The economic outcomes of sons in rural parts of Iowa were less strongly linked to the economic outcomes of their fathers than their peers in the cities in the decades before the Second World War. What’s more, intergenerational mobility was much higher for the grandchildren of foreign-born inhabitants of the state than for the grandchildren of the American-born inhabitants, suggesting very fast rates of economic assimilation.
These are among the findings of new research by James Feigenbaum on the links between the education, earnings and occupations of one generation and the next. The study, which is published in the July 2018 issue of the Economic Journal, matches fathers from the Iowa State Census of 1915 to their sons in the 1940 Federal Census, the first state and federal censuses with data on income and years of education.
The analysis reveals that rates of intergenerational mobility varied in intriguing ways, perhaps a reflection of the changing economic landscape in the early twentieth century. Overall, they suggest that there was more mobility in Iowa – and perhaps the United States as a whole – then than there is today.
Was intergenerational economic mobility high in the early twentieth century in the United States? Intergenerational economic mobility measures how correlated children’s outcomes are in adulthood to their parents’ outcomes a generation before. To construct mobility estimates, one needs to observe the economic outcomes – education, earnings, occupation – of two generations with links between parents and children in each.
Comparisons of mobility across time or space are complicated by the constraints of what data was collected historically: One linked sample may record only occupation, another labour earnings or another education. Are the estimated rates of mobility comparable or driven by the types of variables considered?
The new study constructs a new historical linked sample in which it is possible to measure rates of intergenerational mobility in a variety of ways. Across many measures of intergenerational mobility, the research documents broad consensus that rates of persistence were low in Iowa in the early twentieth century. The author matches fathers from the Iowa State Census of 1915 to their sons in the 1940 Federal Census, the first state and federal censuses with data on income and years of education.
With this linked sample, it is possible to estimate rates of intergenerational mobility in the early twentieth century based on earnings, education and occupation, as well as name-based methods, which extract economic and social information from the names that parents chose to give their children.
Unfortunately, because parents and children are linked from one census to another by name, state of birth and year of birth, it is necessary to exclude women from the sample, as most daughters of Iowa are likely to change their surnames between 1915 and 1940 on marriage.
Rates of mobility vary in intriguing ways within the sample, perhaps a reflection of the changing economic landscape in the early twentieth century. Rural sons from Iowa –many growing up on farms hard hit by the Great Depression – had more intergenerational mobility than their urban peers. In other words, the economic outcomes of rural sons were less strongly linked to the economic outcomes of their fathers before them.
In addition, while much of the sample is made up of American-born sons and American-born fathers, there is a lot of variation in the nation of birth of the grandparents represented in the sample. One third of the sons in the sample had four foreign-born grandparents and another third had four American-born grandparents. Mobility was much higher for the grandchildren of the foreign-born than for the grandchildren of the American-born, suggesting very fast economic assimilation rates in the early twentieth century.
Relative to comparable contemporary estimates, these results suggest that there was more mobility in Iowa in the early twentieth century than there is today. But these low persistence rates come with a few caveats.
First, how representative of the nation was Iowa in the early twentieth century? Iowa is ranked in the middle of the states in terms of population density, urbanisation and rural share, as well as sex ratio. But in other ways, Iowa was exceptional, starting with a very high level of education – it was not randomly the first state to collect data on years of education among its population. In addition, Iowa was a more agricultural economy than the nation as a whole, and it was more than 99% white in both 1910 and 1920.
Second, the linking procedure that the study uses to connect one historical census to another may introduce erroneous matches, and linking fathers to the wrong son will falsely suggest high rates of mobility.
Still, as the author documents, the high rates of mobility that he finds are unlikely to be explained away either by Iowa’s uniqueness (the historical estimates are compared to a contemporary Iowa-like sample) or by noise in the data (the historical data cannot be made any cleaner, but when the contemporary data are made noisier, the comparison still holds).
‘Multiple Measures of Historical Intergenerational Mobility: Iowa 1915 to 1940’ by James Feigenbaum is published in the July 2018 issue of the Economic Journal.