This one has actually been in the works a while. I've presented this several places, including an IZA conference. It didn't have much luck in the journals for some reasons that weren't entirely surprising to us, and I've been the one moving it along so I wanted it out in some form before I left. The abstract:
"Growing up poor strongly predicts poverty and poor adult outcomes. This study explores two primary reasons poverty may persist across generations: risk behavior in adolescence and dropping out of high school. Results suggest that risk behavior and dropping out help perpetuate poor economic outcomes for children from single-parent families but are less important for children who grow up in low-income families. The findings suggest that policies directed at reducing youth risk behavior and dropping out can improve economic outcomes when targeted to youth from single-parent households."
It's kind of got a kitchen-sink feel to the regressions, but it tries to do some interesting things with direct and indirect effects of poverty. Often these mechanisms are a black box for economists.
This is an off-shoot of a project funded by the Department of Health and Human Services. It uses the 1997 National Longitudinal Survey of Youth, which is a really great dataset.