How population is changing in Wisconsin and the Upper Midwest since the pandemic

map showing population change from the 2020 census to July 2023 in the subcounty geographies of selected midwestern states

(This post also appeared in The Recombobulation Area).

The official census count occurs just once every ten years, and it’s out of date by the time it gets released. 

The last census was officially conducted on April 1, 2020, so it missed essentially all the population changes caused by the COVID-19 pandemic. But each year, the U.S. Census Bureau also releases “intercensal” population estimates. These are based on carefully collected administrative records (births, deaths, tax returns, etc.), and they give us the best look at how our current population is changing. The estimates cover 12-month periods beginning on July 1. The latest data covers the year from July 1, 2022 to June 30, 2023.

The pandemic and its aftermath is still the big story in this data. Cities across the country shrank. Some, like Detroit, have begun to grow. Despite Milwaukee Mayor Cavalier Johnson’s well-publicized goal of “1 million Milwaukeeans,” the city’s population recovery has yet to begin.

The latest estimates show positive signs for Wisconsin, but continued struggles for Milwaukee, relative to our midwestern peers. Wisconsin has largely returned to pre-pandemic form, while Milwaukee County is continuing to shrink at twice the rate of the 2010s.

Wisconsin added 20,000 net new residents from July 2022 to July 2023, a growth rate of 0.35%, which is practically identical to the state’s average growth rate during the previous decade. That growth was nearly double the state’s increase of 11,000 in 2022, which came on the heels of a 17,000-person loss in the first year of the pandemic.

Population change can be broken down into two components — net migration and natural change (births minus deaths). The next graph shows why each state grew or shrank over the past three years. Wisconsin’s 2023 growth rate falls below Minnesota and Indiana but above Michigan and Iowa. Illinois is still shrinking badly.

bar plot showing the components of population change for selected midwestern states

Each state has followed a different trajectory. 

  • Wisconsin had slightly more deaths than births in 2021 and 2022, before flipping to slightly more births in 2023. The bigger change has come from improving migration numbers. The state lost 16,000 net migrants in 2021 but gained 15,000 in 2022 and 19,000 in 2023.
  • Michigan has followed a similar trajectory as Wisconsin, but with more negative natural increase and slightly less migration.
  • Minnesota’s net migration has bounced around over the past few years, but its strong birth rates have kept it from shrinking much.
  • Iowa is close to flat—but slightly positive—in both natural change and net migration.
  • Illinois has maintained slightly positive natural change, but it gets hammered on net migration. A net of more than 100,000 people left the state in both 2021 and 2022. The outbound tide slowed to 43,000 in 2023, but Illinois remains the only state in this set to have negative net migration.
  • Indiana has attracted more than 20,000 net new residents in each of the past 3 years—some of them, doubtlessly, former Illinoisans.

Milwaukee County’s population fell by an estimated 1,800 during the 12-month period ending July 2023. That was an improvement over the previous two years, when the population fell by 6,200 and 14,300, respectively. Still, the county shrank by 0.2% in 2023, compared with an average annual decline of less than 0.1% throughout the 2010s.

The maps below show components of population change for each county.

maps showing county level components of population change for selected midwestern states

The strongest position is in the top left map, which shows counties with more births than deaths and positive net migration. It includes the counties surrounding Indianapolis and the Twin Cities—though notably not any of those cities themselves. The growth emanating (though not shared by) the Twin Cities is so strong that it reaches as far as Eau Claire, Wis. Other areas in Wisconsin in this strong position include Dane and Marathon counties, the Fox Valley, and the greater La Crosse area.

Only a handful of counties have negative migration but enough of a positive birth rate to keep growing nonetheless. The largest in this category is Hennepin County, home to Minneapolis. In Wisconsin, they include Clark, Lafayette, and Trempealeau counties.

Many more counties have aging populations, resulting in negative natural change, but enough in-migration to create population growth. Broadly speaking, this includes a great swathe across northern Michigan, Wisconsin, and Minnesota. Perhaps the remote work boom is finally creating the conditions for population growth across the Northwoods.

