Land-use and socioeconomic time-series reveal legacy of redlining on present-day gentrification within a growing United States city

Peter C. Ibsen, Anna Bierbrauer, Lucila M. Corro, Zachary H. Ancona, Mark A. Drummond, Kenneth J. Bagstad, Jay E. Diffendorfer•View original

OverviewBalancedshimmer voice

In the late 1930s, a federal agency called the Home Owners' Loan Corporation drew color-coded maps of American cities. Each neighborhood received a grade from A to D, ranging from "desirable" to "hazardous." These grades determined who could get a mortgage. In Denver, Colorado, the spatial logic embedded in those maps is still shaping who lives where today. That's the central finding of a study by Peter Ibsen and colleagues, who used a high-resolution land-use time series running from 1937 to 2018, paired with eight decades of U.S. Census data, to track what actually happened inside those Home Owners' Loan Corporation boundaries over time. Denver serves as a useful lens because it's one of the fastest-growing cities in the western United States, with an annual growth rate of 1.48 percent and a population now approaching 715,000. This growth means there's significant pressure on neighborhoods, and that pressure reveals fractures. The first thing the team found is that the divergence between Home Owners' Loan Corporation grades didn't start with the maps. The maps captured something already in motion. Areas graded A and B were already more dominated by single-family housing in 1937, while C and D areas showed more mixed land use, including commercial parcels, light industrial lots, and multifamily buildings. The bureaucrats drawing those maps weren't inventing inequality; they were encoding it. What happened next is where it gets damning. Over the following eighty years, the divergence deepened. In grade A neighborhoods, single-family residential cover rose from 68 percent in 1937 to 93 percent by 1997, creating a locked-in landscape. In grade D neighborhoods, single-family housing started at just 49 percent and actually declined slightly to 45 percent, while multifamily housing, light industrial use, and retail filled in around it. To measure this mix, Ibsen and colleagues used the Shannon-Wiener diversity index, a metric borrowed from ecology that captures how many different uses share a space and how evenly they're distributed. Higher scores indicate more variety. Grade D increased in land-use diversity every twenty years, while grade A decreased. By 1977, the gap was statistically unambiguous, with the Home Owners' Loan Corporation category showing a highly significant effect on diversity across every decade analyzed. This is not just a story about zoning maps; it's a story about who got to accumulate wealth and who didn't. The demographic data follows the same trajectory. The proportion of non-White residents rose significantly with both time and Home Owners' Loan Corporation category, and the gap between A and D neighborhoods widened across the study period. Grade D had significantly higher proportions of non-White residents than grades A and B from 1980 onward. Educational attainment tracked the same gradient; grade D consistently lagged grade A in college-educated residents for the entire study period, showing a highly significant category-by-year interaction in the statistical models. Income diverged too. By 1980, grade D had significantly higher proportions of low-income households, and those proportions stayed elevated through 2000. The authors are direct about what drove this: redlining was a key mechanism of divergence, as were the urban renewal policies that followed in the mid to late twentieth century. Denver's Urban Renewal Authority began operating in 1958, and those projects, concentrated in C and D areas, compounded the original sorting. It wasn't one policy; it was policy layered on policy, decade after decade, each reinforcing the spatial logic of the one before. Then comes the twist. The same features that made grade D neighborhoods sites of disinvestment — low home values, mixed land use, diverse zoning histories — turned them into targets for gentrification in the 2000s and 2010s. Ibsen and colleagues define gentrification eligibility precisely. A census tract was eligible if, in the year 2000, both its median household income and median home value fell in the bottom 40th percentile of the metro area. That's the statistical fingerprint of historical neglect. A tract was then classified as actively gentrified if, by 2019, its inflation-adjusted home value had increased, its median home value ranked in the top one-third of metro tracts, and its growth in college-educated adults also ranked in the top one-third. All three conditions had to be met. The results are stark. One hundred percent of grade A areas were ineligible for gentrification between 2000 and 2019 — they were already too wealthy, too stable, and too locked in. Grade D areas contained significantly more gentrification-eligible land than any other category. Within that eligible land, the changes were dramatic. The fraction of college-educated residents in grade D rose by 147 percent between 2000 and 2019. High-income households increased by 58 percent. Residential land cover jumped from 19 percent to 40 percent when comparing 1997 to 2018 data, and critically, the increase in residential land use was positively correlated with increases in high-income households only in grade D neighborhoods. This relationship didn't appear in grades A or B; it was specific to the historically hazardous zones. The authors put it plainly: areas that had been marginalized for decades were able to accept these demographic changes due to their diverse and changing land-use patterns. That sentence deserves a moment. The land-use flexibility that resulted from disinvestment — the mixed lots, the light industrial parcels, and the absence of protective single-family zoning — became the economic opening that capital moved through. The wound became the entry point. So, what does any of this mean for policymakers? Ibsen and colleagues are careful here. They frame the study as a diagnostic, not a prescription. But the diagnostic is genuinely useful. The gentrification eligibility framework they build, combining Census-derived income and home value thresholds with the fixed spatial boundaries of Home Owners' Loan Corporation zones, can function as an early-warning system for displacement risk. Mapping where eligible tracts cluster and then watching for the signs of active gentrification gives planners a lead time they wouldn't otherwise have. The deeper insight is about timescales. Land-use diversity differences were measurable as early as 1937. Racial and income disparities became statistically significant by 1980. Gentrification pressures emerged between 2000 and 2019. Each phase of the story took decades. That means one-off policy interventions, like a single rezoning decision or a single affordability program, are unlikely to reverse trajectories that compounded over a century. The spatial logic of redlining is stubborn, and countering it requires equally sustained attention. The authors are also honest about what their study can't do. The high-resolution land-use dataset they used isn't available for most cities, which limits direct replication. They couldn't access parcel-level records of Home Owners' Loan Corporation loan distributions in Denver, so they can't trace the financial mechanism directly. Their gentrification measure has a 2000 baseline, which means any displacement that happened before that year renders a tract ineligible in their framework, potentially undercounting the true extent of change. Denver's specific drivers — including the particular history of its Urban Renewal Authority, its downtown amenity growth, and its regional migration patterns — shape local outcomes in ways that don't automatically transfer elsewhere. What does transfer is the framework itself: pairing a multi-decadal land-use time series with standardized Census metrics over fixed neighborhood boundaries gives a spatial picture of vulnerability that no single snapshot could provide. The Home Owners' Loan Corporation zones are convenient because they're historically meaningful and already digitized for most major American cities. However, the logic works with any consistent boundary system. The point is to look back far enough to see how the present was made. Gentrification is often treated as a market event — capital flows in, prices rise, and people get displaced. What Ibsen and colleagues show is that gentrification is also a long historical process, one whose conditions were assembled over decades through government maps, mortgage policy, urban renewal, and zoning decisions. The neighborhoods gentrifying fastest in Denver today are the ones that absorbed the accumulated costs of those decisions. Understanding that doesn't resolve the policy dilemma, but it does change the terms of the conversation. You can't address where a city is going without reckoning with how it was built.

