Culture and Modernization in the United States, 1880-1930
Testing Sowell's cultural persistence hypothesis
Thomas Sowell argued in Black Rednecks and White Liberals that population differences in socioeconomic outcomes were the result of cultural persistence. Specifically, he argued that before the great migration of southern blacks to northern cities in the early 20th century, neighborhoods and schools in the north were in the process of becoming integrated. Blacks in New England enjoyed relative upward mobility as the second industrial revolution gained steam. This process was halted and reversed by the great migration as southern blacks brought southern culture to the north that was unacceptable to northerners, particularly in New England. With the arrival of what they considered to be blacks lacking in culture, whites abandoned mixed neighborhoods and schools.
In general, following David Hackett Fisher, Sowell argued the case for strong cultural persistence and the theory of first settlement. The older regions of the United States were first settled by people from different parts of Britain and that heritage not only persists to the present day but structures regional folkways and socioeconomic outcomes. New England was first settled by people from East Anglia in the early 1600s; the Chesapeake Bay by people from the south of England in the late-1600s; the Delaware Valley by people from North Midlands in the fifty years around 1700; and Appalachia and other areas in the backcountry from northern England, Scotland and Ireland in the mid-1700s.
Sow
Sowell is a pretty unabashed New England supremacist. His Roman contempt for the barbarians beyond Hadrian’s Wall is open and scathing.
The cultural values and social patterns prevalent among Southern whites included an aversion to work, proneness to violence, neglect of education, improvidence, drunkenness, lack of entrepreneurship, reckless searches for for excitement, lively music and dance, and a style of religious oratory marked by strident rhetoric, unbridled emotions, and flamboyant imagery. This oratorical style carried over into the political oratory of the region in both the Jim Crow era and the civil rights era, and has continued into our own times among black politicians, preachers, and activists. Touchy pride, vanity, and boastful self-dramatization were also part of this redneck culture among people from regions of Britain “where the civilization was the least developed” [Fisher].
Thomas Sowell, Black Rednecks and White Liberals, 2006.
Is there such strong cultural persistence? Has cracker or redneck culture persisted since the arrival of north Britons to the backcountry in the 18th century? And is the cultural particularism of the south responsible for its relative underdevelopment?
One is immediately reminded here, not only of Vance’s Hillbilly Elegy, but also the debate over culture and modernization in the global south. The lower productivity of southern farmers is explained here by their tendency to “gather around the courthouse and country stores to smoke, chew, talk politics, and, in general, to waste time.” Many southern businessman, Sowell cites McWiney’s Cracker Culture, “were unreliable about either paying their bills or delivering goods and services when promised.” This generalized incompetence is pretty much what is posited to explain the failure of the industrious revolution to spread to the global south. Gregory Clark showed that, in 1910, each New England handloom worker manned 2.97 machines, each Southerner, 2.65. In England, there were 2.04 machines per worker; 0.50 in India, 0.48 in China; and so on. Differences in factory discipline, Clark argued, explain world order — a greater capacity for self-discipline was the key to modernization and high income.
Bailey (1960)’s effective temperature (ET) is the strongest environmental conditioner of socioeconomic outcomes. Neither Clark nor Sowell control for the structuring effect of ET. But it is clear that the North and the South, globally and within the United States are differentiated by ET. One reason both southerners and Indians need frequent breaks is that it gets hot — they have to stop in order to obey what’s called the human thermal balance equation. And ET is just the top conditioner. In general, both have failed to test their own hypotheses. They couldn’t figure out how to control for region fixed-effects and directly test the cultural channel of historical causation. Of course, that is not even a question if you just have cross-sectional data.
For the US, we do have microdata from IPUMS for the crucial period when the second industrial revolution exploded: 1880-1930. These six censuses contain information on both the socioeconomic index but also, crucially, last names. We outline a strategy of using rare surnames as a way to get a cultural persistence signal. We then show that we can test Sowell’s hypothesis of cultural persistence by looking at what happens to cultural Southerners and New Englanders, black and white, in the United States, independent of the region where they live.
The way to construct the cultural signal is to ask: if you hear an uncommon surname, can you tell with good odds that the bearer of the surname must be from a specific region? We construct a novel dataset of 380,453 observations from census microdata in IPUMS’s 1 percent samples for 1880-1930 — available on my GitHub. We select all surnames in the 1880 sample for non-Hispanic blacks and whites. We compute the odds that they are from one of the 9 census subregions where they were most frequent during 1880-1930. We then choose a subset where the odds ratio is greater than 2 and significant at the 5 percent level. These odds ratios will serve as our cultural persistence factor because higher odds imply a higher probability of cultural persistence. This provides us a dose-response relationship with origins in different cultural regions, the crucial conditioner hypothesized by Sowell. We can then control for region fixed-effects and test two predictions of Sowell’s hypothesis.
Sowell’s hypothesis says that the socioeconomic retardation of the South is due to Southern cultural particularism and resistance to the adoption of New England’s folkways. His hypothesis predicts that whites with New England roots should do better than people with ancestry in other regions, in particular, whites with Southern roots — independent of where they live. These premia should be recoverable even after controlling for region fixed-effects.
In his account, New England blacks suffered downward mobility as a result of the arrival of southern blacks in large numbers during the great migration that began in the period under study. However, given his insistence of cultural persistence, it follows from his account that blacks with origins in New England should do better at the margin than southern blacks.
