From Fatih Guvenen, Greg Kaplan and Jae Song:
Our findings suggest large differences across sectors in the cyclicality of earnings growth. While average earnings growth (across workers in all sectors) is very similar for individuals in the top 1 percent compared with the bottom 99 percent, there are striking differences across sectors. Most of the cyclicality of top earners is driven by those in FIRE, while FIRE workers in the rest of the distribution do not have particularly cyclical earnings. Services on the hand, the largest sector, is actually less cyclical in the top 1 percent than in the bulk of the earnings distribution.
The analysis in this paper has only scratched the surface of issues surrounding the nature of the business cycle risk faced by top earners. Guvenen, Kaplan and Song (2014) use an even larger sample from the same panel dataset that allows them to study individuals in the top 0.1% and 0.01%, as well as to distinguish between several two-digit SIC industries: for example, separating health services (mostly doctors), and professional services (lawyers, engineers, accountants, various research services, etc) from the broad category of services, as well as distinguishing between finance, insurance and real estate. Such distinction is important, especially because these categories occupy much larger shares of the top 1% and 0.1% than their population share. This larger dataset also enables further exploitation of the panel dimension to study membership in the top percentiles of average earnings over horizons that are longer than 5 years, such as lifetime earnings.