Downzoning Chicago: How Local Land Use Policy Has Reduced Housing Construction and Reinforced Segregation

1. Questions

Localities use land-use policy, including zoning, to shape the built environment. With elected officials worried about rising housing costs, some cities have recently “upzoned” to accommodate construction (Freemark 2020; Pendall, Lo, and Wegmann 2022). But postwar US local development policy was dominated by “downzoning”—meaning altering land-use controls in a way that limits the ability to build new structures, particularly higher-density apartments (Morrow 2013; Whittemore 2012). Homeowners who sought to protect neighborhood aesthetics and property values supported these downzonings (Been, Madar, and McDonnell 2014; Fischel 2005), as did renters who worried that new development could increase housing costs (Hankinson 2018). The dominance of residents who were wealthy, white, and homeowners in planning processes likely contributed to these policy choices (Einstein, Palmer, and Glick 2019; Lo and Freemark 2022).

Research examining citywide zoning changes indicates that downzoning is associated with increased housing costs and less building (Stacy et al. 2023). This suggests that neighborhood-level downzonings could contribute to the underprovision of housing. To explore the conditions that affect downzoning and downzoning’s neighborhood-scale consequences, in this paper we examine tract-level data for the city of Chicago to answer:

• What was the incidence of downzoning? In what sorts of neighborhoods did it occur?

• What were downzoning’s effects in terms of housing availability, housing costs, and neighborhood demographics?

2. Methods

We leverage a parcel-level dataset of downzonings across Chicago between 1970 and 2016; these data pinpoint the decade when downzonings occurred. We classify historic landmarking as a form of downzoning, as it may be even more onerous for developers than downzoning—it “freezes” assets, providing little opportunity for redevelopment, while downzoning typically allows at least some reconstruction.^[1] This research builds on a descriptive study of zoning policy by Chicago Area Fair Housing Alliance (2018). Chicago experienced dozens of neighborhood- or parcel-level downzonings beginning in the 1970s, often responding to resident concerns about overdevelopment or neighborhood decline (Schwieterman and Caspall 2006). We combine these data with tract-level, Census-produced housing and demographic data.^[2]

Not every community reacts similarly to equivalent zoning changes. A downzoning might have more damaging effects on housing availability and affordability in high-wealth neighborhoods—where there is demand for construction—than in low-wealth areas (Freemark 2023). This factor may be particularly relevant in post-1970s Chicago, where many buildings were abandoned across the increasingly impoverished South and West Sides. Using citywide data may violate the assumption of parallel pre-trends, important for establishing comparability between “treated” (downzoned) and “untreated” (non-downzoned) tracts.

We calculated the share of each tract’s land area that was downzoned, by decade.^[3] We then evaluated tract-level housing supply changes between 1960 and 1970, before mass downzonings began; a t-test of means shows a statistically significant difference in housing supply growth between tracts citywide that experienced any downzonings in the following two decades and those that did not (supplementary table S1). To create samples that enable a fair comparison between treated and untreated areas in our fixed-effects regressions, we thus subsetted the citywide data into two study groups whose later downzoned and not-downzoned tracts exhibited no significant difference in their housing supply growth pre-trends: (a) tracts that had a greater-than-median white population share and greater-than-median housing values in 1970; and (b) tracts located downtown or on the North Side (figure 1).^[4] We examine these areas using a panel dataset. For each tract and decade, we identify both the share of the tract that was cumulatively downzoned during previous decades and during the same decade, because we hypothesize that both could influence outcomes.

Figure 1.Map of Chicago, showing downzoned areas and study groups

Source: The authors, based on data from US Census, 1970, and George Kisiel, 2024.
Note: The two study groups have some overlap, primarily in the far northern portion of the city. But a majority of tracts in each group is only in that group.

