Yesterday, minding my own business on Twitter, I came across an article that made me, as well as about half of the internet, furious. Someone was wrong on the internet. It turned out that an undergraduate was wrong on the internet, and despite the generally low esteem I hold for Quillette as a publication 1 and the generally acute contempt I hold for Thomas Sowell as an intellectual — his mediocre oeuvre is conspicuously overrepresented in the footnotes of the essay — I didn’t really want to dunk on an article written by an undergraduate. After all, I said plenty of stupid things as an undergraduate, and while I’m still pretty stupid, I like to think people won’t hold me excessively responsible for all the stupid things I said back then.
However, there is one part of the argument I don’t feel inclined to let through to the keeper, because it’s a point of economic history. I was rather stunned to read from the article’s author, Coleman Hughes, that:
‘Although it is true that the median income of white men more than tripled between 1939 and 1960 (rising from 1,112 dollars to 5,137 dollars), the median income of black men more than quintupled (rising from 460 dollars to 3,075 dollars).’
An increase of 568% in incomes for African-American male workers would indeed be incredible. China, usually considered the star pupil of the second half of the 20th century, has, at its most impressive, only ever managed to scrape a 334% increase in real per capita income over a 21-year period, according to the 2018 Maddison Project database (the period in question ran from 1991 to 2012, if you’re interested.) Singapore, which Hughes singles out as a growth miracle, maxed out at roughly the same level: the rosiest 21-year period in the city-state, from 1958 to 1978, saw the Singaporean economy grow a mere 335%. Japan actually does somewhat better: 440% growth in the period between 1949 and 1970: good, but nowhere near what is claimed for African-American male workers.
So why don’t we teach courses about the African-American male income growth miracle, 1939-1960? Why aren’t there books about it? The reason, I think, is easily discovered. While the source for this claim is given as Jason Riley’s book False Black Power?, which I don’t have access to, the original source is clearly the US Census bureau. And the data they report are, clearly, nominal wages, not real wages. After all, US per capita GDP only increased 73% in the same time period, so unless there was a truly stupendous decline in the capital share of income, it’s hard to see where the wage growth would be coming from. So this data must clearly be adjusted for inflation before it can be used to attack Richard Rothstein, who wrote in The Color of Law (an excellent book, by the way) that ‘African American incomes didn’t take off until the 1960s’. Using CPI to deflate the income series in the Census Bureau’s report, we find that the median white male wage increased from $1,112 (in constant 1939 dollars) to $2,410; for black males, the median wage increased from $460 to $1440. 2 While this is still a significant increase, it’s much less dramatic than Riley or Hughes claim. It’s also worth pointing out that since they were starting from a larger base, white male workers actually saw a larger absolute increase in their living standards — an additional $1298 dollars (in 1939 money), compared to an additional $980 for African-American male workers.
Of course, here we have assumed, as one normally does, that the same price index is appropriate for both African-Americans and whites. In this instance, though, there’s a good reason to suppose that this assumption is inadequate, for the simple reason that much of the increase in the median nominal wage of African Americans was not because they somehow defeated racism, but rather because they moved away from the part of the United States most completely drenched in it. The Second Great Migration, described movingly in Isabel Wilkerson’s book The Warmth of Other Suns, was one of the great internal population movements in American history. Southern African-Americans moved en masse to Northern cities where nominal wages were higher than in the South. However, with higher nominal wages came less agreeable things. Infant mortality, for example, was higher in the North for migrating African-Americans, as, more importantly for the per capita income estimates, were prices.
What this means in practice is that the ‘median’ African-American male in 1960 may not have been living anywhere near the ‘median’ African-American male in 1939, and may in fact have been living in a city with a much higher price level. If so, then even adjusting his increased wage for CPI inflation will overstate his actual income increase, because it doesn’t take account of the fact that he’s facing a new, higher relative price level.
