Reimagining Cities-Disrupting the Urban Doom Loop

REIMAGINING THE CITY: THE OPTIMAL REAL ESTATE MIX AND HOWTOGET THERE “There are no solutions. There are only trade-offs.” – Thomas Sowell 102

GOLDILOCKS PRINCIPLE Our earlier analysis revealed that WalkUPs, and by extension their cities, perform better across multiple metrics as the real estate portfolio becomes more diverse. Generally, we find that rents and valuations perform better when there is a better balance of product types, which, on average, means less Work and more Live and Play than currently exists. But this is not necessarily a guarantee as there can be “too much of a good thing” for any property type, including too much Live . Indeed, an optimal balance is one that does not have too much or too little of any given product type—the Goldilocks Principle . For example, without enough office, urban cores would lack the central product where many 21st century knowledge jobs continue to be performed. Keep in mind the best place to locate to maximize the number of workers within any given commute zone remains somewhere near the Downtown WalkUP of the city, where over half of WalkUPs’ Work inventory is located. The optimized product mix portfolio is the best way to reverse a doom loop for a WalkUP, ideally transforming the WalkUP

performance into a Virtuous Spiral. For a true Virtuous Spiral, the outputs must benefit the city and its economy (GDP) as well as the private sector investors creating and managing the built environment (PPSF). We calculated the optimal real estate portfolio by balancing economic output (as measured by our place-based GDP) and real estate valuation performance (PPSF), 103 employing a random forest machine learning model. 104 In this methodology, the random forest algorithm creates a predictive model that simultaneously considers the relationship between different product categories (Live, Work and Play) on both GDP and PPSF outcomes. Essentially, the model learns from the dataset by examining observations of WalkUPs with different product mixes. It then identifies patterns and relationships that indicate which combination of Live, Work and Play shares tend to produce the best outcomes for both GDP and PPSF at the same time. The model uses these patterns to identify an optimal mix. We condition our findings on each WalkUP being able to achieve at least the minimum share of each category, as defined by the lower end of the 90% margin of error.

102 Thomas Sowell, A Conflict of Visions: Ideological Origins of Political Struggles. 103 Information about how place-based GDP and PPSF were calculated can be found in the Real Estate Performance section and in the Methodology Appendix. 104 Liaw, A. and M. Weiner. (2002) “Classification and Regression by randomForest.” R News 3(2): 18-22.

74 Cushman & Wakefield