U.S. Markets closed
  • S&P 500

    3,934.38
    -29.13 (-0.73%)
     
  • Dow 30

    33,476.46
    -305.02 (-0.90%)
     
  • Nasdaq

    11,004.62
    -77.39 (-0.70%)
     
  • Russell 2000

    1,796.66
    -21.63 (-1.19%)
     
  • Crude Oil

    71.59
    +0.13 (+0.18%)
     
  • Gold

    1,809.40
    +7.90 (+0.44%)
     
  • Silver

    23.68
    +0.43 (+1.87%)
     
  • EUR/USD

    1.0545
    -0.0014 (-0.1371%)
     
  • 10-Yr Bond

    3.5670
    +0.0760 (+2.18%)
     
  • Vix

    22.83
    +0.54 (+2.42%)
     
  • GBP/USD

    1.2250
    +0.0012 (+0.0980%)
     
  • USD/JPY

    136.5500
    -0.0800 (-0.0586%)
     
  • BTC-USD

    17,155.28
    -56.10 (-0.33%)
     
  • CMC Crypto 200

    402.11
    -4.14 (-1.02%)
     
  • FTSE 100

    7,476.63
    +4.46 (+0.06%)
     
  • Nikkei 225

    27,901.01
    +326.58 (+1.18%)
     

As Eviction Cliff Nears, Here's How Millions Could Keep Their Homes

  • The path to forecasting evictions is rocky, and more than 3 million Americans say they are at risk of eviction. The actual number of evictions, which could be as low as 130,000 households, depends on the speed of relief distribution, pace of economic recovery and decision-making by landlords.

  • More than 8.3 million U.S. renters said they were behind on rent in the Census Bureau's latest Household Pulse Survey, with 16.8% believing they are 'very likely' to be evicted in the next two months.

  • The federal eviction moratorium that has kept many renters in their homes during the pandemic is set to expire March 31.

Prior to extension of a federal eviction moratorium, some 3.4 million Americans feared they were at some risk of eviction this spring, and as many as 664,000 households - or as few as about 130,000 - could have ultimately been evicted, according to a Zillow analysis. The wide range of possible outcomes is a testament to the very murky path forward for evictions as future policy guidelines and landlords’ behaviors remain open questions.

Covid-19 has dramatically increased housing insecurity for renters — especially renters of color — who have been disproportionately impacted by job disruptions and economic hardship. More than 8.3 million U.S. renters reported that they were behind on rent as of March 15, according to the Census Bureau's Household Pulse Survey (HPS). Of those, 1.41 million (16.8%) said they thought they were "very likely" to be evicted in the next two months. These are sobering numbers, but it's important to note that only a small fraction of those fearing eviction will likely be evicted.

The tricky part is determining how small that fraction will ultimately be — and in crunching some basic numbers to illustrate how it might be even smaller, should landlords choose to work with struggling tenants rather than pursue eviction.

There are important differences between a renter at risk for eviction, an eviction filing and an eviction judgement:

  • At-risk eviction candidates are renters behind on rent and of the belief they are likely to be evicted.

  • An eviction filing is a request for a court hearing, where a judge will rule if an eviction is ultimately correct or necessary.

  • An eviction judgment is a judge's ultimate decision to remove a tenant from a property.

Expectations vs. Reality

What this tells us is that just because a tenant is behind on rent, that does not definitively mean their landlord will file for eviction. And not all eviction filings result in removal of a tenant from a property. This distinction is borne out in the HPS data, where respondents have thus far overestimated their eviction risk.

In the December 21, 2020 HPS survey, for example, 6,917 households (15,309 total people) from Connecticut said they were "very likely" to be evicted in the coming two months. But according to data from the Eviction Lab's Eviction Tracking System, there were only 1,229 eviction filings in Connecticut between December 27, 2020 and February 21, 2021 — only 17.8% of self-reported eviction candidates. And all of those filings will not result in an eventual eviction — historically, an estimated 75.4% of Connecticut eviction filings result in an eviction judgement to remove tenants from their Connecticut homes. Similarly, over the same period, 39,447 households (62,987 total HPS respondents) from Indiana indicated they were "very likely" to be evicted, but there were only 8,712 filings — just 22.1% of those Indiana renter households that believed they were "very likely" to be evicted. And again, while the ultimate number of eviction judgments is not yet known, historically in Indiana just 53.1% of eviction filings result in actual eviction judgements.

