The last two times I’ve taught my quantitative economic history course, I have assigned a micro-data collection project based on the 1901 Census of Canada. All in all, this data collection was a good experience for the students given they got some direct experience collecting primary data, coding it and then analyzing it. Moreover, I now have a small data set of household heads from the 1901 Census with data on age and total earnings. The 1901 Census asked what the earnings were from an occupation or trade as well as any extra earnings from other than the chief occupation or trade. As a result, I have been able to construct age-earnings profiles that unlike modern ones show rather steep declines in earnings after the mid-forties. Modern profiles show relatively moderate declines after age 50.
There are 3,357 individuals in this small data set: 91 percent are male; the average age was 44 years with a range from 15 to 91 and average total earnings were $488.96. About 72 percent reported non-zero total earnings. Since it is not always possible to ascertain if zero reported earnings actually means zero or if there was non-compliance, I have decided to drop the zero earnings from the analysis. The data is spread out over urban centers across Canada as the accompanying Figure 1 shows but much of the data is from Ontario. When the earnings are ranked by urban center, there is a strong east to west gradient with earnings generally higher in the west and lower in the east (See Figure 2).
The earnings-age profiles are estimated using LOWESS and are quite interesting. Figure 3 plots the overall earnings-age profile for only earnings greater than zero as well as a profile including the zero earnings and they show similar peak ages and similar rates of decline. Once age 40 was reached, earnings began to decline quite rapidly. Figure 4 plots the profiles (only for earnings>0) by major religious affiliation and they show quite different levels of income across the denominations though rapid rates of decline after peak earnings. However, the peak seems to vary a bit by group. Catholics, Baptists, Methodists and Anglicans all reach peak earnings at about 40 years of age but Presbyterians do so at about age 50.
Finally, Figure 5 plots the profiles again for a few of the occupational categories. These five categories account for nearly 50 percent of the people in the data set with general laborers the largest at 15 percent of the individuals. In terms of peak incomes reached, professionals and transportation workers seem to do the best. Professionals reach their income peak at about age 60 whereas transportation workers do so at age 50. Individuals employed in manufacturing, building trades and general labor reach their peaks at or just before age 40. The high income level in transportation (which includes utilities and communications by the way) seems a bit odd relative to professionals (teachers, doctors, lawyers, etc..) but then transportation was a big growth sector in the nineteenth and early twentieth century as railway infrastructure was being put into place.
Of course the next step is regression to estimate earnings-age profiles while controlling for many of these characteristics. I’m looking forward to the completed student term papers...
UPDATE: December 5
In response to Donald's query about the age distribution of the individuals in the data, see Figure 6 below.