My book, Build or Die: How America Is Suffocating Its Cities and What to Do About It, will be published by Princeton University Press on December 8, 2026. Here’s what people are saying:
“Build or Die is destined to be a classic, a book that those who care about making our cities genuine engines of opportunity will continue to reach for in years to come.”—Steven Teles, professor of government and policy, Johns Hopkins University, and senior fellow, Niskanen Center
“This is a bracing diagnosis of the housing crisis from one of the best qualified housing journalists of our time.”—Jerusalem Demsas, editor-in-chief of The Argument
“In prose as vital and engaging as the cities he celebrates, Resnikoff offers a compelling guide to fixing our urban spaces—and our democracy.” — Yoni Appelbaum, deputy executive editor of The Atlantic and author of Stuck: How the Privileged and the Propertied Broke the Engine of American Opportunity
Last week, scholars with the Public Policy Institute of California made a bold claim in a new report: Prop 47, a 2014 ballot measure that downgraded some nonviolent crimes from felonies to misdemeanors, has “contributed to increases in unsheltered homelessness and drug use as well as a decline in drug treatment.” That finding has since been making the rounds in the state press, with big writeups from CalMatters and the New York Post’s recently established California bureau. (It should be noted that Prop 47 was partially rolled back by 2024’s Prop 36; it’s a bit early to say what the impact of this partial rollback would be.)
It’s not much of a mystery why the PPIC report is getting so much buzz. For years, various pundits on the right have insisted that California’s homelessness crisis is largely a result of the state’s permissive drug laws and relatively generous social safety net. That is one of the core arguments that animate Michael Shellenberger’s 2021 book San Fransicko, along with much of the agitprop that helped launch Christopher Rufo’s career in the right-wing think tank world. More recently, the Cicero Institute—a personal project of billionaire venture capitalist Joe Lonsdale—essentially dictated a homelessness strategy to the Trump administration that is based on similar principles. If homelessness is driven by housing scarcity, then it follows that we can largely solve homelessness by building more housing and moving people from the streets into stable homes; if homelessness is what happens when people of weak character realize that they can do drugs all day and mooch off the state instead of finding a job, then we can only solve homelessness by showing those people the whip.
Shellenberger, Rufo and Lonsdale are charlatans; you can read my response to San Fransicko here and to some of Lonsdale’s arguments here if you’re interested. But the team at PPIC is not made up of MAGA ideologues or serial fabulists. These are serious researchers doing serious work. And so, when those researchers make a claim that runs directly counter to my own work on homelessness, I can’t dismiss it out of hand.
So let’s dig into exactly what PPIC’s Magnus Lofstrom, Shannon McConville, and Sean Cremin found. (Full disclosure: Magnus was one of my econ professors in graduate school.)
Using biennial point-in-time (PIT) counts, PPIC compares the rate of homelessness in California to a composite rate pulled from various comparison states. They find the trend in California’s homelessness rate largely tracked the trend in those comparison states, up until the passage of Prop 47; afterward, they diverged fairly sharply.
Thus, in the words of Magnus and his co-authors, “We estimate that the reform led to a roughly 10 percent increase in California’s unsheltered homelessness rate or about 7,000 additional unsheltered people experiencing homelessness, accounting for approximately 20 percent of the growth in California’s unsheltered homeless population between 2015 and 2019.”
But there are a couple of caveats. The first is that the PIT count process was reformed in 2015 to better capture the US unsheltered population, right around the same time as Prop 47 implementation and the subsequent divergence in homelessness rates. “It is possible changes in how local agencies conducted the counts could have contributed to the increase in homelessness that we observe, especially for unsheltered counts after Prop 47,” Magnus and co-authors write. “As long as these efforts were no greater in California than in comparison states, our synthetic-approach estimates would not be affected. However, our within-California estimates would be overstated if these improvements to the count were broadly implemented across local agencies after Prop 47.”
The PIT count is a flawed metric in any number of ways, and one of its major flaws is that it tends to substantially undercount unsheltered populations. It’s a lot easier to take a census of people who are signed into shelters and assigned designated beds. Unsheltered people, in contrast, are notoriously difficult to track down. So a change in how PIT counts are conducted, if it leads to a more accurate count, could cause the baseline estimate to jump in areas with unusually large unsheltered populations—such as California, which, thanks to its temperate climate and inadequate shelter bed capacity, has by far the largest unsheltered population in the United States. (The PPIC report’s technical appendix doesn’t list the comparison states, but New York, for example, has a very high rate of homelessness, a much less forgiving winter season, and a legal right to shelter in its largest city.)
And there’s another important caveat. PIT counts exclude people who are hospitalized, incarcerated, or otherwise institutionalized; if you’re sitting in a jail cell, you are not considered homeless for the purposes of the count. So while it’s theoretically possible that a lot of Californians got into hard drugs once illicit drug use was no longer a felony, a substantial share of the PIT count increase may simply come from the fact that a lot of people who previously got sucked into the prison system, and therefore excluded from the PIT count, are no longer getting felony convictions.
