Mistaking motion for progress in urban policy

Why policy conversations around housing and transit keep getting stuck on the wrong metrics.

A recent report on—stay with me here—the U.S. military campaign against the Houthis contains a detail that feels familiar to anyone who's worked in policy, in almost any field. It’s summed up in this quote:

“The White House began pressing Central Command for metrics of success in the campaign. The command responded by providing data showing the number of munitions dropped.”

That response sounds... nonresponsive. But the issue isn’t that Central Command misunderstood the question. It’s that the real question—is the campaign working?—is very hard to answer on a short timeline. So instead, we reach for what’s easy to count.

This is a familiar move in policy evaluation: when outcomes are ambiguous or hard to quantify, we fall back on inputs—bombs dropped, dollars spent, hours worked—and treat those as stand-ins for success. This isn’t limited to government; you see it in corporate earning calls discussing their use of AI (how much are they spending on Nvidia chips?) and remote work. It’s not a conspiracy or a trick. It’s a coping mechanism for measurement uncertainty and it’s not uncommon in policy.¹ But it has consequences.

The numbers are saying: We tried really hard. That might be enough to avoid blame for tactical decisions, but it’s a poor way to set strategy or judge effectiveness.

This substitution of effort for results shows up all over policy—especially where outcomes are diffuse, lagging, or just hard to track. In this post, I want to explore two such areas—housing and transit—where this dynamic is especially counterproductive.

🏘 I. Housing: Taking the homothetic nature of demand seriously

Economists often describe housing demand as homothetic. In plain English: as people get wealthier, they tend to spend more on housing—in rough proportion to their income. NYU finance professor Arpit Gupta (who has an excellent Substack) explains:

Not all goods behave this way. As income rises, the share of income spent on food typically declines (as shown in the BLS data noted in the quoted tweet). But with housing, the more people earn, the more they spend—for homes that are bigger, better located, and more elaborately outfitted.

This insight matters when evaluating policies aimed at increasing supply—particularly accessory dwelling units (ADUs). Over the past several years, many states have liberalized ADU rules. These changes make it easier to build nominally separate housing units that go by the names of backyard cottages, apartments over the garage, granny flats, alley (or laneway, in Commonwealth countries) houses, and so forth, sited on the same land as the primary structure. I support this trend. ADUs have earned bipartisan attention and think-tank support. This Mercatus Center report notes a wave of pro-ADU state laws, and New America has report describing ADUs as “a major part of the housing economy in the 21st [century].”

But there’s a subtle sleight of hand in how ADUs are discussed. Consider this line from the California Department of Housing and Community Development's January 2025 handbook:

“In 2023, ADUs comprised more than 21 percent of all homes permitted statewide.”

Are these “homes” really functioning as additional dwellings?

Often, the answer is no. A large share of ADUs are better understood as additions—not legally, but functionally. They’re used to expand space for current residents rather than house new ones. Detached or not, many operate more like private expansions—consumption—than as rentals, or true housing production. Though often they are in a separate detached structure in the backyard, they are often the functional equivalent of adding a bedroom or family room to an existing structure.

The output we care about—new housing units occupied by distinct households—is hard to measure. So instead, we count ADUs permitted or constructed. That’s an input.

Some surveys, like a 2017 report from the Urban Land Institute (together with partners), have tried to track ADU use. But as a look at the screenshot below shows, the data is fuzzy: it is not clear what is captured by the categories “not used for anything” and “other,” I don’t have high confidence that survey respondents understood that “primary residence (occupied)” actually means the primary residence of someone other than the homeowner (as opposed to part of the owner’s primary residence), and it’s not clear which are current versus intended uses and how stable either of those is. If you add up all those percentages you get a big number. In other words, the general problems of survey data attach to surveys about ADU use as well.

Add in the time gap—2017 is practically a lifetime ago in housing economics—and the utility of the data declines further.

Post-pandemic, demand for private space has increased, materials and labor costs have surged, and interest rates have more than doubled. Anecdotally, new ADUs—increasingly expensive to build for these fundamental reasons—are rarely rented out. Theoretically, their modular nature—the fact that they normally have separate entrances—could allow them to be used as a platform for an arms-length rental arrangement down the line. That latent flexibility has real social value. But in the short term, many ADUs—especially new ADUs—function more like private amenities than policy solutions.

Don’t get me wrong, I’m still in favor of ADUs. They unlock non-economic (as well as economic) forms of value, like intergenerational or non-traditional living arrangements, in the otherwise fairly inflexible structure we know as the single-family detached house. But let’s be real: in most cases, these aren’t really moving the needle on the housing crunch. They’re a salutary innovation, but mostly an orthogonal one when it comes to housing supply.

Some cities stretch the definition of ADUs for strategic reasons. San Diego, for example, has used legal workarounds that enable full apartment buildings to be built under ADU rules. Creativity in service of laudable policy goals is laudable. My guess is that this is unlikely to be scalable or sustainable (people will figure it out), though stranger things have happened. Nevertheless, to the extent ADUs are adding supply because they are a backdoor to build apartment buildings, I think it’s fair to say that their success is no longer the success of ADUs as such. We are talking about apartment buildings.

In any event, we should not confuse ourselves by considering new ADUs as equivalent, or even directionally aligned, with housing production. Except where they’re being loopholed into apartment buildings, the modal new ADU is probably better understood as more housing consumption for the owner, not more housing production for renters.

🚇 II. Transit: Safety is not just the absence of reported crime

The second theme is safety in public transit—and what we mean when we say a system is “safe.”

