I have not seen the Carroll paper, but actually all the description in the article fully applies to the original Krussel and Smith paper, which I know very well

I have not seen the Carroll paper, but actually all the description in the article fully applies to the original Krussel and Smith paper, which I know very well. One problem here is that actually, heterogeneity does not really matter much for the aggregate! In other words, the model may be more realistic, but the aggregate behavior -- such as consumption and economic growth -- tends to be relatively unaffected by inequality. This happens because of two things: (1) the poor people, exactly because they are poor, do not consume enough to create a meaningful economy-wide impact, and (2) people really hate being poor, and save more during good times so that they don't get hit with the constraints in bad times. The second one can be to some extent accounted for by allowing a sizable fraction of people to be somewhat irrational, and by getting rid of the "infinitely lived" consumers; the first would stay though. I am curious about what is exactly done in the Carroll paper to make it matter.