Why did humans form state societies, and why in such peculiar places like Mesopotamia, Mesoamerica, and China? Were they coerced by elites or warlords, or did they cooperate to provide public goods and trade? @pieterwfrancois and I investigate & arrive at surprising conclusions. We build on an old theory called circumscription theory developed by American anthropologist Robert Carneiro (1927-2020). He suggested that early states form in areas with high farming productivity that were circumscribed by unproductive land, such as Coastal Peru or Egypt.

Neighborhood segregation by race is a fundamental feature of urban America, but how has it changed in the last 30 years? In 1990, most Black households in Detroit, Cleveland and Chicago lived in almost entirely Black neighborhoods. The country has made remarkable progress since then: Detroit is now less segregated than Houston, Miami, or Charlotte were in 1990. Almost all metro areas have become less segregated for Black households.

In 1631, Prague was conquered by Sweden, only to revert back to the Habsburgs six months later. France took Nancy from Burgundy in 1634 and held it for decades. Thousands of such city switches occurred throughout history. What can we learn from them? New research with Eric Weese: Background: over the years 1000-1850, hundreds of states competed over European territory, even though many of them were short-lived. There was a constant stream of entry and exit in the violent market for governance.

This is Rose Avenue near San Francisco. A quiet residential neighborhood, it looks like life on either side of the street should be pretty similar. But it turns out that opportunities and access to public goods are dramatically different. What's going on? Rose Avenue is the boundary between Piedmont (right) and Oakland (left). Piedmont is a wealthy bedroom community with mostly White and Asian residents, whereas Oakland is much larger and much more diverse, both in terms of household income and demographics.

In this year's edition of diff-in-diff in our Econometrics PhD class, I added a new lecture on matched DID and synthetic controls. All updated panel/DID lectures are on my website at the end of this new thread.

Here are some updated examples on the power of matrix visualization.

https://twitter.com/davidfromterra/status/1390649644418584579

A treatment matrix has rows with different units (e.g. people, firms, regions) grouped by treatment pattern (e.g. never treated, treated in third period, etc.) and columns being periods. Then it's easy to see that 2x2 DID can be simply computed by hand:

Between 1997-2008, voters in Los Angeles approved a series of bonds dedicating around $20 billion in funding to the construction, expansion, and renovation of hundreds of schools. What was the impact of this massive investment in public goods? Julien Lafortune and I investigated. Backstory: in the 1990s, LAUSD school facilities were in dire conditions. School facilities often had broken and missing equipment and lacked adequate restrooms. Classroom temperatures upwards of 90 degrees were not uncommon. Many schools were also severely overcrowded.

Happy this is finally coming out! Julien and I assess the largest school construction program in U.S. history, finding that it substantially increased student outcomes with benefits exceeding costs by 62%. #EconTwitter

🧵 on LA school facilities and the value of public goods ⬇️

https://twitter.com/AEAjournals/status/1404822048619253762

After gradual neglect of school facilities in the 1970s and 80s, the 2000s turned out to be a "boom" phase of investment for many districts. In Los Angeles, the average age of facilities fell from just under 60 to closer to 50 years.

Many school facilities in the U.S. are bad. Really bad. This is the case even though in most places, voters get to decide how much to spend -- how can that be? #EconTwitter

🧵 on new P&P article with @BarbaraBiasi and Julien Lafortune: doi.org/10.1257/pandp.… We first show that, despite the dire state of many facilities, spending actually went DOWN in the last 10 years. This is especially true for richer districts who were finally making up for neglect in the 70s through the 90s.

Hey #EconTwitter! Thinking about including modern diff-in-diff in a 1st year econ metrics class? Here are some ideas from a class co-taught with @kbburchardi and @arash_nekoei. Key elements:

1) MATRIX VISUALIZATION
2) UNIFIED NOTATION

🧵⬇️

Most of what I covered in one slide: First off, MATRIX VISUALIZATION. I adapted matrices that crop up in recent papers by @Susan_Athey and Guido Imbens that visualize how units are treated over time, which I call TREATMENT STRUCTURES.