The study validates what has long been a rule of thumb among real estate professionals—the idea that generally speaking, people shouldn’t spend more than a third of their income on rent. That idea is based on the assumption that wages and salaries go up at roughly the same rate as rents—but in many places, that is no longer the case.

Median rent in the U.S. has risen 11% in the past five years, just as homelessness rates have also continued to climb. The study, conducted by (Z) Zillow Research, is an attempt to quantify the size, as well as identify some of the root causes, of the nation’s homelessness crisis.

The study examines federal statistics, such as the U.S. Department of Housing and Urban Development (HUD) estimate that more than 546,000 Americans experienced homelessness in 2017. But a Zillow researcher and several of his colleagues put those figures higher at just over 660,000, believing that the scale of the nationwide homelessness epidemic has been estimated 20% lower than it actually is.

The study identified two rent burden thresholds—22% and 32%—which can affect homelessness in a particular community. When rent affordability rises more than 22%, more people in a community experience homelessness. When the rent burden rises more than 32%, the overall rate of homelessness is likely to rise much faster in a particular area. To put it simply, even a modest rent increase can push many people to the brink, and very quickly, those individual hardships can add up to a community-wide crisis. The clusters of people most at risk for becoming homeless live in Boston, New York, Los Angeles, and Seattle, cities with a combined 15% of the U.S. population.

This type of data can be useful to local legislators and policymakers tasked with expanding and funding programs that offer assistance to residents struggling with a disproportionate rent burden. The rent burden already exceeds the 32% threshold in 100 of the 386 markets included in the study.