Loading Chicago tract data…
An exploration of the relationship between bike lane infrastructure and median rent at the census tract level across Cook County, Illinois (2012 to 2022).
The null result is reassuring: cities can invest in bike infrastructure without directly pricing out residents. But the more important message is that bike lane investment has been flowing to already-privileged neighborhoods. The equity problem isn't displacement; it's that the people who most need transportation alternatives aren't getting the infrastructure.
The dominant driver of rent change is income growth and demographic shift. If you're worried about displacement, bike lanes are not the lever to pull, but they are a signal of the broader gentrification process happening in a neighborhood.
Chicago's bike lane network has grown substantially over the past decade, but that growth has not been evenly distributed. The pre-2016 network was concentrated on the North Side and along a handful of major arterials. Subsequent expansions pushed into more neighborhoods, though as the analysis below shows, infrastructure investment has tended to follow areas that were already experiencing economic change.
Rent increased across nearly all of Cook County between 2012 and 2022, but the geography is uneven. The highest rents are concentrated in a band running through the North Side, Near North, and parts of the Near West Side, the same areas with the densest bike lane coverage. Toggle between years to watch rent levels climb, and switch to bike lane km to see whether the spatial overlap holds.
Rent grew by a median of $322 per tract over the decade. The largest increases cluster on the North and Northwest sides, the same areas where bike lane additions were most concentrated. This geographic overlap is visually striking, but the regression analysis below suggests it reflects shared neighborhood characteristics rather than a direct causal link. Toggle between the two maps and compare the patterns.
Tracts with bike lanes saw higher median rents throughout the period, but they started higher too. The gap between bike lane and no-bike-lane tracts reflects pre-existing neighborhood differences more than infrastructure-driven change. Switch to income quartile grouping to see how strongly baseline wealth predicts rent trajectory, independent of bike lanes entirely.
Lines show median rent across tracts in each group across the three survey years.
Each dot is a census tract. The raw bivariate relationship suggests a positive association, but it explains less than 0.4% of the variation in rent change. Once income growth, demographic shift, and population density are controlled for in the full model, the bike lane coefficient shrinks from $22/km to $5.71/km and becomes statistically indistinguishable from zero. The pattern on this chart is largely driven by confounding, not by bike lanes themselves.
Color encodes baseline 2012 rent. Hover over dots for tract details. Red solid line = controlled model slope ($5.71/km, p = 0.55, n.s.). Gray dashed line = raw bivariate slope ($22.25/km, p = 0.03) before controlling for confounders.
A directed acyclic graph showing the assumed causal relationships in the model. Arrow thickness on the controls to rent path reflects that confounders are strongly significant. The bike lane to rent arrow is dashed because the coefficient (β = 5.71) is not statistically significant at p = 0.55, consistent with the cross-sectional null result. The endogeneity path (controls to bike lanes) represents the placement concern that wealthier neighborhoods attract both bike infrastructure and higher rents, addressed here by the long-difference design.