A useful concept. Grok:
In Barza, Glaeser, Hidalgo, and Viarengo’s Economic Integration and Upward Mobility in Brazil (PNAS 2026; earlier NBER version Cities as Engines of Opportunities), social mixing is not residential mixing. It is workplace mixing of low-wage and higher-wage workers—how often initially poor formal-sector workers share an establishment with people who are not poor.
That is the construct they use to interpret why some Brazilian cities raise the earnings of the initially poor and others do not.
What they mean by it
The paper asks whether developing-world cities are poverty traps or engines of opportunity. Using RAIS linked employer–employee records (the universe of Brazil’s formal workforce), they follow workers who enter formal work earning less than 1.5 times the minimum wage and estimate city-level initial wage premia and medium-term premia (about 7–8 years later), following De la Roca and Puga.
City size alone does not explain the north–south gap. Southern cities look more like European learning cities; northern cities give an immediate wage bump but little subsequent growth. The authors’ preferred mechanism is Marshallian on-the-job learning: the unskilled gain when they work with the skilled, not merely in the same metro area.
Social mixing (Hidalgo’s label for the same object) is the city-level measure of that exposure.
How it is constructed
From the CEPR write-up of the same project and the NBER text:
- Identify the “poor” as workers below a wage threshold (in the cohort analysis, first-year earnings below 1.5× the minimum wage).
- For each metro area, compute the average share of non-poor coworkers in the typical poor worker’s establishment.
- Subtract the city’s overall non-poor share. That residual is mixing beyond what the city’s poverty rate would mechanically produce—i.e., whether poor workers are isolated inside firms or sit next to higher-wage colleagues.
They also report isolation at occupation and industry levels. Controlling for isolation of the initially poor at establishment, occupation, or industry shrinks the north–south premium gap by about 60% in the working-paper results. Hidalgo reports that workplace-mixing models account for roughly 77% of variation in initial city premia for the initially poor and 83% of variation in medium-term premia.
So the constructs that “paint it out” are:
- Establishment-level exposure of low-wage to high-wage workers (the core mixing variable)
- Industry composition and economic complexity (ECI): more complex southern industries combine a wider skill range in the same workplace
- Public-sector concentration of the skilled in the north, which pulls high-wage workers out of mixed private establishments
- Contrast with residential segregation and with city average schooling, which they find do not explain the mobility gap the same way
They are explicit that this is associational, not a causal experiment: workers and firms are not randomly assigned. The claim is that agglomeration effects run through who works with whom, and that industrial structure (complexity vs. public-sector/simple resource processing) shapes that mixing.
In short: social mixing here = excess workplace co-presence of poor and non-poor workers, used as the operational stand-in for skill integration and on-the-job learning.
Me again: Social mixing in neighborhoods also promotes a type of associational capital. It is why people want to buy in neighborhoods where CEOs live. It’s how the elderly get their sidewalks shoveled. It’s how the kids whose parents work still get to Little League practice.
Home Economics 