If the healthiest counties grow from births and migration, then the most troubled counties are shrinking for both reasons. These counties — which have more deaths than births and more leavers than comers — are found most commonly in rural Illinois and Iowa. In Wisconsin they include just Columbia, Crawford, Juneau, and Jackson counties.

Seven of the region’s largest 10 counties fall into the next category — where natural change is positive, but offset by out-migration. To put it reductively, people seem to have kids in these counties, then they leave. This status includes Milwaukee County; Cook, Kenosha, Lake, and DuPage counties in the Chicago metro; Wayne and Oakland counties in the Detroit metro; and Marion County (containing Indianapolis).

The final category, positive migration but an even larger negative natural change, occurs mainly in rural counties — particularly in Illinois.

The Census Bureau further estimates municipality population by tracking new housing unit construction and allocating the county-level population estimate into each town based on the average household size in the 2020 census. Based on this methodology,  Milwaukee city’s population fell by 2,200 in 2023 to a new low of 561,400. The most significant growth came in Oak Creek, which likely added about 1,100 new residents. Most of the remaining municipalities are estimated to have shrunk slightly.

Milwaukee’s decline of 2,200 is an improvement over its loss of 2,900 in 2022 and pandemic-fueled drop of over 10,000 in 2021. But Milwaukee’s rate of decline in 2023 was still tied with Rockford, Ill., for the worst among the region’s 15 largest cities.

Milwaukee’s rate of change since the 2020 census is second-worst, trailing only Chicago. Madison, on the other hand, had the highest rate of growth in 2023 and the second highest since the pandemic (after Fort Wayne).

table showing population trends for the 15 largest cities in selected midwestern states

Most of the 15 largest municipalities in Wisconsin have shrunk since the 2020 census, with the exceptions of Madison, Eau Claire, and Janesville. But all of these cities do show signs of improvement in 2023. Eau Claire has shown particularly robust growth, passing Waukesha to become the state’s seventh largest city in the latest estimates.

table showing population trends for the 15 largest cities in Wisconsin

By comparing Wisconsin with this set of neighboring states, I hope to better place our demographic situation in context. The bright spots in Wisconsin extend well beyond Madison. Many regions surrounding smaller cities like Eau Claire, Wausau, and the Fox Valley are doing quite well. The rural Northwoods is attracting enough migration to offset the natural decline of its aging population. Our rural communities are in a much healthier place than those of Illinois or Iowa.

Still, even if Wisconsin’s outlook seems better than Illinois’, Milwaukee nonetheless appears to be on the same trajectory as Chicago. In both, the population has fallen by about 3% since the pandemic began. The culprit is the same. The birth rate in each city is positive, but more people choose to move away than to move in.

Continue ReadingHow population is changing in Wisconsin and the Upper Midwest since the pandemic

Highlights from the 2024 Spring Election in Milwaukee County

Only a few of the major races on Milwaukee’s spring ballot were competitive. Most notably, the latest Milwaukee Public School District funding referendum passed by fewer than 2,000 votes. Both Biden and Trump put up relatively weak showings in their respective pro forma presidential preference votes. Evan Goyke handily won a landslide victory against the incumbent Milwaukee City Attorney.

The following discussion is based on complete, but unofficial, election night vote totals. All demographic data is from the 2020 census. Registered voter statistics are based on the number of registrants entering election day; they do not include same-day registrants.

MPS Referendum

MPS referendum ward results

The referendum received about 41,600 “yes” votes to 39,900 “no” votes. The No’s actually won more wards—172 to 164.

As the map below shows, support for the referendum was strongest in neighborhoods along the lake and on the near west side. Opposition was strongest on the far south and southwest sides. Most north and northwest side wards also voted against the referendum.

These patterns do somewhat follow Milwaukee’s racial divisions. The “Yes” vote won majority non-Hispanic white wards (53.2% for “Yes”) while narrowly losing both majority Black wards (48.4% “Yes”) and majority Latino wards (49.4% “Yes”).