The Eviction Tracking System is arguably the most complete data source for current eviction information nationwide, but even so it only currently includes filing data for five states — Connecticut, Delaware, Indiana, Minnesota and Missouri — and 15 metropolitan areas, a data reality that creates a murky picture of the eviction crisis at both the state and national levels.

Simulating the eviction crisis

It's clear that a federal eviction moratorium that has been extended several times since first enacted is largely achieving its goal and keeping renters in their homes, but the question remains: Even though it was recently extended, what will happen once that moratorium is eventually lifted? Given limited public data, an uncertain policy track going forward and the simple fact that the current national situation has virtually no historical precedent, the number of nationwide evictions that will ultimately occur is very difficult to forecast.

The best we can reasonably do is simulate different futures based on available data and simple assumptions. Additionally, because there is a limited historical baseline between HPS data (which only begins at the start of the pandemic) and eviction filings, any simulation must reflect the very uncertain nature of the problem. But assuming landlords in all states evict at the same rate as the 5 states tracked by the Eviction Lab, our analysis suggests there is a 95% probability that the total number of eviction filings will be between 188,000 and 378,000 in April and May, with an average of 284,000.

There were 2.35 million eviction filings nationwide in 2016 (the most recent year for which complete data is available from the Eviction Lab), an average of 392,000 every two months. So our simple simulation based on matching all states to the eviction rates from the 5 available states is roughly 70% of the historic baseline — likely an indication that landlords today are filing fewer evictions than they have in the past, either through increased compassion for their tenants, local/federal restrictions or both.

Adding additional uncertainty to the simulation to reflect the fact that landlords in all states will not behave like the five for which we have data, the number of likely evictions (and the amount of uncertainty) increases: A 95% probability that the number of national eviction filings will be between 251,000 and 1.18 million in April and May. The average number of filings in this new simulation would be 577,000 over two months, about 47% higher than the 2016 benchmark. In this more uncertain simulation, an average of 310,000 households are actually evicted — just more than half the number of eviction filings — and there is a 95% chance that the number of evictions will be between 130,000 and 664,000.

Our main takeaway? The number of ultimate evictions is a very uncertain and moving target.

An Alternative to Eviction

The main sources contributing to eviction uncertainty are federal policy, the economic recovery and landlord decision-making. On the political front: Will the Biden Administration extend the current moratorium, and if so, for how long? On the economic front: When will quality jobs for renters return? The recent $1,400 stimulus payments and $300 federal supplement to weekly unemployment benefits are keeping renters afloat, but their long-term housing stability depends on a robust labor market recovery. And perhaps the most critical unknown lies with landlords themselves: Will they unilaterally move to evict those tenants behind on their rent? Or will they choose to work with tenants to find mutually beneficial solutions?

Every tenant-landlord relationship is different, but evictions based strictly on owed back rent may not always be in everyone's best interest. For example, after going through the time-consuming eviction process, landlords may find themselves with a rental unit that is vacant for several months — costing them still more money while they search for a new tenant. Tenants need to find a new home, which is both personally disruptive and expensive when moving costs, security deposits, utility fees and upfront application/first month/last month costs are factored in. A reasonable alternative to eviction, then, may be to amortize back rent into current monthly payments, keeping tenants in their homes while landlords effectively serve as private creditors as they recover back rent over an agreed-upon time frame.

Consider a tenant on a $1,500/month lease that owes six months of back rent, or $9,000 total. If the landlord evicts, they can expect roughly $1,750 in monthly costs for each month that the unit is vacant, according to an analysis by TransUnion's SmartMove. If the apartment is vacant for 6 weeks, the landlord will have incurred $2,625 in costs for evicting the tenant, not to mention the additional time required to advertise the apartment, screen new applicants, show the unit and conduct background checks. If this landlord evicts without recovering back rent, they will face a total loss of $11,625 ($9,000 in unrecovered back rent + $2,625 in sunk costs).