That's the central finding of a study by Peter Ibsen and colleagues, who used a high-resolution land-use time series running from 1937 to 2018, paired with eight decades of U.S. Census data, to track what actually happened inside those Home Owners' Loan Corporation boundaries over time. Denver serves as a useful lens because it's one of the fastest-growing cities in the western United States, with an annual growth rate of 1.48 percent and a population now approaching 715,000.

This growth means there's significant pressure on neighborhoods, and that pressure reveals fractures.

The first thing the team found is that the divergence between Home Owners' Loan Corporation grades didn't start with the maps. The maps captured something already in motion. Areas graded A and B were already more dominated by single-family housing in 1937, while C and D areas showed more mixed land use, including commercial parcels, light industrial lots, and multifamily buildings.

The bureaucrats drawing those maps weren't inventing inequality; they were encoding it.

What happened next is where it gets damning. Over the following eighty years, the divergence deepened. In grade A neighborhoods, single-family residential cover rose from 68 percent in 1937 to 93 percent by 1997, creating a locked-in landscape.

In grade D neighborhoods, single-family housing started at just 49 percent and actually declined slightly to 45 percent, while multifamily housing, light industrial use, and retail filled in around it. To measure this mix, Ibsen and colleagues used the Shannon-Wiener diversity index, a metric borrowed from ecology that captures how many different uses share a space and how evenly they're distributed. Higher scores indicate more variety.

Grade D increased in land-use diversity every twenty years, while grade A decreased. By 1977, the gap was statistically unambiguous, with the Home Owners' Loan Corporation category showing a highly significant effect on diversity across every decade analyzed.

This is not just a story about zoning maps; it's a story about who got to accumulate wealth and who didn't. The demographic data follows the same trajectory. The proportion of non-White residents rose significantly with both time and Home Owners' Loan Corporation category, and the gap between A and D neighborhoods widened across the study period.

