Warning, this post is a bit cynical. Okay, it is a LOT cynical. Sorry.
I suppose this is obvious but the traditional way of increasing prices is to lower the supply. If you want SWE's to get more money, first pass a rule that a SWE has to have some EU certification, and make the certification something reasonably expensive and not easy to get, say a certificate than can only be granted by accredited authorities after some number of years of study/testing/work.
Back in the day that was the role of guilds. They trained and tested individuals and at some point granted and individual "master" status. Consumers paid a premium for goods and services provided by a "master ."
These days, 6 weeks of "coding boot camp" or maybe spending your days hanging out on mailing lists and watching youtube videos can give you the illusion of being a "SWE." I say it is an "illusion" because while a person doing this can write code it doesn't really mean they can write code that will be durable. Much like someone who considers themself to be an automotive engineer because they grew up doing their own car maintenance.
That said, there are a LOT of programming jobs for which someone who learned coding casually is perfectly appropriate. Just as there are jobs where such people would be completely inappropriate. Unfortunately we call both sets of people "SWEs" when the job requirements are quite different.
An interesting exercise would be to survey jobs and salaries for software engineers an see if you could correlate salary with job responsibility. For example, does a web site "front end" engineer make more or less than a "operations/infrastructure" engineer? Does an engineer writing embedded software for modem chips in phones make more or less than an engineer writing embedded software for a toy. Do engineers writing software for vehicle control systems make more or less than software engineers writing drivers for USB peripherals?
My point here is that because software is eating the world "SWE" has grown as a category to span jobs that are trivial and those that require years of study and experience to do.
I would not be surprised at all if the "simple" jobs were much more prevalent than the "hard" jobs and so your dataset's salary numbers would be more interesting to look at as percentiles rather than as an overall average.