When does a small bias toward the richer trader split people into rich and poor? At the defaults (wage 3, living-cost 2) everyone gains 1 a tick, mean-wealth climbs steadily, and poor-count and rich-count stay at 0.
Try this Set wage to 2 so it equals living-cost. It takes effect live, but it's clearest from a fresh start, so press Reset if the run has been going. After 500 ticks about half the people are poor and rich-count is around 35; by 2,000, about 88 of the 120 are poor and most of them have nothing.
About this model Where it ends up
A random-exchange model of inequality: people earn a wage, pay a living cost, and trade single units of wealth with whoever is nearby, with the odds tilted toward the richer party. A rich minority and a broke majority appear readily when wage equals living-cost; with a net income they need a much stronger tilt and more trading.
Each tick, in random order, everyone takes a random step of move-step and adds wage − living-cost to their wealth (floored at 0). Then each person, with probability trade-chance percent, picks a random person within 2 patches; if that person has any wealth, two independent draws follow. The initiator takes 1 unit with probability 50 + unfairness / 2 percent if it is the richer of the two (50 − unfairness / 2 if poorer) and gives 1 unit with the complementary probability, so on average the richer side gains unfairness / 100 units per trade. Finally everyone is classed poor below half the mean wealth and rich above 1.5 times it. Only population and initial-wealth need a Reset.
This is not the model of the same name in the NetLogo library, which is a Sugarscape-style model of people harvesting grain. With wage equal to living-cost and unfairness at 0, it is close to the conserved random-exchange models of econophysics, such as Drăgulescu and Yakovenko's (2000).