Now, consider a case where the landlord works with the tenant on a repayment plan and amortizes the back rent into ongoing rent payments. If the tenant repays that $9,000 debt over two years at an interest rate of 6% (an admittedly ballpark figure for unsecured debt that is roughly twice current mortgage interest rates, but less than interest charged on most credit cards), their monthly payment would be $398.89, on top of their contracted $1,500/month, for a total of $1,898.89/month (assuming no additional rent hikes along the way). In this case, the landlord would collect $9,573.25 in back rent and interest, in addition to ongoing rent, rather than writing off a $11,625 loss. If the repayment period were three years, that total monthly payment would come down to $1,773.80, and the landlord would collect $9,856.71 in back rent and interest.

These kinds of arrangements might not be feasible for landlords in need of a steady cash flow to make mortgage payments and remain afloat, and in certain cases generating that cash flow might be faster by evicting and starting over. And regardless, some elevated level of evictions should be expected under most scenarios whenever local and federal eviction moratoria do eventually expire. But if/when eviction alternatives are pursued, both sides can realize benefits. The benefit to the tenant is avoiding an eviction, albeit with a significant rent increase for a fixed period of time. The benefits to landlords are keeping their units occupied and avoiding turnover time/cost, and — over time (which is a tradeoff of its own) — being made financially whole and then some, rather than writing off an immediate loss. Again, every tenant-landlord relationship is unique and landlords are ultimately in charge of their own businesses, but the above analysis shows that the choice to evict based on back rent alone is not clear cut when there are reasonable alternatives.

Methodology

The number of renters at risk for eviction is estimated by assigning probability distributions to verbal responses like "very likely," "somewhat likely," "not very likely" and "not at all likely." Specifically, we use the Beta distribution, which is a family of probability distributions on the unit interval, and parameters are chosen so that the expected probabilities assigned to each response type align with the verbal probability literature.

Table 1. Eviction probability distributions

Response

Mean Eviction Probability

2.5-percentile

97.5-percentile

Very Likely

0.875

0.779

0.952

Somewhat Likely

0.675

0.537

0.799

Not Very Likely

0.125

0.048

0.225

Not At All Likely

0.02

0.000

0.069

For example, the mean eviction probability for a response "very likely" is 0.875, with 95% uncertainty interval spanning 0.779 to 0.952. In other words, respondents indicating that it is very likely that they will be evicted in the next two months, on average, believe that means there is an 87.5% chance of an eviction happening, but it could be as low as 77.9% or as high as 95.2%. Each respondent likely has a different interpretation of these verbal probabilities, and our simulation endows each respondent with their own probability distribution. A Monte Carlo simulation is used to generate a binary "at-risk" or "not-at-risk" classification for each respondent given their interpretation of the categories, and the total number of at-risk renters is then the sum of the person-level at-risk binary variables.

From the total number of at-risk renters, the number of likely eviction filings is simulated based on data from the Eviction Lab's Eviction Tracking System. Specifically, we estimate the proportion of at risk respondents whose landlords file for eviction in five states: Connecticut, Delaware, Indiana, Missouri, and Minnesota, the only states for which current data is available in the Eviction Tracking System. Specifically, we use Household Pulse Survey Responses from Week 17 (ending October 26, 2020) and Week 21 (ending December 21, 2020) and tabulate the number of evictions in the two months following each of those Household Pulse Surveys to calibrate the number of eviction filings based on the number of respondents at risk. We use a Bayesian mixture model of Beta-Binomial distributions to estimate this fraction, and then simulate the number of likely eviction filings from the most recent Household Pulse Survey (Week 26 – ending March 15, 2021) by applying the estimated fraction of at-risk renters that result in eviction filings. One nuance is that we consider cases where each state is a blend of the existing five states for which historical data exists, and a case where each state behaves quite differently from the current five, and simulate a new probability from a diffuse prior distribution for the rate of eviction filings from at-risk renters.

We simulate the number of evicted households with a second stage Beta-Binomial model. Historical Eviction Lab Data at the state-level is used to estimate the state-specific historical proportion of eviction filings that result in eviction judgements. Again, this estimated fraction of eviction filings that result in actual evictions is then applied to the most recent Household Pulse Survey data to simulate the number of evicted households from March 15 – May 15.

The post As Eviction Cliff Nears, Here's How Millions Could Keep Their Homes appeared first on Zillow Research.