I’m particularly curious about the relative decline in black life expectancy to white life expectancy. I’ve been reading Robert Putnam’s new book, The Upswing, and he has a chart on page 204 which shows that the ratio of black life expectancy to white life expectancy in 1921 was around .875 or so, but then drops sharply to around .75 in 1924. (He cites the National Center for Health Statistics, “Death Rates and Life Expectancy at Birth” for his info).
What accounts for this dramatic drop? Was it a latent affect of the 1918 pandemic? Were black communities so disproportionately affected that it had such a huge impact on their relative life expectancy to whites? Was it some after effect of World War I? A statistical anomaly? Or maybe I’m thinking about it wrong, maybe white life expectancy rose quickly during this period and black life expectancy still rose but at a much slower pace?
Thanks! And if anyone is interested in commenting generally on Putnam’s book from a historian’s perspective I would very much welcome the comment.
I do not have a complete answer, but I would caution against over-interpreting some of the early historical data. I suspect the early fluctuations are based on early sampling issues. Between 1900 and 1933, the states used to estimate lifespans changed, and the addition of states with different relative life expectancies probably had an effect on the 'US' estimates you're seeing.
I don't know what data Putnam was using, but I don't think 'true' nation-wide data exists for that period. "Prior to 1933 mortality data were not available on a country-wide basis because the registration of deaths was so incomplete in some states as to be decidedly unreliable." Instead, for most purposes (e.g. the Census), data from a selection of states with 'good' reporting was used. These started with 10 "death registration" states in 1900 and this selection gradually grew to encompass the entire nation in 1933.
Infant mortality has a disproportionate impact on life expectancy calculations, and there was evidence to indicate that the "registration of infant deaths [was] less complete than the registration of deaths at older ages". So life-span estimates would have been dependent on both the death registration data and the birth registration data.
Pages 96 and 97 of this publication identify when states were added to the death and birth registrations. The selection of states used to calculate these statistics was changing pretty drastically during this period. For example, in 1923-25, Florida was added to birth registration, Alabama was added to the death registration, and Iowa, North Dakota, and West Virginia were added to both the birth and the death registrations.
If you look at the lifespans on this site (Appendix B), you will notice some decidedly dramatic ups and downs during this period, and I suspect that at least part of these fluctuations were due to the changing sample of states used.
Lifespan estimates:
| Year | White | Black |
|---|---|---|
| 1921 | 61.8 | 51.5 |
| 1923 | 58.3 | 48.3 |
| 1925 | 60.7 | 45.7 |
| 1927 | 62.0 | 48.2 |
| 1929 | 58.6 | 46.7 |
As a more minor note, it was also not certain that adding a state to the death and birth registration data would add data for all races. For example, these estimates of lifespans (p. 45) stated that data from Massachusetts in 1920-21 was based only on White people.