In a recent interview with David Roberts, transit consultant Jarrett Walker challenged the idea that transit systems are especially dangerous. He cited data showing that reported crime on transit is relatively low. And that’s important: bringing evidence into a conversation dominated by vibes is a welcome move.

But there’s a catch: reported crime doesn’t capture the full picture.

It misses unreported incidents and, crucially, the experiences of people who avoid transit entirely due to, among other factors, fear. Measuring a concept like safety by using a proxy like reported crimes is like assessing walkability by counting pedestrian injuries—technically informative, but very incomplete. If I tell you a given stretch of road features a low number of pedestrian injuries, I could be describing an urbanist paradise or an interstate highway. The data point has limited information value because it’s not measuring what we want to know.²

Walker acknowledges that transit won’t work for everyone. Some people just won’t feel comfortable. That’s pragmatic. But it raises a tough question:

Should we accept a system that feels unsafe to large swaths of users—especially women and other vulnerable groups?

No one would openly advocate for a transit system designed exclusively for able-bodied men aged 16 to 54. But would we tolerate it de facto?

Women comprise a slightly disproportionately large share of transit riders, so in one sense it’s inaccurate to say the system excludes them. Still, in many systems, the comfort and safety of women is de-prioritized. That’s a massive mistake—not just from an equity standpoint but from a practical one. A system that feels safe to everyone (or nearly everyone) will attract a much larger share of the population. Including women: how many more women would ride transit if it felt safer? Women earn less than men, have less wealth, and live longer; for these reasons alone, in expectation there should be more women on transit than men, so the fact that there are doesn’t tell us much. This is the kind of question—how many more women could be riding transit if transit felt safe—that looking at reported crime rates doesn’t give you much purchase on.

This is a moral and equity concern—but it is also a practical one for all potential users and beneficiaries of transit systems, which is to say society at large. A system that feels safe and welcoming to everyone will attract more riders. That’s essential for achieving a host of goals, including mitigating greenhouse gas emissions and other negative externalities of the car as well as the fiscal health of transit systems.

So how do we make transit feel safe?

Assume a can opener

Some advocates argue that frequency solves everything: more service → more riders → more safety. Walker makes this pitch. There’s some truth to it. It’s the “eyes on the vehicle” equivalent of Jane Jacobs’ “eyes on the street.”

Politicians frequently turn to law enforcement, which has a role. But in addition to the use-of-force risks of amping up police presence in transit systems, in transit it is also expensive and widely viewed as ineffective as a general solution.

You make systems safe by adding riders, and you add riders by running more frequent service, but how do you get people to ride in the first place?

This Catch-22 has another layer: safety matters for operations, too. Transit agencies are struggling to hire drivers, as Walker observes. A leading reason is safety concerns. Operators don’t want to deal with unruly or threatening passengers. It is they, after all, who are tasked with maintaining order on buses and ensuring passengers pay their fare. You could try to infer this from the documented increase in crimes against transit operators—which are probably more reliably reported than crimes in transit in general—but you would also want to talk to operators and would-be operators. Anecdotally, the post-pandemic decline in social behavior—including but not limited to actual crimes—is a major reason that the appeal of the job has declined.

So, even if transit enjoyed the more generous fiscal support that is transit advocates’ preferred solution, labor shortages—driven in part by safety fears and in part by macroeconomic trends—would constrain expansion. Of course, that support is not generally forthcoming (though as Walker notes local referenda provide some bright spots). In fact, nearly every major transit system is facing a fiscal cliff right now with the expiration of federal pandemic funds.³

So, how to solve this? Well, the “frequency” answer is essentially a call to assume a can opener: if you start from a premise of well-funded and competently administered frequent service, it’s easy to reason your way to a virtuous circle. More seriously, safety cannot be treated as secondary or optional. It must be embedded in system design and operations. Capital improvements—like fare gates, lighting, and staff presence—can make a meaningful difference. They’re not silver bullets, but deployed in concert with improved service they should expand the market for transit riders.

It helps to treat transit as a mass market public service rather than a niche product. The addressable market for transit isn’t every single person, but it includes people who currently think transit is unsafe but are open to taking it. In the persistent shortage of not just passengers but operators we see individuals picking up on something advocates and scholars of transit are sometimes reluctant to acknowledge: unease. Surveys tend to confirm that drug use and antisocial behavior—far more prevalent on many transit systems today than pre-pandemic—is driving people away. This may to some extent show up in reported crime stats (which by the way are up significantly in many systems on a per-rider basis vs. pre-covid), but the two are hardly coterminous.

🧭 The bottom line: a relentless focus on goals and outcomes

For those focused on policy analysis and applied research (as opposed to, say, political organizing), calls for ADUs and transit frequency as solutions to the housing shortage and safety, respectively, suggest a common limitation: the need to acknowledge real-world constraints.

Perhaps we rely on inputs and proxies not only because they’re easier to count but because they let us tell a story we want to hear. But new ADUs bear only a loose relationship to new housing units and reported crimes provide a coarse measure of safety or comfort. To some extent, these measures are not only incomplete but irrelevant. People who care about cities should care about making the substance of policy work. With cities facing the greatest challenges they have faced in generations, this may be more important than at any other point in our lifetime.

1. Together with co-author Jonathan Levine, I discuss this confusion in the context of transportation policy in this book chapter.

2. For this reason, Peter Furth of Northeastern has developed a capacious measurement for roads of bicycling suitability called Level of Stress. Notably, it goes beyond injuries and fatalities.

3. Smaller transit systems face fiscal challenges too, but they never relied on fares for much of their funding to begin with and so did not rely on federal funding to replace that stream. Systems like the CTA in Chicago did, and now that funding is gone.