Adjusting the Levels.fyi End of Year Pay Report
for COL and Taxes
Levels.fyi recently released their end of year pay report for 2020. It was a very insightful report and I deeply appreciated the section outlining pay by level across employers in particular. That being said, one issue I saw was that the pay by location section did not account for cost of living and taxes. This is understandable given how difficult it is to compare cost of living across jurisdictions and that estimating these is deeply personal; however, it’s still valuable in my opinion to estimate cost of living and taxes in order to consider the merits of one particular location over another.
To provide an adjusted chart of Levels.fyi’s pay by location section, I took cost of living data from Numbeo (“A single person estimated monthly costs are X without rent.“), as well as the average rent both inside the city center and outside the city center. I also calculated the taxes for each jurisdiction using Neuvoo or government tax calculators when they were usable in English. From this information, I used the formula “salary - cost of living * 12 - rent * 12 - tax” to generate the following charts:
There’s nothing too insightful here. American cities are generally better than non-American cities for software engineers. Although, I am surprised by the performance of Zurich, Tel Aviv, and Vancouver especially given that not too long ago the municipal government of Vancouver bragged about the low salaries software engineers make in their city. I am also surprised by the under-performance of Portland and Boston considering that I thought any American cities would outperform any non-American city but Portland and Boston act as exceptions to that.
That being said, it should be noted that Levels.fyi is certainly subject to sampling bias as there is a certain population of software engineers willing to put their salaries on Levels.fyi and these software engineers presumably earn higher salaries than the software engineers who don’t put their salaries on Levels.fyi. The phenomenon of sampling bias should also apply to Numbeo and it’s possible that Neuvoo isn’t as accurate as I would like it to be. Furthermore, average cost of living isn’t necessarily reflective of how you would live in particular city. Certain lifestyles are more expensive in some cities than others and this is something that should be kept in mind.