But these differences are small. The presence of children in a ward correlates much more strongly with the referendum vote.

In wards where fewer than 20% of households include a child under 18, the “Yes” vote won by nearly two-thirds of the vote, 65.8%. But in places where more households have kids, it lost. In wards where 20% to 40% of households have a kid, 45.9% of voters supported the referendum. Where 40% to 60% of households have a kid, the “Yes” vote took 46.9%.

These are just correlations. We don’t have data on how parents themselves voted. But we can say that the MPS referendum was most popular in the parts of the city with the fewest children.

Presidential Preference

No candidate remained campaigning against Biden or Trump in their Wisconsin presidential primaries. Still, both candidates put up relatively weak showings.

Across the entire county, Trump received 73.2% from Republican primary voters. Nikki Haley’s defunct campaign still got 16.6%. In 2020 (with no other named candidates on the ballot), Trump won 97% support among Republican primary voters.

Biden received 84.5% of the vote, compared with 12.2% for uninstructed delegates, and 2.5% for Dean Phillips. The last primary featuring an incumbent Democratic president was 2012. In that year, Obama won 99.1% of the Milwaukee County vote, and uninstructed delegates received 0.7%.

maps showing the presidential preference results in milwaukee county

The table below shows the primary results in each municipality.

A slim majority of voters participated in the Republican primary in Hales Corners and Franklin. Democrats were in the majority everywhere else.

Among Republican primary voters, Trump generally did best in the southern suburbs, garnering 80% or better in West Milwaukee, Oak Creek, Cudahy, and Hales Corners. Haley did best in the wealthier north shore suburbs, receiving 30% or better among GOP voters in Shorewood, Fox Point, Whitefish Bay, and River Hills.

Activists opposing Biden’s handling of the war in Gaza campaigned in support of the “uninstructed delegation” option in the Democratic primary. “Uninstructed” support varied between municipalities with the most support coming in two usually quite different places. Shorewood is the most Democratic municipality in the entire county, and 15% of its Democratic primary participants supported the uninstructed campaign. But “Uninstructed” support was even higher, 16%, in Franklin, one of the county’s most conservative suburbs. In general, uninstructed support was higher in the southern suburbs and lower on the north shore.

Within the city of Milwaukee, the uninstructed campaign generally performed best in the Riverwest neighborhood, with pockets of significant support scattered elsewhere in the city.

table showing presidential preference results in milwaukee county municipalities

Milwaukee City Attorney

map showing milwaukee city attorney ward results

Evan Goyke won 63.3% of the vote, defeating Tearman Spencer in the race for City Attorney. He performed particularly well in the Washington Heights neighborhood and the neighborhoods along the lake. Spencer’s support was mainly limited to parts of the north side of the city.

Goyke carried 211 wards (to Spencer’s 124) and 10 of the city’s 15 aldermanic districts. Goyke won more than three quarters of the vote in the 3rd, 14th, 10th, and 4th districts.

Spencer’s best district was the 1st, where he won two-thirds of the vote, followed by the 2nd, where he won 63%.

Overall Turnout

map showing ward turnout as a share of registered voters

Interpreting turnout is always challenging, because factors like the mix of elections on a ballot and the competitiveness of those races always vary from one election year to another. Turnout in April 2024 appears to have been middling, compared to recent past cycles.

  • In 2020, about 210,000 voters, or 41% of registered voters, cast a ballot.
  • April 2022 saw about 145,000 voters, or 29% of registered voters.
  • This year, 165,000 voters participated, about 34% of those registered.

As a share of registered voters, turnout was highest on the north shore. The top-5 municipalities were Glendale, Greendale, Shorewood, Fox Point, and Bayside, among which turnout ranged from 42.1% to 45.4% of registered voters.

The lowest turnout came in West Milwaukee (22.6%), Cudahy (30.7%), West Allis (31.4%), and Milwaukee (31.6%).

Continue ReadingHighlights from the 2024 Spring Election in Milwaukee County

Landlords use many different LLCs. A new tool,, connects them.

