More than 50 years after the passage of comprehensive anti-discrimination legislation, American cities remain highly segregated. The nation’s capital is a glaring example: The D.C. area’s African American residents are concentrated in the Northwest D.C. neighborhoods of Brightwood, 16th Street Heights and Petworth — and, above all, in Northeast D.C. and east of the Anacostia River, where 25 census tracts (the U.S. Census Bureau’s geographic subdivisions) have African American population shares exceeding 90 percent.

Yet Washington is also the most rapidly gentrifying metropolitan area in the United States. Since 2000, 22 percent of D.C. census tracts have seen a large influx of wealthier residents. Gentrification has demographic implications, too. Between 1990 and 2010, the two tracts covering the section of Columbia Heights between 14th and 16th streets saw the black share of the population drop by 20 and 30 percentage points, respectively. The white share in each jumped by more than 20 points.

As they renovate dilapidated buildings and attract new businesses, “gentrifiers” are having a positive impact on many communities. Yet a common downside of gentrification is the displacement of historic low-income residents, particularly renters, who struggle to keep up with a rising cost of living. Now evidence suggests that a major financial regulation enacted to promote financial inclusion may in fact be accelerating displacement, at least in the D.C. area.

The Community Reinvestment Act (CRA), passed in 1977, sought to stem “redlining,” the systematic exclusion of minority communities and neighborhoods from access to credit. Redlining made it hard for black Americans to buy homes and move closer to economic opportunities. Its legacy persists in the form of a $154,000 median wealth gap between whites and blacks, much of it explained by differences in home values.

The CRA requires banks to demonstrate a record of lending in low-income communities and gives bank regulators the authority to block bank mergers if the banks fail to perform. Banks have a strong incentive to get high CRA marks.

According to the National Community Reinvestment Coalition, banks have made a cumulative $6 trillion worth of CRA-related commitments since 1992. In the District, banks lent out $2.7 billion worth of mortgages eligible for CRA points in 2017. But making loans in compliance with the CRA is no guarantee that they will reach historically underserved residents, because loans in a low-income census tract might still be going to people with high incomes who would qualify even without the CRA.

In fact, using detailed mortgage loan data for 2017, my colleague and research associate Andrew Forrester and I found that two-thirds of the mortgage loans eligible for the CRA in the District went to higher-income borrowers living in low-income areas — gentrifiers. In previous years, the gentrifiers’ share of CRA mortgages frequently exceeded 70 percent.

Not only is CRA lending failing to reach its target population, but also evidence suggests it has accelerated the displacement of minorities. Between 2012 and 2017, an additional percentage-point increase in CRA loan volume was associated with a three-percentage-point decline in the minority share of that census tract’s population. However, when we limited our analysis to borrowers, we found that CRA lending did not correlate with an increase in the minority share of those getting mortgage loans.

In plain English: The CRA appears to be accelerating the displacement of minority renters in gentrifying communities, without increasing the proportion of minority residents who can buy homes. That is quite a disappointment to those who see the CRA as an instrument for financial inclusion.

It would be a mistake to blame banks for this failure. Banks have a mandate to lend in the communities where they have branches without incurring risks that might cause them to fail. They seek to make viable loans throughout the communities where they operate, in compliance with regulation and their own underwriting standards. That their CRA-eligible loans mostly go to gentrifiers is likely the result of attempting to reconcile many different objectives at once.

Although mandating banks to lend to low-income borrowers may seem the obvious solution to address the CRA’s shortcomings, history suggests this approach is fraught with risks. In the run-up to the 2007 financial crisis, government steadily raised the share of mortgages going to low-income borrowers, some of whom could not afford to repay. This cost taxpayers hundreds of billions of dollars and forced millions of vulnerable people from their homes.

Instead of using the CRA’s perverse consequences as reason to give politicians more power to dictate who gets loans, we should ask what effective financial inclusion policy looks like in 2019.

For example, many of the 8.4 million U.S. households that lack a bank account — disproportionately black, Latino and low-income — say they do so not because they live far from branches but because they deem fees too high and banks untrustworthy. Why not let other businesses — such as retailers and tech firms — provide basic banking services? In Africa, nonbanks offering mobile money accounts have brought tens of millions of people into the financial system. The same solution could help resolve America’s “unbanked” problem. Policymakers are also right to examine ways that “credit-invisible” consumers, typically young and minority, can make relevant credit information more easily available to lenders. These people are often creditworthy, but they lack the history and assets to qualify using traditional credit scores only.

The legacy of redlining continues to have an impact on the economic opportunities available to different communities across the country. And while there has been progress thanks to growing competition and innovation, decades-old policies such as the CRA no longer work well in many low-income urban areas. It is time to try something different.