Thread by @mattkahn1966

Let's return to Jim Rauch's key 1993 Journal of Urban Economics paper that started all of the action in urban economics. His paper must have been motivated by Lucas and Romer's work on human capital spillovers.
sciencedirect.com/science/articl…

https://twitter.com/mattkahn1966/status/1366804480537239554

Some notation , person i in city j at time t. The researcher observes her wage, a vector of her demographics X and her city's average education level. Using OLS, estimate B2 ;

Wage_ijt = B1*X_it + B2*Education_jt + U_ijt

B2>0 ==> human capital beneficial local spillover

In the early 1990s as the credibility revolution (Angrist, Card, Katz, Krueger, Imbens) unfolded in applied micro, reasonable concerns could be made that the average human capital in a city is not an exogenous variable in a Mincer regression.

In our Superstar Cities economy, people are not randomly assigned to cities. Paul Krugman and Paul Romer are both in New York City (and I'm in Baltimore). A high skill city may signal high unobserved productivity shifters! (selection , not treatment effects)

Recognizing this point, urban economists sought "instrumental variables" --- In English, a variable that determines a city's skill level that isn't correlated with unobserved productivity or amenity factors that are represented by the "U" in the regression model.

Enrico's important insight appealed to history and the persistence literature. The Morrill Act of 1862 created Land Grant Universities. These are points on a map that create skill. Path dependence arguments predict that this creates skill clusters.

sciencedirect.com/science/articl…

At @JHU_Cities, Mac and I have argued that college towns tend to have great quality of life as they have sports teams, green space, enthusiastic young people, music, culture and the arts. In our emerging Work from Home economy, new clusters will form in these places.

Urban economists note my point. If University towns have great quality of life that the econometrician does not capture in estimating a Mincer regression, the the Land Grant IV fails the exclusion restriction. Why? See my next tweet

Sherwin Rosen and Jennifer Roback taught us that places with great amenities have higher rents and tend to offer lower wages as people pay a compensating differential to live there. I deeply admire the IV paper but old instruments always require a new look. #rainfall revisited

Here is Jennifer Roback's classic urban economics paper that is now almost 40 years old.
citeseerx.ist.psu.edu/viewdoc/downlo…

To summarize, in a spatial economy where we cannot randomly assign different people to cities, how do estimate what would be the gain for different people if there is a "helicopter drop" of skilled people who move into the city? What is the causal effect of upskilling a city?