There is a new working paper by Mark Stelzner and Sven Beckert The Contribution of Enslaved Workers to Output and Growth in the Antebellum United States that attempts to estimate the significance of enslaved labor to the American economy. The essence of the approach is to use slave prices to estimate slave production. The price of a slave should reflect the value of output that the enslaved person is expected to produce. They then use estimates of production in 1839 and 1859 to estimate the contribution to growth.
My initial reaction is that recognition that cotton was not the only good produced by enslaved people is long overdue, and their approach generally makes sense, though I need more time to consider the specifics. And I find the results plausible.
That said, their estimate of the percentage of commodity output attributable to enslaved people seems somewhat low. They state that “For the United States as a whole, slaves created between 14.8 and 16.5 percent of total commodity output in 1839 and between 13.0 and 14.8 percent in 1859.” Enslaved people only accounted for 12% of the population, so why would I regard this estimate as low? While enslaved people accounted for about 12% of the population in 1859, they accounted for 21.7% of the labor force. Given that enslaved people were engaged primarily in commodity production, I would have expected their share of commodity output to be close to their share of the labor force. The difference may arise from the fact that much of the over-representation of enslaved people in the labor force comes from women and children being forced to work at higher rates than free people. Their estimates are actually very close to the share of the male labor force accounted for by enslaved men: enslaved men accounted for about 13.6% of the male labor force (labor force share are from Historical Statistics Millennial Edition Tables B11-10 and Ba 40-49). To the extent that the women and children were less involved in commodity production and their prices were lower than for adult males the estimation procedure would lead to estimates of output share that are less than the labor force share. The bottom line, however, is that their estimates of the commodity output share attributable to enslaved people is similar to estimates of the percentage of the labor force accounted for by enslaved people.
Although I find the results reasonable, I was surprised by the authors claims about their implications. They seem to suggest that it provides support for the earlier claims of Baptist and Beckert and refutes their critics. In conclusion, they state that “These estimates do not consider economic activity, like in insurance, banking, transportation and industrial sectors that were stimulated by the slave economy, and thus represent a lower bound estimate for the overall importance of slavery. However, they do show, as argued by Du Bois (1935), Callender (1902), Schmidt (1939), North (1961), Darity (1990), Johnson (2013), Beckert (2014), Baptist (2016), Beckert and Rockman (2016), and Stelzner (2020), that slavery was important historically for US economic development.” In other words, they frame their argument against a straw man who claims that slavery was unimportant instead of framing it against actual people who criticized claims by Baptist, Beckert and others that slave produced cotton was not just important but was the driving force behind economic growth, accounting for nearly half of all output. Ironically, they even criticize scholars who challenged claims about slave produced cotton as the driving force of economic growth for focusing too much attention on cotton. Although, I agree that the history of slavery in American economic development needs to move beyond cotton, if you are going to make misleading claims about the cotton economy you should expect people to challenge them with evidence about the cotton economy.
The claims about slave produced cotton being the driving force behind growth rely on the theory that there were large spillovers from cotton onto the rest of the economy. Many recent historians of slavery and the economy cited Douglass North’s theory that interregional trade linkages made cotton exports the driving force behind growth (this is also the argument of Callender and Schmidt). But research in the 1960s and 1970s had already shown that the available evidence did not support the argument. More recently arguments about linkages have shifted toward finance and services, but the evidence on these linkages remains largely anecdotal. That these connections were large has been asserted rather than established. To the best of my knowledge, quantitative estimates of their significance are still lacking. As they note in their conclusion, Stelzner and Beckert do not provide any evidence regarding them.
In sum, the paper provides a new approach to estimating the economic significance of slavery that begins with the labor of enslaved people rather than the production of specific goods. It finds that that the contribution of enslaved people to production was similar to their share of the labor force, which seems reasonable to me. It does not, however, provide any support for the theory that Baptist and Beckert built their original stories around? In fact, to the extent that the share of commodity production is less than the share of the labor force one might regard the result as consistent with arguments about the negative economic effects of slavery.