Big Cities See Big 'Donut Effect' Hollowing Out CBDs

Since the start of the pandemic, there has been a lot of analysis about what populations and real estate have done in and around cities. For example, suburban office product has outperformed that in central business districts and the pattern held true for the first quarter of 2023. Suburban restaurants outperformed CBD locations because of low office utilization.

Not everything is quite so clearcut. There have been indications that suburban living has started to lose its appeal, particularly for single professionals.

But a new study from Nicholas Bloom, an economics professor, and Arjun Ramani, a researcher, both at Stanford University, suggests that even with some uneven observed results, there has been a definite effect of the Covid-19 pandemic on migration patterns and central business district real estate.

The authors looked at rent and real estate price data from Zillow and then matched that to United States Postal Service National Change of Address data and also used longitudinally-linked household-address data from Data Axle to find where households moved from and to.

There have been three main observations. The first was based on rents and real estate prices acting as proxies for demand for household, business, and real estate. Demand moved from central business districts to suburbs and exurbs where there was lower population and utilization density. In other words, CBDs have been getting hollowed out, something that is probably not a surprise to those working in the office space industry and shifting to suburban rings. Hence, the authors called it the “donut effect.” (Not doughnut, which they passed over because, as they explained, it longer.)

“Rent growth in the CBDs of the largest 12 US metros has been about 15 percentage points less relative to growth of the bottom 50% of zip codes by population density after accounting for pre-pandemic trends,” they wrote. “Similarly, home price growth in CBDs has been more than 20 percentage points less compared to price growth in bottom half of zip codes by density.” In the CBDs of the top dozen largest U.S. cities, net population and business outflows were between 9% and 16%.

The second observation is that the donut effect is mostly a problem of larger cities. They demonstrate the greatest shift in populations and businesses. Mid-sized cities — the 13th to 50th largest cities after the big first dozen — demonstrate smaller donut effects. Metros 51 and up to 365 see little to no effects.

Finally, the third effect is that people moved from large metros to smaller ones and rural areas. “This is especially case for the most agglomerated metros like New York and San Francisco,” they wrote. “Nonetheless this pattern is much less pronounced when measured with rental growth or home price growth. This might be explained by the fact that housing supply is more elastic in less dense metros and rural areas, so population flows do not have major effects on prices.”

A few caveats. One is that this is a working paper, so hasn’t yet gone through formal peer review and often change significantly after that process. Another is that data from Zillow isn’t guaranteed to be accurate for specific properties. As Zillow notes, the national median error rate for on-market homes is 2.4%; for off-market homes it is almost 7.5%. Estimated prices within 5% of a sale price for a given MSA can range from 24.3% to 92.8%. For New York City, the figure was 80.2% for on-market, while for on-market San Francisco it was 62.5%.

The study raises the question of whether developers should abandon CBDs. Some economics would suggest no. Demand has dropped in some big CBDs, but that suggests prices will drop to match the new dynamics of the areas. Also, a big city could lose such percentages of business and population and still have a lot left. But it does mean that development and investing has to be smarter and do more research.