Grade D had significantly higher proportions of non-White residents than grades A and B from 1980 onward. Educational attainment tracked the same gradient; grade D consistently lagged grade A in college-educated residents for the entire study period, showing a highly significant category-by-year interaction in the statistical models.

Income diverged too. By 1980, grade D had significantly higher proportions of low-income households, and those proportions stayed elevated through 2000. The authors are direct about what drove this: redlining was a key mechanism of divergence, as were the urban renewal policies that followed in the mid to late twentieth century.

Denver's Urban Renewal Authority began operating in 1958, and those projects, concentrated in C and D areas, compounded the original sorting. It wasn't one policy; it was policy layered on policy, decade after decade, each reinforcing the spatial logic of the one before.

Then comes the twist. The same features that made grade D neighborhoods sites of disinvestment — low home values, mixed land use, diverse zoning histories — turned them into targets for gentrification in the 2000s and 2010s.

Ibsen and colleagues define gentrification eligibility precisely. A census tract was eligible if, in the year 2000, both its median household income and median home value fell in the bottom 40th percentile of the metro area. That's the statistical fingerprint of historical neglect.

A tract was then classified as actively gentrified if, by 2019, its inflation-adjusted home value had increased, its median home value ranked in the top one-third of metro tracts, and its growth in college-educated adults also ranked in the top one-third. All three conditions had to be met.

The results are stark. One hundred percent of grade A areas were ineligible for gentrification between 2000 and 2019 — they were already too wealthy, too stable, and too locked in. Grade D areas contained significantly more gentrification-eligible land than any other category.

Within that eligible land, the changes were dramatic. The fraction of college-educated residents in grade D rose by 147 percent between 2000 and 2019. High-income households increased by 58 percent.

Residential land cover jumped from 19 percent to 40 percent when comparing 1997 to 2018 data, and critically, the increase in residential land use was positively correlated with increases in high-income households only in grade D neighborhoods. This relationship didn't appear in grades A or B; it was specific to the historically hazardous zones.

The authors put it plainly: areas that had been marginalized for decades were able to accept these demographic changes due to their diverse and changing land-use patterns. That sentence deserves a moment. The land-use flexibility that resulted from disinvestment — the mixed lots, the light industrial parcels, and the absence of protective single-family zoning — became the economic opening that capital moved through. The wound became the entry point.

So, what does any of this mean for policymakers? Ibsen and colleagues are careful here. They frame the study as a diagnostic, not a prescription.

But the diagnostic is genuinely useful. The gentrification eligibility framework they build, combining Census-derived income and home value thresholds with the fixed spatial boundaries of Home Owners' Loan Corporation zones, can function as an early-warning system for displacement risk. Mapping where eligible tracts cluster and then watching for the signs of active gentrification gives planners a lead time they wouldn't otherwise have.

The deeper insight is about timescales. Land-use diversity differences were measurable as early as 1937. Racial and income disparities became statistically significant by 1980.

Gentrification pressures emerged between 2000 and 2019. Each phase of the story took decades. That means one-off policy interventions, like a single rezoning decision or a single affordability program, are unlikely to reverse trajectories that compounded over a century.

The spatial logic of redlining is stubborn, and countering it requires equally sustained attention.

The authors are also honest about what their study can't do. The high-resolution land-use dataset they used isn't available for most cities, which limits direct replication. They couldn't access parcel-level records of Home Owners' Loan Corporation loan distributions in Denver, so they can't trace the financial mechanism directly.

Their gentrification measure has a 2000 baseline, which means any displacement that happened before that year renders a tract ineligible in their framework, potentially undercounting the true extent of change. Denver's specific drivers — including the particular history of its Urban Renewal Authority, its downtown amenity growth, and its regional migration patterns — shape local outcomes in ways that don't automatically transfer elsewhere.

What does transfer is the framework itself: pairing a multi-decadal land-use time series with standardized Census metrics over fixed neighborhood boundaries gives a spatial picture of vulnerability that no single snapshot could provide. The Home Owners' Loan Corporation zones are convenient because they're historically meaningful and already digitized for most major American cities. However, the logic works with any consistent boundary system. The point is to look back far enough to see how the present was made.

Gentrification is often treated as a market event — capital flows in, prices rise, and people get displaced. What Ibsen and colleagues show is that gentrification is also a long historical process, one whose conditions were assembled over decades through government maps, mortgage policy, urban renewal, and zoning decisions. The neighborhoods gentrifying fastest in Denver today are the ones that absorbed the accumulated costs of those decisions.

Understanding that doesn't resolve the policy dilemma, but it does change the terms of the conversation. You can't address where a city is going without reckoning with how it was built.