In New York’s Affordable Housing Lottery, Fewer Choices Could Make Better Matches

Runaway demand and an open, accessible system fail to put keys in the hands of New Yorkers who could benefit most from affordable units

Print this page

The United States is in the midst of an affordable-housing crisis. According to a 2016 report by Freddie Mac, the federal home loan mortgage corporation, affordable rental stock, defined by the agency as units costing no more than 30 percent of monthly income, fell by 60 percent between 2010 and 2016, from 11 percent of units to just 4. The shortage is particularly acute in major urban areas, where much of the job growth has been concentrated since the Great Recession.

In New York City, even as the rental market has softened, median rents have risen at twice the rate of wages since 2010. Mayor Bill de Blasio made affordable housing a major issue in his reelection campaign, promising to deliver an additional 100,000 affordable units by 2026, in addition to the 200,000 units already planned for the end of 2022.

Many of these units are developed as part of the Affordable New York program, which offers a 35-year tax exemption to developers of buildings with over 300 units that set 25 to 35 percent of units aside for lower-income tenants. These are allocated via NYC Housing Connect, an online portal launched in 2012 that allows New York residents to see the stock of available units and enter lotteries for them in mere minutes. So far, over one million New York City residents have registered on the site.

This convenience, however, has come at a price. According to a new working paper by Nick Arnosti of Columbia Business School and Peng Shi of the University of Southern California, the outcomes of New York’s lottery differ little from random assignment of tenants to units. What’s more, the lottery’s very accessibility is partially to blame. A simple fix, like capping the number of lotteries applicants could enter, or allowing winners to reserve their priority for another unit, could markedly improve outcomes.

“The system isn’t working as well as it could,” Arnosti says. “And we have some ideas for how to improve it.”

The lottery complements the city’s waitlist for public housing, but the prime locations and high quality of the units attract much broader public interest. Vying for a federal award last summer, city officials credited Housing Connect with attracting an “unprecedented” volume of lottery applications — a 14-fold rise since its launch — many of them the result of New Yorkers entering contests for buildings with a single click.

At present, many aspiring tenants, daunted by the vanishingly low odds of winning anything, apply to everything — any unit for which they qualify, in any building type, in any neighborhood. And, because it is a tremendous gamble to turn down an offer, lottery winners are incentivized to accept any apartment they’re allotted, even ones that don’t meet their needs.

Ironically, part of the city’s impetus for digitizing its lottery system was the concern of property developers that they’d lose money while waiting for eligible applicants to find their listings. The internet, however, has turned this problem on its head.

Between October 2015 and May 2016, for instance, the city received almost 2.5 million applications for the 2,628 new rental units it made available. That spring, total registered users on Housing Connect numbered roughly 700,000 — a number that has since grown to1.7 million — indicating that many people sent in multiple bids.

Confirming this fact was Vicki Been, a senior New York housing agency official, who told city councilors in May 2016 that many users “check the option for ‘apply to everything’ — even if it’s senior housing and they’re not seniors. … They just apply.”

With levels of interest so high, even after screening out unqualified applicants, efficiency is not primarily a matter of finding enough people, but rather of finding those most in need, and those who are the best match for each individual unit.

Meeting one objective does not always satisfy the other. A lottery win could help a struggling family avoid becoming homeless, but could place them miles away from their support network, school, or work. Meanwhile another household, one better suited for that particular building, could be left to wait for the next draw. Situations of this kind are costly for both households and their communities, fraying the very social fabric that affordable housing aims to preserve.

Policymakers often face a tradeoff between targeting the neediest cases and making the best matches, Arnosti and Shi demonstrate. That is particularly true for repeated lottery systems like New York’s, which allows participants to enter any number of contests at no additional cost. Incentives created by such systems precipitate actions that, when repeated by hundreds of thousands of rational participants over time, push the system to a suboptimal equilibrium. A repeated lottery of this kind might somewhat succeed at placing the neediest tenants in new homes, the authors find, but it does a poor job of matching tenants with units that meet their needs and preferences.

An ideal system would account for applicants’ preferences over dimensions such as location, accessibility, and pet-friendliness, but asking applicants to rank every possible combination of these attributes is not practical. Furthermore, this approach would rely on applicants providing this information honestly, even when it might not be in their best interest to do so.

Fortunately, there exists a simpler way to capture applicants’ preferences: cap the number of lotteries that each person can enter. A moderate restriction — say, to five entries per year — would shift the system to a better equilibrium by forcing applicants to prioritize, leaving units that they consider less desirable to those who value them more. This change, Arnosti and Shi show, would produce significantly better matches than the current approach.

Although there is no data on exactly how many tenant-apartment mismatches occur under existing conditions, anecdotal evidence suggests they do take place. Writing on a web forum devoted to New York City housing, for instance, one recent winner related her experience of submitting multiple bids and getting selected for a building in the Bronx. Though she was “not too crazy” about the surrounding industrial area and wary of relocating there from Queens, she signed the lease — despite receiving, and ultimately declining, an eleventh-hour callback for a different lottery in her own borough.

Few applicants are equally fortunate in drawing two winning tickets, but the strategy of applying everywhere makes perfect sense under the existing system, Arnosti and Shi write. A system encouraging all agents to pursue only their favorite buildings, would benefit everyone by reducing misallocation and somewhat tempering runaway demand. Barring that, checking every box on Housing Connect feels like the right thing to do.

The researchers’ recommendations are particularly timely given City Hall’s current push to increase the housing stock. If the challenge of construction in a famously crowded city can be tackled, why not a simple tweak to the lottery portal’s software? Says Arnosti, “We may not have enough affordable housing, but at least we can make sure we do a good job with the housing we do have.”

About the researcher

Nicholas Arnosti

Professor Arnosti teaches the MBA core class B6102 Operations Management. His research focuses on market design, with particular emphasis on online marketplaces (such as...

Read more.
articles by Topic