> "Capital in the 21st Century."
It's weird because prior to the 21st Century this used to be almost entirely how wealth was generated.
But I think the argument of inheritance being a major wealth driver should be simple to understand from a few base principles. This is invariant of money and is related to capital in the broader sense (i.e. resources) which is also going to be invariant to many economic systems. One of the most important ones is momentum.
Momentum seems to often not be considered in these discussions but makes discussions of trickle down economics rather insane. Let's look at a more realistic discussions. Your stock portfolio should not be bounds, full of bonds, and nothing but bonds. You have to diversify and actually even include some risky investments. We usually talk about this in percentages of investments but this doesn't actually make any sense because it's actually percentage of risk vulnerability. E.g. Elon can lose 244B of his 245B and still gave essentially the same quality of life (similarly doubling his wealth won't change anything in a meaningful way). But let's also look deeper at percentages. Let's say I make an investment of $10k (a nice size for someone early on in their career) and I'm about to do really well and get 500% return on that (killer value!), I now have $60k to play with. Now let's say another person puts a $10M investment and only gets a measly 1% return. They gained $100k so have $10.1M to play with.[0] The other problem with percentages is that you can "increase your luck" by taking on far more risk. If we both take our 10{k,M} and diversity into 1k pots and each pot has a x% chance of returning 500%, which person do you think ends up with more winners? Or if we do a fixed number of pots but divide by percent of money, you're going to find a very different expected return value.
Consider this factor when you're also considering cost of living arguments. I'll take that SV software job at $300k/yr where I can only invest 10% of my income (30k) per year over a $100k/yr job in middle America where I can invest 20% of my income (20k). COL calculators are always based on percentages and so not taking this into account (as well as a lot more abstract things) and it doesn't matter of in SV I'm spending 70% of my income on housing but 30% elsewhere if this is the end result (only focusing on building wealth for a clearer argument. Other factors matter and wealth isn't everything. But employers, you can't just use COL adjustments, it's a bad deal for the employees and they know it)
Now I said it is invariant of money, because capital is capital. Land, factories, gold, potatoes, whatever, that's all capital. It can't ever be evenly distributed, even with all good faith actors, because the value of those changes wildly and fast (value != money, its more abstract). Throw in how non-homogeneous everything is (e.g. land, housing, location are not fungible resources) and you got a crazy fucking mess. It's no wonder you can't make a successful 5 year plan, the world is chaos and the models need to be exceptionally complex. You have to embrace the chaos in some way.
TLDR: math isn't always intuitive because neither percentages nor raw values exclusively tell the whole story (don't get me started on mean, median, and variance). People love to (often unintentionally) lie by using whichever fits a better narrative. A holistic approach is needed (i.e. context) with clear definitions of objective functions and more importantly clear understandings of the metrics (a whole other can of worms that people think is far simpler than they actually are).
[0] To put it another way, if you're retiring with $10M in investments you're retiring with a passive income north of an L3 at Google. This is why John Bogle is happy with only $50m (which grew) because he understood that there's very little more money can do for him and that a simple 2% a year on that gets him a million every year and so the value is not a high score but being human. This isn't really a philosophical argument because literally capital generally follows an S curve in its utility, not linear and usually not exponential (though S looks exponential at some points).