Most landlords are small, but the largest 1% of networks own more than 40% of rental units.

By John Johnson and Mitchell Henke.

During the late 2010s, companies owned by a single landlord, Curtis Hoff, generated 5% of all evictions filed in Milwaukee despite owning just 0.5% of the rental stock. His properties racked up code violations at a rate three times that of other large landlord operating in the same parts of the city. The annual number of evictions they filed exceeded 90% of their total housing units.

Despite posting these eye-watering numbers, Hoff’s companies largely flew under the radar until a series of investigative articles in the early 2020s brought them to light, just as he was getting out of the business. That’s because Hoff’s properties were distributed among about 20 different limited liability corporations (LLCs), each of which was the official, legal owner of a different small set of properties.

It’s not that Hoff was trying to hide. Each of these companies began with the letter “A,” a practice dating back to the era when the courts heard each day’s eviction cases in alphabetical order. They all had their taxes mailed to the same address on Good Hope Road—an office building with a large “Anchor Properties” sign.

Still, if you searched the city’s property ownership dataset or the court system’s database, you would have no simple way of telling that all these properties were connected. And so, for years, few people figured it out. As the executive director of the Legal Aid Society of Milwaukee told the Journal Sentinel, “Maybe we’re not connecting the dots like we should be. Nobody’s able to officially track who the problem landlords are.”

To help address this, we’ve built, a website that lets anyone quickly discover the connections between legal property owners that already exist in publicly available data. A user can simply enter the address of any landlord-owned property in the city, and the website will show that parcel’s official owner, the other owners connected to it, and the total list of properties in the “ownership network.” We also include network-level code violation and eviction annualized rates.

Our process works by standardizing owner names, the addresses at which they receive tax bills, and their corporate registration addresses. We then use network analysis software to identify connected owner names. More methodological details are available on the “About” page. All our data sources are public, and we publish the complete code and source data in this public repository. The data updates each weeknight with the latest version of the city’s property ownership database.

Curtis Hoff no longer owns any properties in Milwaukee, but here is what our website might have shown in 2019, if it had existed. Each green triangle shows a different owner name and each orange circle shows a tax bill address. The lines show the number of times each of the two nodes are connected.

network graph showing companies connected to Curtis Hoff

Although Hoff is gone, practically all other large landlords also use multiple companies. For instance, Milwaukee’s largest landlord is Joe Berrada, well-known as the “boulder guy” for his unique style of landscaping. His companies own around 9,000 rental units spread across more than 800 parcels in the City of Milwaukee. Our website identifies over 100 distinct companies associated with Berrada, each of which individually own fewer than 300 units.

Research from many cities shows that landlord size corresponds to different rental practices. Large landlords are much more likely to file evictions. They may also raise rents more aggressively than small landlords. On the other hand, small landlords may employ more discriminatory tenant screening methods than larger, more professional companies. In any case, there is much variation between the behavior of large landlords. As with Curtis Hoff, a handful of companies can account for a greatly disproportionate of eviction cases, for instance. All this means that targeted interventions by housing advocates can be quite successful.

In Milwaukee, we identify about 49,000 landlord-owned parcels in the city containing around 148,000 housing units. This is comparable with the Census Bureau’s estimate of tenant-occupied and vacant housing units. There are about 21,500 distinct owner networks in the data. Of those networks, 15,500 (or 72%) own just a single property. Only about 130 ownership networks own more than 25 rental properties. Nonetheless, those 130-odd networks collectively own more than a fifth of the city’s rental units.

Put another way: the majority of landlords own just 1 or 2 rental units, but the typical tenant rents from a much larger landlord. Our data shows that 44% of rental units are owned by the largest 1% of landlord networks.

bar plot showign landlord network size

We hope this website will be useful to a wide variety of users, including government agencies, community organizations, prospective homebuyers, and tenants; and we are committed to maintaining and improving the project for the foreseeable future.

Continue ReadingLandlords use many different LLCs. A new tool,, connects them.