Next, I explore selective out-migration. I document that local soil heterogeneity prompted farmers who who depended on social networks to migrate elsewhere. This finding is consistent with the DID results

Second, I provide implicit evidence by studying production outcomes, focusing on a historical context in which American farmers learned and adjusted their practices—the adaptation of wheat cultivation to marginal environments

Exploring channels, I first document the negative impact of soil heterogeneity on farmers’ agricultural learning and adaptation. First, I use an explicit indicator of learning—the rate of fertilizer adoption in farms

I also explore the long-run impact on children’s communal attachment, and find that a child of a migrant to a soil-heterogeneous county was less likely to reside in it as an adult and less likely to marry a wife born in the destination state

Using DID, I explore the short-run impact on in-group identity and document a decrease in communal identification among farmers who migrated to soil-heterogeneous counties compared to those who migrated to soil-homogeneous counties, with no impact on non-farmers

The impact has weakened over time but remains evident today

First, I document a robust negative historical relationship between soil heterogeneity and close-knit communities with a county-level analysis and show that it is unlikely to be driven by selection on unobservables

In the appendix, I also use three indicators focusing on kinship tightness. A separate 🧵 on all the historical indicators, along with data links is now forthcoming on Bluesky 😀

I also develop novel outcome variables measuring different cultural and psychological aspects of historically close-knit communities using census data: the Local Name Index (LNI), focusing on in-group identity and measured using children’s first names

I use detailed and highly-granular geo-referenced soil data to construct a county-level Soil Heterogeneity Index (SHI), capturing the average dissimilarity of soil across neighboring locations in the county

I provide the first empirical evidence supporting Shannon's (1945) Social Learning Hypothesis

However, substantial local soil heterogeneity in certain areas implied that optimal farming practices were highly localized, thereby limiting the effectiveness of social learning. The American historian Fred Shannon argued in his 1945 book that this weakened communal ties

One possible strategy was "learning by doing" (i.e. individual trial and error). Another and potentially more efficient strategy was to engage in social learning and to build on the experience of their neighbors

Historical background: during the 19th century millions of American farmers migrated to unfamiliar environments. They settled vast areas with no prior history of agriculture. Success required a tremendous amount of learning to discover optimal location-specific farming practices

❓Does environmental heterogeneity effects social structure and the strength of social ties? ❓

Celebrating the forthcoming of my paper, "Soil Heterogeneity, Social Learning, and the Formation of Close-knit Communities" in the JPE @JPolEcon 🥳 with a (long) 🧵