The Challenge | Source of income discrimination is illegal in New York, but still happening.

“Lawful source of income” is a protected class under New York State Law and includes any federal, state, or local public or housing assistance towards the payment of rent. This law means that it is illegal for a landlord or broker to not rent to someone because they receive rental assistance in the form of housing vouchers. However, landlords and real estate brokers continue to turn away potential tenants who rely on subsidies to pay rent.

Data indicates that small landlords are the most common perpetrators of source of income (SOI) discrimination, making it difficult for the New York Office of the Attorney General (NY OAG) to track and prosecute. NY OAG reached out to my team at BIT to see if we could help prevent discriminatory behavior before it happens.

Landlords think SOI discrimination is normal— and they don't think they'll get caught.

We worked with NY OAG and a range of housing justice stakeholders to identify four behavioral drivers of SOI discrimination:

1. Landlord’s beliefs about rental assistance recipients & programs. Most landlords would prefer a tenant who doesn’t use rental assistance rather than one who does. Further, some landlords are not aware of (or are not persuaded by) the proposed benefits of renting to rental assistance recipients.
2. The economic landscape & competition in the rental market. Landlords in “desirable” neighborhoods have a larger pool of prospective tenants, which may make it easier for them to reject rental assistance recipients in ways that seem less explicitly discriminatory (e.g., using high-income requirements, “ghosting” potential tenants, or steering them to other neighborhoods).
3. Landlord’s expectations about enforcement of SOI non-discrimination law. Some landlords believe that the chance of being caught is very low, and that the penalties (if caught) are unlikely to be significant. Others have the (false) belief that discrimination is not illegal or that they are not liable for the action individually (e.g., if they work with a broker that screens out tenants receiving rental assistance for them).
4. Existing social dynamics that have normalized SOI discrimination. Negative perceptions about rental assistance recipients and inaccurate beliefs about enforcement reinforce a social norm that SOI discrimination is “OK.” This can exacerbate discrimination, for example, causing brokers to assume that landlords want to avoid tenants who will use rental assistance, without asking the landlord.

Approach | We needed to shift enforcement perceptions and reach landlords at an actionable moment for them to change their behavior.

In order to maximize the behavioral impact of a message about SOI discrimination, I wanted our intervention to be tied to live listings— in other words, I wanted my team to reach landlords and brokers at a moment when they are about to have the opportunity to discriminate (or not) for a specific unit.

I knew this would be logistically challenging, but also knew this touchpoint would optimize our chances of changing behavior. We ultimately designed an intervention that would deliver an email and a text message (basically a nudge to read the email) to the broker or landlord who lists a rental unit on the day they list it.

Design | I designed an intervention that incorporated timely action, social norms, and deliberate choice framing.

The intervention itself (particularly the email) included a range of evidence-based behavioral science concepts, guided by the drivers of discrimination my team had identified. These included:

• “Deliberate choice” framing: Because discriminatory behaviors typically take the form of inaction (not responding to a call, not following up about a viewing), we framed non-compliance with SOI discrimination as a deliberate act. For behaviors like vaccines that fall into this category, where one option is viewed as a behavior (getting a vaccine) and the other as a non-behavior (not getting a vaccine), it is important to reframe both choices as equally intentional and deliberate.
• Timeliness: Connecting the intervention to a specific listing increases the sense of being monitored and provides the nudge right before the behavior might take place.
• Dynamic social norms: The intervention notes that more and more cases are being prosecuted. Our goal here was to dispel the idea that this is something every landlord or broker does, or something lots of landlords and brokers get away with.
• Identity-based relatability: the intervention flags cases of prosecution for discrimination “in your area” or specific borough.

Evaluation | Results pending!

As of December 2024, the intervention has launched. We designed it as a randomized controlled trial (RCT), which is the gold standard for understanding the impact of an intervention because it ensures a random, representative sample and random allocation into a treatment vs a control group.

We partnered with another organization called the Housing Rights Initiative, who are supporting the evaluation by conducting 'paired tests'— essentially, one person calls a landlord or broker about a listing and pretends they will pay with a voucher, and another call about the same listing but pretends they will pay in cash. Both callers collect data on how they are treated and the ease of setting up a viewing. If all goes well, the intervention should lead to a smaller difference in the way the two groups in each paired test are treated.