How Sampling Error Created America's Housing Crisis

Community representatives are often anything but

In previous posts on Covid-19 [Standard Errors] and homelessness [Standard Errors] in Los Angeles, I have pointed to the original sin of Southern California policy, the issue from which nearly all others flow: the incredibly high cost of housing. The reasons California housing is so expensive is quite simple: there are not enough houses for people who want them. The reason behind it is well-known to observers of the state, which is that local politics has imposed a hammerlock on the construction of new housing in urban California for decades. As the population has grown and demand has grown, the limited supply has squeezed prices ever-higher.

What is less-well-understood is that while housing construction is anathema to local governments, housing construction is generally quite popular. I like to think of this contradiction as fundamentally a flaw driven by statistical sampling practices. The processes local governments have set up to gather community input end up selecting for outlier opinions that do not reflect the community. The process ends up as a form of unrepresentative democracy.

---

Local activists oppose new homes

My graduate school classmate Weihuang Wong [Twitter] opens a great working paper on local housing politics [SSRN] with a shocking and yet completely typical and unremarkable anecdote that hit close to home (literally):

On February 4, 2014, the city council in Santa Monica, California, approved by a narrow margin a proposal to redevelop a shuttered pen factory into a mixed-use complex with homes, shops, restaurants, and offices. The factory had been closed since 2005 and Santa Monica, a seaside town at the heart of Southern California’s “Silicon Beach” west of Los Angeles, was experiencing a real estate boom. Hines, the developer that acquired the site in 2007, had planned to build more than 400 homes and 400,000 square feet of commercial space, across the street from a planned light-rail transit stop. The proposal was the culmination of a process spanning four years and numerous public hearings. Three months later, in the face of a referendum on the project, the city council reversed its decision. By the spring of 2015, Hines had sold the property to new owners, who subsequently redeveloped the site as a creative office complex with no residential component.

The referendum battle was fought by Santa Monicans for Renter’s Rights, which complained that the project did not offer sufficient affordable housing. Instead, SMRR won and zero units of affordable housing were built. The prospect of similar protracted battles looms large in the face of anyone considering building housing in California. Conor Dougherty detailed the incredible story of the battle over an apartment building in Lafayette [NYT] - in compliance with zoning - which has been ongoing since 2012 and took out the city manager as collateral damage.

---

Neighborhood Defenders

There is a popular sentiment that local control is closest to the people and will be most representative of local interest. An emerging strand of research on local politics, especially with regards to land use decisions, is centered on the empirical reality that local politics is often less representative of local interests than state or federal politics. Instead, it is often dominated by local busybodies who have the combination of vested interests and free time to show up at local city council meetings and complain.

These local busybodies are less-judgmentally known as “neighborhood defenders”, after the 2019 political science book [Amazon]. The authors find that the adoption of practices centered around local community input and control tend to be dominated by small groups of residents. These residents tend to be motivated to participate by a high degree of attachment to the status quo, and coordinate effectively to block new housing or indeed policy change of any sort. When the local government opens up a public hearing on new development to community input - at 8 PM on a school night - these neighborhood defenders are the community who show up to give their input. Even when the community would benefit as a whole from new housing, the fact that local government is dominated by “who shows up” often means that the more local a government the more it can be swayed by idiosyncratic individuals.

Parks and Rec is the best depiction of local government in practice

The method by which local governments solicit public input on housing policy could charitably be described as a “convenience sample” [Wikipedia], which is exactly what it sounds like. I’d go even further to say that it represents an adverse selection [Wikipedia] problem, in that it is accidentally designed to draw input from people who do not accurately represent the public’s opinions. (Who shows up to a public meeting to comment on housing development? Not a normal person). This is why being seen as development-friendly is political death in most local governments, even though a large majority of the public supports building more housing in their community [Cato].

Ultimately, you can consider the near-lockdown on home construction in most of America’s cities as a function of terrible sampling practices. Land use is one of the most important decisions a community can make, so it is only right that decisions are made with community interests and input in mind. However, the mechanism that most local governments have settled on to sample the community to assess their interests is…really bad. It has led to policy outcomes that diverge further and further from actual public opinion.

It can never be repeated enough, but whenever deciding to gather information or input it always pays to ask how you are sampling that information or input. Casual assumptions that the data will be representative can easily lead to disaster.