This yard-sale model isn't really how the economy works though, people don't continuously bet 20% of their net worth. It's something similar going on though, I think a lot of these variables would seem to be explained by considering the lognormal distribution for personal wealth.
In a normal distribution, the shape of the distribution comes from a "random walk" left and right from a large number of steps of varying size.
In a lognormal distribution, on the other hand, the random steps are not additive but multiplicative: e.g you multiply the previous figure by a (Gaussian) random variable many times.
This seems to reflect economic reality that people make decisions proportional to the scale of their current wealth. If I make 10k, it would take 2k extra to entice me to a different job. If I make (or lose) 10% on an investment, etc. It's all multiplicative.
The lognormal distribution also has a fatter "right tail" than a Gaussian, which is what we see IRL.
I'm not sure it's intended to reflect the real economy. It's a useful model that shows how success and wealth inequality can arise for reasons of pure luck, and can have nothing to do with meritocracy. I'm not sure any of the points you raise really change the fundamental dynamics of what's shown here.
In a normal distribution, the shape of the distribution comes from a "random walk" left and right from a large number of steps of varying size.
In a lognormal distribution, on the other hand, the random steps are not additive but multiplicative: e.g you multiply the previous figure by a (Gaussian) random variable many times.
This seems to reflect economic reality that people make decisions proportional to the scale of their current wealth. If I make 10k, it would take 2k extra to entice me to a different job. If I make (or lose) 10% on an investment, etc. It's all multiplicative.
The lognormal distribution also has a fatter "right tail" than a Gaussian, which is what we see IRL.