Depending on the research question being asked, some degree of heterogeneity is required in a model. Sometimes this modeling requires distinguishing between those who provide capital and those who work. This is obviously an abstraction, because in reality these "capitalists" may just be shareholders who also work on the labor market. In fact, they often are, and they save and invest for various purposes, like self-insurance, retirement or bequests. But making households purely capitalists and workers can sometimes prove useful in making a result emerge more clearly, as long as one is conscious of the abstraction. In some circumstances, it is useful to explain why these "classes" emerge, like differences in access to credit or in subjective discount rates. But again, these are abstractions useful for modeling.
Alberto Russo takes this abstraction very seriously. In his model, people are born capitalists or workers, which translates in households either investing in an activity with a multiplicative risk or working for a wage with additive risk. Why that is so and what should be achieved with this is left unexplained. Households face an additional risk: they randomly switch between classes, the probability depending on wealth. The model is "closed" with exogenous and distinct propensities to consume for both classes. The model is then calibrated with parameters values not related to anything observable. The simulations reveal that if one starts with everyone having the same wealth, wealth heterogeneity then emerges. Well, that was unexpected... Even back in 1993, Mark Huggett had a much better model to explain heterogeneity in wealth.