I was looking at data from the 1980, 1990 and 2000 censes (as well as the ACS in later years). There is a dramatic change in the number of `chief executive officers’ (occ1990 category 4) starting in 2000. The case count changes from several hundred in 1980 and 1900 to many thousands in 2000 and onwards (the increase also happens in the percentage of occupations so it is not just a sample size issue). For example, there are 255 observations of occ1990 = 4 in the 1990 data (representing around 0.02% of occupations) compared to 12,329 in 2000 (representing around 0.74%). I couldn’t really find any explanation in the provided crosswalks. Basically, either something substantial changed in the way these are coded or there was some substantial change in people calling themselves CEO’s. I am not sure which it is.
The Census Bureau significantly restructured its occupational coding scheme in 2000, which makes comparing some values of OCC1990 before and after this restructuring problematic. Specifically, you will want to take a look at the Census Bureau’s 1990-2000 occupation crosswalk found here. Note that the 2000 Census code “001: Chief Executives” can be found under multiple 1990 Census code categories. Most notably, 461,131 persons identified as “022: Managers and Administrators, n.e.c.” in 1990 would have been identified as “001: Chief Executives” in 2000. Summing the “001: Chief Executives” population in 1990 gets you nearly 500,000 persons. An increase from 500,000 CEOs in 1990 to 1.2 million CEOs in 2000 is much more plausible than the increase you are seeing with OCC1990.
Hope this helps.