Did Slavery Impede the Growth of American Capitalism? Two Failed Natural Experiments

Two natural experiments reveal the limits of using causal inference to assess whether slavery impeded the growth of American capitalism in the nineteenth century. Superficially, both seem to produce meaningful results. A spatial re- gression discontinuity design (RDD) finds that farm values per acre showed no differences at the free-slave state border. A difference-in-differences (DiD) event study meanwhile suggests that abolition affected farm values more nega- tively where slavery had been more intense before the Civil War. Nonetheless, both of these findings are invalidated due to fundamental problems in the re- search designs. There are systematic violations of the Stable Unit Treatment Value Assumption (SUTVA), given that North and South were in a union for 70 years. Balance tests fail at the border, moreover, indicating that the as-good- as-random assumption does not hold. The failure of these natural experiments thus illustrates how difficult it can be to apply the methods of causal inference to a major question of economic history.