Comparing price levels across cities is of course hazardous, and there’s no perfect series we can use to check, but, as a kind of first pass, I used the food portion of Bob Allen’s respectability basket basket (182kg of bread, 34kg of dried beans, 26kg meat, 5.2 kg butter, 5.2kg cheese, and 52 eggs per person per year) and historical food prices by city, as well as gross median rent by state from the Census bureau to come up with a very rough indication of the cost of feeding and housing a family of 4 in 1959-60 in various cities in the United States. As I suspected, the Southern cities (Atlanta and Houston) were the cheapest; Los Angeles, which was a destination for some Southern African Americans during the Second Great Migration, was the most expensive of the cities I looked at.
|City||Food & rent|
So what might this mean in practice for our measures of wage income? Let’s pretend, for the sake of argument, that we have an African-American man working in 1940 in Atlanta earning that national median salary for African-American men. His wage would be $460. To figure out how well-off he is, we need to calculate the cost of his food and rent basket, which is tricky, since the sources for food prices I’ve found don’t go back that far. So rather than trying to find equivalent sources to collect intercity food prices for 1939-40, I assumed that price growth was constant across areas, just not the price level — hence I deflated the cost of the 1960 basket using national CPI. The cost of that basket plus rent for the year in 1940 would be $288 + $156 = $444. In 1960, let’s assume that this man has moved to Los Angeles, and still receives the national median wage, $3075. In total, he has to pay $1921 in food and rent for his family. Thus, over time, this ‘median’ man’s real income has increased from a welfare ratio of $460/$440=1.05 to one of $3075/$1921=1.6, an increase of 52%.
This is still an appreciable increase in living standards. But it is nowhere near as large as the 213% increase in real income we allocate to our median worker if we assume that he did not migrate. Now, of course, you might say that it would be unlikely that he received the national median wage in both Atlanta and Los Angeles, which is probably true; nominal wages are partly determined by the local price level. However, remember that the increase in the national African-American median wage is in fact measuring the impact of people on lower nominal wages moving to cities with higher nominal wages, so at least some of the increase in the overall median wage for African-Americans was being eaten up by higher prices in the places they moved to. In order to figure out the magnitude of the price-level problem, we’d need to recalculate the real wage using a CPI basket in which local prices were weighted by the African-American population of that city, or, at the very least, calculate the African-American real male wages for various cities (using local prices). It’s a task that’s somewhat time-intensive, but when I get more time I might do it.
By the way, the attempted rebuke to Richard Rothstein falls flat regardless of these other problems when you read the full passage. Rothstein’s claim that “African American incomes didn’t take off until the 1960s” comes several paragraphs into a chapter that begins with the phrase “From the end of the Second World War”:
FROM THE end of World War II until about 1973, the real wages and family incomes of all working- and middle-class Americans grew rapidly, nearly doubling. African Americans, however, experienced the biggest growth toward the end of that period. In the 1960s, the income gap between them and white workers narrowed somewhat.
This is perfectly true; trying to debunk this with statistics that refer to the growth in incomes between 1939 and 1960 won’t do. The gap between wages paid to white and black workers expanded in the 1950s, but began to shrink in the 1960s. And racial wage gap compression in the 1940s had a great deal to do with Roosevelt’s executive order that prohibited racial discrimination in hiring for military contracters: in other words, direct government intervention over the economy, not a cultural improvement drive, was responsible for improving the incomes of African-American workers. Measuring using annual growth rates, the African-American real wage grew 7.1% from 1939 to 1947, but only 4.6% a year from 1947 to 1960 — and this isn’t accounting for the inflation effects of migration I mentioned above.
A similar misreading comes with the second snippet of Rothstein’s that Hughes takes issue with: “black workers did not share in the income gains that [white] blue collar workers realized” in the mid-twentieth century”. Again, you have to read the entire paragraph:
The construction trades continued to exclude African Americans during the home and highway construction booms of the postwar years, so black workers did not share with whites the substantial income gains that blue collar workers realized in the two big wage growth periods of the mid-twentieth century.
Rothstein’s claim is explicitly about the construction sector; and you can’t use aggregate figures to show anything about blue-collar workers, either.
- When deciding on my favourite batshit article I’m torn between the one that gave a list of eight thinkers, one of which was Albert Einstein and another of which was Brendan O’Neill, and the one which began “In an interesting way, the Starbucks incident parallels one of the historical cruces of the scientific revolution. In 1613, when Galileo published his most important findings on planteary motion…” You’ll have your work cut out to persuade me that there’s an uninteresting way in which those two incidents could be parallel. [return]
- It’s also worth pointing out that the Census data that Riley and Hughes used to get these figures doesn’t actually break down income for African-Americans; it merely lists ‘white’ and ‘non-white’. The number of non-whites who were not African-American was relatively small, but would have been larger if Latinos had been counted separately from other whites. [return]