The choice of these study groups reflects that Chicago neighborhoods experienced contrasting postwar trajectories. Tracts with high white population shares and high housing costs in 1970, as well as those on the North Side and downtown, gained housing value in the following decade; those elsewhere declined in value (supplementary figure S1). We thus focus on these study groups, since they feature a variety of “treated” and “untreated” tracts. Moreover, given that zoning controls are most relevant in high-demand neighborhoods, these study groups best reflect where these regulations might matter.

We use OLS regressions to estimate associations between tract-level characteristics and a tract’s likelihood of being downzoned, by decade, between 1970 and 2016. We then use tract- and time-level fixed-effect models (sometimes with time trends) to estimate downzoning’s effects on housing availability, share white population, housing values, and rents.

3. Findings

We examine the incidence of downzoning by regressing several tract-level characteristics on a binary dependent variable: whether at least 10 percent of a tract’s land area was downzoned in a decade (table 1).^[5] We find the city was more likely to downzone areas in tracts that had higher housing values (this relationship was statistically significant for three decades); this suggests that higher-income residents and their elected representatives successfully leveraged downzoning to maintain the status quo. We also find that lower homeownership rates were associated with downzoning, contrary to previous research (Been, Madar, and McDonnell 2014); this may result from the fact that downzonings largely did not impact the city’s edges (figure 1), which were predominately zoned for single-family homes and not easily downzoned.

Next, to descriptively understand the relationships between our variables, we compare changes between 1990 and 2010 among tracts that experienced a previous downzoning (between 1970 and 1990), and those that did not. In both study groups, downzoned tracts had substantially lower housing supply growth (table 2). On average, tracts on the North Side or downtown that were downzoned between 1970 and 1990 gained only 65 units between 1990 and 2010, while non-downzoned tracts gained 470 units on average, seven times as much.^[6]

The comparison in table 2 also shows that the median downzoned tract had slower loss in its share population white than non-downzoned tracts. We identify lower rent increases in downzoned areas, an issue to which we return below, and uneven trends in terms of housing values.

To confirm these outcomes, we estimate downzoning’s supply impacts using fixed-effects regressions that leverage our panel data (table 3). Among all tracts, downzoning was associated with increasing housing availability (models I–II), but this likely results from divergent preexisting conditions, as noted. Examining just the study groups, on the other hand, we find consistently negative housing supply effects associated with higher levels of downzoning (models III–VI). These effects occur both as a result of previous-decades downzonings and also same-decade downzonings (effects are generally larger from downzonings that occurred further in the past). Outcomes manifest whether or not we include time trends.^[7] This provides evidence that downzoning substantially reduced housing supply.

Finally, we examine downzoning’s impacts on other neighborhood characteristics (table 4). Downzonings were associated with higher white population shares and higher housing values (models I–II and IV–V, with varying statistical significance and effect magnitude), outcomes that suggest that downzoning reinforced segregation and concentrated wealth. We hypothesize that downzoning has been associated with increased investment in, and competition for, ownership units, resulting in higher home prices.

We also find that downzoning is associated with lower rents (table 4, models III and VI). Downzoning displaces housing construction into other areas that were not downzoned; we find a significantly lower share of rental units built in recent decades among tracts that were downzoned (supplementary table S2). On average, tracts that experienced any downzoning between 1970 and 2000 lost about 160 rental units over that period, while those that were not downzoned gained 25 or more rental units on average. As new rental units typically charge higher rents (Damiano and Frenier 2020), we hypothesize that the lower rents in downzoned areas are a product of declining typical rental unit quality, though further research is needed to substantiate this claim.

This study offers further evidence for downzoning’s impacts, building on previous citywide analysis. Neighborhood-scale downzonings are associated with less housing supply, increasing home values, and concentration of white residents in the most sought-after neighborhoods. Cities seeking to increase access to housing and improve neighborhood diversity should avoid downzoning.

Acknowledgements

We thank the Metropolitan Planning Council and the Chicago Community Trust for their support funding this research. We thank Lydia Lo for her collaboration on our broader work on Chicago’s zoning policies. We thank the anonymous reviewers for their feedback on drafts of this manuscript.