When deciding whether to rent to a tenant, most property managers and landlords check the applicant’s rental history, credit and income, and criminal background. As data on these metrics becomes more available and as companies search for a way to set their screening reports apart, many companies have developed algorithmic models that assign tenant risk scores based on available data.

Little public evidence shows risk scores accurately predict rental outcomes, and which histories are most clearly linked to lease violation remains unclear. Risk scores are unregulated, and advocates have highlighted that companies may include incorrect, outdated, and minor records that can prevent applicants from finding housing. As a result, some states and cities have regulated the data allowed in screening reports. 

To understand how these policies might affect prospective tenants’ ability to find stable housing, the tenant screening and housing access simulator allows users to simulate tenant screening regulations and their effects on imputed risk scores. Here, we highlight key findings and recommendations for policymakers interested in expanding access to housing.

How We Did It 

Using data from the Administrative Office of Pennsylvania Courts, we mirrored three common categories of policies regulating the data used in tenant screening: 

• Eviction masking policies, which seal or mask eviction filings under certain conditions, such as when a tenant is found “not guilty”;
• Clean slate policies, which seal or expunge certain criminal records, such as low-severity convictions, nonconvictions, and decriminalized convictions; and 
• Fair Chance legislation, which limits how certain criminal histories, such as misdemeanor convictions, can be used by landlords or tenant screening companies. 

Because these risk scores—and the algorithms that produce them—are proprietary, we developed “inferred risk levels” (low, moderate, high, and very high) to approximate the effects of these policies. 

Five Actions Policymakers Can Take to Support Housing Outcomes 

Our tool allows us to modify common risk criteria and change the duration of the look-back period—that is, the number of years a record can be included in a risk assessment. Then, we can see how the number of individuals defined as “moderate” or “high” risk changes. From this analysis, we synthesized five policy actions that local and state officials can take to support greater access to rental housing. 

1. Limit the use of nonconvictions, nonevictions, and decriminalized records in screening reports 

   Policy goal: Minimize the negative effects of the least severe kinds of records. Limiting access to records where an individual was found “not guilty” of the charge or filing and for decriminalized marijuana charges reduces rental housing barriers for “moderate risk” households substantially. Depending on the look-back period, sealing these records reduces the proportion of “moderate risk” households by 4 to 37 percent, with imputed risk declining more with shorter look-back periods. Because these records reflect unproven offenses or charges no longer criminalized by many states—for example, more than 20 states have legalized possession of certain amounts of marijuana and seven states have decriminalized marijuana possession—limiting their use in tenant screening can make a large impact with little cost.

2. Reduce look-back periods to align with perceived severity or relevance of the record

   Policy goal: Expand housing opportunities for people with low-severity crimes without changing regulations for severe crime. Policymakers can further target the visibility of low-severity crimes by adding misdemeanor charges and all eviction filings to the trio of record types above. Adjusting the look-back period to five years for these records shifts 45 percent of “moderate-risk” individuals to “low-risk” instead. Reducing the look-back period further, to three years, shifts 63 percent of “moderate-risk” individuals to the “low-risk” group.

   Additionally, reducing look-back periods for a single type of record has a larger effect on tenant risk than maintaining a modest look-back threshold across multiple policies. For example, completely removing access to misdemeanor records reduces those labeled “moderate risk” by 97 percent, while a consistent look-back period of five years for nonconvictions, misdemeanors, and decriminalized records reduces those labeled “moderate risk” by 43 percent. Policymakers can also consider using different look-back periods for different records to ensure policies fit local context. 

3. Align look-back periods with federal guidelines on other types of records

   Policy goal: Reduce barriers for renters with older records. Under current federal law, criminal convictions do not have defined look-back periods, meaning a single record from years ago can pose housing barriers to tenants indefinitely. However, other federal guidelines limit look-back periods broadly to seven years for standard credit reporting. Applying a seven-year look-back period to nonconvictions, decriminalized marijuana charges, misdemeanors, and nonviolent felonies increases the share of those labeled “low risk” from 3 percent to 16 percent. When look-back periods are reduced to five years, the share of people labeled “low risk” jumps to 25 percent. 

4. Limit or eliminate look-back periods for eviction filings 

   Policy goal: Expand housing opportunities for people considered “high risk” while targeting the smallest number of records. Reducing the look-back period of all eviction filings, regardless of outcome, decreases risk the most for individuals labeled “high-risk.” When a look-back period of three years is applied, 16 percent of those labeled “high risk” shift to the “moderate” or “low” risk group. If all eviction filings are sealed immediately, those labeled “high-risk” fall by 68 percent. Regardless of the look-back period (three years or immediate), most of those shifted out of the “high-risk” group move to the “low-risk” group. Given that eviction filings data tend to be low quality, limiting how these data are used in tenant screening risk scores may be the most significant way to help open doors to rental housing. 

5. Look upstream in legal and criminal processes

   Policy goal: Reduce housing disparities for different  race and ethnicity groups. Though screening tools themselves do not discriminate based on race and ethnicity, people of color face higher rates of arrest, conviction and eviction, making them more likely to be screened out of rental housing opportunities on the whole. We find that policy levers controlling the data included in tenant screening algorithms show little difference in the racial and ethnic distribution within each risk group before and after policy implementation. This may be because bias shows up earlier within the criminal-legal system, such as in arrests and convictions. For example, Black people make up an estimated 25 percent of individuals in our data, while Pennsylvania’s population is only 12 percent Black.