 250 74,350 17,770 56,580 19.35% 500 73,330 16,990 56,340 19.09% 1,000 71,340 16,820 54,520 18.57% 2,500 69,190 15,570 53,620 18.02% 5,000 68,050 15,700 52,350 17.72% 10,000 68,720 15,470 53,250 17.89% Table 1: Households at risk of eviction by household type based on capacity of rental assistance agencies, in the case where rehousing costs are funded by a program. The percentage of renters is based on the number of renters in the model, 384,060. We also modeled the demand for assistance versus the number of appointments available. The number of appointments is based on both the agency capacity as well as the agency funds. If an agency has no more funds, it won’t be able to assist households even if it has capacity. Note the increase in demand for assistance during the pandemic. The number of appointments granted falls to 0 around the end of the moratorium because the agencies are out of funds, as demonstrated in Figure 6, where we show the available funds per month, primarily obtained through the ERA programs, for three municipally-based agencies. The Maricopa agency represents the municipalities of Tempe, Gilbert and Chandler. These funds were spent during the pandemic to keep households current in their rental payments. Note that when agency capacity increases, funds drop more quickly. Essentially, this model describes the consequences of the steady accumulation of housing-related debt during the course of the eviction moratorium. Previous simulations have demonstrated the importance of a full economic recovery before the end of the moratorium so that households could begin to pay down their housing debt. The model predicts significant housing debt at the end of the moratorium, which will continue to increase as evicted households are rehoused, but still not recovered financially enough to pay full rents, and these projections are given in Table 2. Note that this model does not include the expected increase in rental costs, partly because we expect an increase in the supply of rental units at the end of the moratorium, which may reverse that increase. We don’t yet know whether this surge in evictions will lead to a similar increase in the number of households experiencing homelessness. However, a preliminary regression analysis of Homeless Management Information System (HMIS) data by David Little and Michael Simeone of ASU Libraries shows that from 2017 to 2018, one household out of every 3 to 4 households evicted entered the continuum of care system that supports those in housing crisis. August 2021 2023 2022 2021 0.25 percent of renters evicted 0.20 capacity 0.15 0.10 250 500 1000 2500 5000 10000 0.05 Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan 0.00 month 2023 2022 2021 Figure 2: Percentage of rental population without housing due to an eviction but with rehousing costs not funded by a program. monthly evictions 60000 capacity 40000 250 500 1000 2500 5000 10000 20000 Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan 0 month Figure 3: Evictions per month once moratorium ends August 2021 2023 2022 2021 percentage of renters evicted 0.15 type 0.10 capacity 0.00 250 500 1000 2500 5000 10000 Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan 0.05 multiperson single month 2023 2022 appointments requested vs granted 2021 Figure 4: Percentage of evicted entities that are single individuals or multiple individual appointment 40000 granted requested capacity 20000 250 500 1000 2500 5000 10000 Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan 0 month Figure 5: Requests for rental assistance vs rental assistance appointments available. August 2021 Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan money to disperse in $millions Maricopa 9 6 3 0 Mesa 9 capacity 6 250 500 1000 2500 5000 10000 3 0 Phoenix 9 6 3 0 month Figure 6: Available assistance funds. August 2021 2023 2022 2021 Capacity Total Arrears Rental Arrears Utility Arrears Per Capita Arrears 250 500 1,000 2,500 5,000 10,000 1.015 billion 977 million 926 million 841 million 828 million 827 million 857 million 824 million 781 million 708 million 698 million 697 million 158 million 153 million 145 million 133 million 130 million 130 million 2,643 2,544 2,411 2,190 2,156 2,153 Table 2: Amount of household debt as of July 2021 from utility and rent arrears. The per capita arrears represents the total amount divided by the number of renters in the model (384060). References Sean Bergin, J Applegate (2021, July 31). “Model of Rental Evictions in Phoenix During the Covid-19 Pandemic” (Version 1.0.0). CoMSES Computational Model Library. Retrieved from: https://www.comses.net/codebases/4db84f11-90a2-4d2e-a2c9-ea43574b5c46/releases/1.0.0/ Lamar, Shea, Lora Phillips, and Patricia Solís. (2021). “Visualization: Maricopa County Evictions Dashboard.” Resilience Data Dashboard Series, Knowledge Exchange for Resilience, Maricopa County Justice Courts, 2016-2020. Available from Arizona State University, https://resilience.asu. edu/evictions-dashboard. Citation Applegate, J M and Sean Bergin. (2021). “COVID-19 & Evictions Agent-Based Modeling Results” [White Paper.] Resilience Briefings Series: Shelter and Housing, Knowledge Exchange for Resilience. Available from Arizona State University Library, https://hdl.handle.net/2286/ R.2.N.160733. Acknowledgements Thank you to the ASU School of Complex Adaptive Systems, the staff at the Maricopa Association of Governments and the many community partners providing utility and rental assistance who provided deep knowledge about the housing system. Authored by: J M Applegate, PhD, Assistant Research Professor, School of Complex Adaptive Systems, Arizona State University; Sean Bergin, PhD, Assistant Research Professor, School of Complex Adaptive Systems, Arizona State University Design support: Abby Johnson, Multimedia Specialist, Knowledge Exchange for Resilience, ASU Prepared for: Maricopa Association of Governments August 2021 Disclosure The ASU Knowledge Exchange for Resilience is supported by Virginia G. Piper Charitable Trust. Piper Trust supports organizations that enrich health, well-being, and opportunity for the people of Maricopa County, Arizona. The conclusions, views and opinions expressed in this article are those of the authors and do not necessarily reflect the official policy or position of the Virginia G. Piper Charitable Trust. August 2021