Complexity and colocation of the college-educated: Sort of complements

Kremer (1993) proposed that complementarity between highly skilled workers who are employed in firms with complex production processes would motivate both the firms and the workers they employ to colocate geographically, but thus far empirical work has focused on complementarity at the firm and industry level. This paper uses Census and American Community Survey data between 1960 and 2020 on 561 cities to present new evidence on high-skill complementarity among college-educated workers at and below the city level. The findings include: (1) More complex occupations and industries employ higher college shares; (2) Cities with higher overall college shares employ higher college shares within an industry or occupation; (3) More complex industries and occupations tend to exhibit larger gradients of college share with respect to city college share; (4) Cities with higher college shares tend to house more complex mixes of industries and occupations; (5) Below the city level, (a) college share in an industry tends to be higher, the higher the college share in industries up- and downstream; (b) college share in an occupation tends to be higher, the higher the college share in the remainder of the industry; and (6) More complex industries and occupations tend to exhibit larger (sub-city) gradients. Finally, analysis of college wages supports the hypothesis that productivity-related factors are at work. • Sectoral variation is used to shed new light on high-skill complementarity in cities. • Higher-skill cities have higher-skill sectors at the intensive and extensive margins. • Industry-city skill is positively related to skill in industries up- and downstream. • Occupation-industry-city skill is positively related to industry-remainder skill • Complementarity tends to be higher in more complex, higher-skill endeavors. • Analysis of college wages suggests that productivity-related factors are at work.