The CalMatters writeup of the PPIC report acknowledges this, noting: “It’s unclear how much of that increase in homelessness is tied to increased drug use, and how much instead is because unhoused people who would have been sleeping in jail before Prop. 47 now are sleeping on the street instead.” But CalMatters adds that Magnus “suspects it’s a bit of both.”
But now we’re in the realm of supposition, because there’s nothing in the PPIC report that allows us to disaggregate the two effects. PPIC had previously reported that California’s jail and prison population fell by approximately 13,000 inmates within two years of Prop 47’s passage; in contrast, as mentioned above, the newer report attributes an increase in the unsheltered population of about 7,000 people to Prop 47. The most parsimonious explanation is that, prior to Prop 47, jails and prisons were a sort of shelter of last resort for many homeless people who were caught committing nonviolent offenses. That strikes me as considerably more plausible than the claim that criminal justice reform somehow encouraged homelessness. (And again, this is all assuming that the captured rise in the unsheltered population is 100% real, and not in part an artifact of the change to how PIT data is collected.)
As we’ve covered in this newsletter many times before, there doesn’t appear to be any meaningful correlation across states and metropolitan regions between drug use rates and rates of homelessness; West Virginia, for example, has one of the most pervasive drug problems in the country but a well below average rate of overall homelessness. That’s because housing is relatively cheap in West Virginia, and housing affordability remains the critical factor that determines homelessness rates.
At this point, someone might say: Who cares about the exact mechanism linking Prop 47 implementation to higher homelessness rates? If there are more people on the street as a result of Prop 47, for whatever reason, then that still demonstrates that we should start charging nonviolent drug users with felonies again.
To which I’d say, okay, fine. It is inarguable that moving someone from homelessness to prison removes them from the PIT count. But does that really count as a success? If imprisoning nonviolent drug offenders is the “solution” to their homelessness, that implies we could solve the entire homelessness problem in California by rounding up the unsheltered population and dumping it into concentration camps. That solution may suit the fine people of the Cicero Institute, but I’d be hard pressed to call it a solution; both because it’s morally appalling and because a Gulag First policy wouldn’t address the affordability problem that is pushing Californians into homelessness in the first place. I’m reminded a little of the early days of the COVID-19 pandemic, when President Trump said he wanted to prevent a cruise ship full of infected people from being transferred to onshore medical facilities, because putting those people on American soil would cause them to be counted in the country’s overall case rate.
There’s one more finding in the PPIC report that I want to highlight. Magnus et al. write: “We find that, in any given year, counties with higher rents and lower housing vacancies have significantly higher rates of homelessness. However, we do not find a statistically significant association between changes in local homelessness rates and changes in fair market rents or housing vacancy rates over the past two decades.” In other words, homelessness in, say, Los Angeles does not necessarily decline when rents in the city drop. Which would seem to suggest that, despite what I’ve been arguing in this post and for the past several years, making cities more affordable does not have an appreciable impact on their rates of homelessness. What gives?
To answer that question, I decided to take a look at fair market rents in the parts of California with the highest concentration of homelessness: San Francisco and Oakland, and Los Angeles. Here is a chart that Claude produced for me using HUD fair market rent (FMR) data on two-bedroom units, adjusted for inflation. (Note that the methodology for calculating FMR has changed over time.)
In Los Angeles, you don’t see a ton of fluctuation in fair market rents between 2000 and 2026; it’s more of a steady climb. But in Oakland and San Francisco, while the overall trend line points up, rents are pretty volatile. They also appear to be correlated with the business cycle, so that you see a Bay Area FMR drop a couple of years after the collapse of the first tech bubble, followed by a rapid spike after the second tech boom took off. And then another drop not long after the COVID-19 pandemic’s onset. FMR data tends to have a built-in lag of a couple of years, which explains the gap between these economic downturns and the decline in housing costs.
Given all of that, it shouldn’t come as a big surprise that these FMR declines don’t map onto homelessness declines. Yes, rents fell in the early 2010s and early 2020s, but those falls were the result of serious economic shocks; in both cases, a lot of people were being forced out of work simultaneously. If your rent falls 20% but your income completely zeroes out, then your housing has become less affordable even if it is objectively cheaper. This point is consistent with another key finding from the PPIC report: that “places that saw a larger decline in SSI benefit [federal cash assistance] caseloads over time also saw a larger increase in their homeless population.”
The sort of rent decreases we need in order to reduce homelessness are not the periodic drops that occur as a result of economic crises. California’s housing deficit has accumulated over decades, beginning with the downzonings of the 1960s; as a result, the general trajectory for housing costs has been upward, despite some recession-induced fluctuations. We shouldn’t expect those fluctuations to yield sustained reductions in homelessness; those sustained reductions will only come once California achieves a stable growth rate in housing supply and a commensurately steady decline in rents.
Which brings us back to the main takeaway regarding Prop 47: for homelessness to decline, homeless people need to have somewhere other than shelters or the streets where they can lay their heads. Instead of a jail cell, that should be a permanent home.