We first present a null model of socioeconomic status that we will control for in order to test Sowell’s hypothesis. In this model, we include fixed-effects for sex, race and census subregions. We find a very large effect for sex (b = -9.0, P < 0.001) that is 4-5 times as a large as the race fixed effect (b = -2.0, P < 0.001). The interpretation is that gender bias in socioeconomic competition was dramatically more pronounced than the racial bias. We also find a significant time-trend, consistent with socioeconomic status rising in general through the secondary industrial revolution.
We take Hubbart’s Older Middle West (“East North Central”) as the neutral region. The later-settled Mid West (“West North Central”) was poorer (b = -0.8, P < 0.0001); the Pacific was richer (b = 2.5, P < 0.001), but not as rich as New England (b = 2.9, P < 0.001). The South was uniformly poorer. The old hearth in the coastal south (“South Atlantic,” b = -1.8, P < 0.001) was poorer than the Mountain region (b = -0.2, P = 0.255), or even the later-settled Mid West. The poorest was the region of the Mississippi Delta (b = -2.5, P < 0.001). In the early 20th century then, the United States was regionally polarized with New England at the top and the southern backcountry, the home of cracker culture, at the bottom.
These fixed effects were visible to contemporaries and researchers later. The question is whether cultural persistence can explain the polarization. In order to test the hypothesis, we split the sample by race and keep the fixed-effects for regions in order to bloc the flow of information from environmental differences. We also keep our control for sex. We stratify by ancestral region associated with the odds ratio. We then introduce random effects, which we model as linear functions of our cultural persistence factor — the log odds of being from the most likely ancestral region. If Sowell’s theory of cultural persistence is right, then there should be a clear dose-response relationship between New England ancestry and Southern ancestry on the one hand, and socioeconomic status on the other: the slope for the odds of New England ancestry should be significantly positive; the slope for the odds of Southern ancestry should be significantly negative.
The next table displays the random effects coefficients for US-born, non-Hispanic whites. We find the predicted dose-response relationship. Higher odds of New England ancestry predict higher socioeconomic status in the early 20th century even after controlling for region fixed-effects (fitted but not reported), although the coefficient is only marginally significant (b = 0.1, P =0.078). Higher odds of being from both the Mississippi Delta (b = -0.3, P = 0.280) and South Atlantic ancestry predict lower socioeconomic status (b = -0.9, P < 0.001), although only the latter is significant. This pattern of statistical significance is probably because southern surnames are very likely to have originated in the coastal south. So there is some evidence for New England and Southern cultural persistence among whites predicted by Sowell.
Curiously, we find that higher odds of western MidWestern ancestry significantly predict lower socioeconomic status (b = -0.5, P = 0.047). The slopes for the odds of ancestry in the Mountain region (b = -1.`1, P = 0.292) and Pacific (b = 0.4, P = 0.292) bear the expected signs but are not significant.
The next table displays the random effects coefficients for US-born, non-Hispanic blacks. We find that, contrary to the most straightforward interpretation of Sowell’s account, New England ancestry among blacks predicts lower socioeconomic status (b = -0.1, P = 0.002). No other slopes of the persistence factor are significant. In particular, southern ancestry among blacks is not at all predictive of socioeconomic status. These results cast doubt on Sowell’s happy story of the advancement of blacks from New England and the retardation of blacks from the South.
The washout that this reference frame reveals for blacks may be due to a poorer phylogenetic signal in black surnames. Since the vast majority of blacks in 1880-1930 were descended from enslaved people who were given surnames by their owners, it is should be not be surprising that black surnames do not contain a good signal of ancestry. In order to be sure, we drop the fixed-effects and refit the model. We find the same pattern as before. We also control for literacy, with the same result. The counter-intuitive result that black people with surnames from New England had lower socioeconomic status is robust. Could it be because slave-owners from New England in the South were particularly likely to run plantations? Does this signal contain information on the culture of black field hands in large plantations in the Mississippi Delta controlled by New England money? We leave this as an open problem for future work.
For US born whites in 1880-1930, we have seen that higher odds of southern ancestry predict lower socioeconomic status while higher odds of New England ancestry predict higher socioeconomic status. This is consistent with Sowell’s aesthetics, even if his account of divergence by regional ancestry among blacks is not borne out by the data.
Cultural persistence may explain the long-run hegemony of New England and the decadence of the South. How could it be otherwise? If southern England was able to dominate the world despite her size, if New England could discipline the vast continent of the United States, there must be something that made them especially dynamic. We don’t have to fall for Sowell’s New England supremacism to appreciate that some people in some periods were more dynamic than others. We’re seen before that the deep cultural persistence of Todd’s family systems may be a strong conditioner of world order. Is it really surprising to find that cultural persistence conditions the social and regional order as well?
I should have added that the problem with the "black rednecks" thesis is that the areas that the Borderers settled didn't have lots of slaves. The black areas of the map don't include the Virginia Tidewater area or much of the Carolina lowlands, much less Alabama, Mississippi or the more fertile parts of Louisiana and Georgia, which are the parts of the South where the black people were made to live.
In fact, the Borderers were more inclined to be pro-Union (e.g. West Virginia), because they saw no reason to fight so that the plantation owners could keep their slaves, not to mention the federal government was further away and less bothersome than the state capital.
Bad cat.
Well, there was a black justice of the peace in Massachusetts before the Civil War, in 1870, a black man was elected to the Massachusetts state legislator (and was later appointed a judge) and say what you want about a Charles Sumner, he was no racist.
For that matter, the Freedman's Bureau was liberally stocked with graduates from the best New England colleges and universities.