The paper can be found at Saez's website: The Evolution of High Incomes in Northern America: Lessons from Canadian Evidence .
Their evidence for long term trends comes from income tax returns. Essentially Saez and Veall have looked at income tax returns for upper income persons in Canada, and compared the tax income estimates to overall estimates of national income from other sources. The limitations of the data prevent them from saying anything about the details of income changes below the 90th percentile.
The figure shows the proportion of income accruing to top income percentiles (90-95, 95-99, and 99-100). The data for the 90-95 percentile grouping only becomes available after the World War.
Saez and Veall see interesting things in the diagram:
- There are strong counter-cyclical movements in the proportion of income accruing to the top percentiles prior to the World War.
- The top income shares dropped during the World War.
- The top income shares were fairly stable from the War to the mid-seventies.
- The income share accruing to the very highest level rises strongly after '75.
Why was the top income share counter-cyclical before the war?
This evidence suggests that well-compensated employees formed a very important fraction of the top 5 percent of income earners, and probably the overwhelming majority of the P95–99 group. If wages are nominally rigid in the short run, this can explain why the P95–99 share is so clearly counter-cyclical in Canada’s prewar period, as sharp downturns of the prewar period were associated with sharp deflations.
Why did the top income share decline during the war?
This fall can be explained, in part, by the fiscal shock in the corporate sector. As part of financing the war, Canada substantially increased taxes on corporations. Moreover, corporations reduced their payout ratios during the war because of the high demand for investment, and perhaps also to avoid the personal income tax, which imposed extremely high marginal tax rates (in excess of 90 percent) on the highest incomes. Hence very top incomes, composed primarily of dividends, declined during the war. The shares of income groups P90–95 and P95–99, composed mostly of well-compensated employees, also fell. Saez and Veall (2003) confirm these results by showing that salary earners gained significantly relative
to nonsalaried employees in terms of employment and compensation during the downturns of 1920–1921 and the Great Depression, but lost significantly during World War II
What about the increase in the top income share from the mid-seventies? It could have a technological explanation. But look at the similarity between top income movements in the U.S. and Canada:
There is something else, too. The sources of upper level income have changed considerably over the years. In recent years, "the self-employed business proprietors and capital income earners have been in large part replaced by highly compensated employees at the very top of the income distribution."
With this background in mind, Saez and Veall like another explanation for the recent increase in the share of income received by the top percentile:
A second possible explanation might be competition for highly skilled executives driven by the U.S. market. Canadian executives and other professionals can relatively easily move and find jobs in the United States as part of what is sometimes called the brain drain. Therefore, Canadian firms might attempt to retain their best-paid employees by increasing their salaries.