Wagle (W. Michigan) -- Food stamps increase recipient incomes. Food stamps also reduce work hours supplied by families. Given that, food stamps tend to increase the likelihood of families being poor while increasing incomes. It may help non-poor families more than poor families. "The role of FSP on food and economic security cannot be over-estimated."
Mandell (UMBC) -- "Fuzzy" regression discontinuity to estimate SNAP on food insecurity and obesity -- Virtually no one above the cut off uses SNAP, contrary to Wagle's findings. Most estimates show SNAP increasing food insecurity, with no effect on obesity.
Discussant 2 (Tufts) -- The time to consider good instrumental variables is while USDA et al are preparing their surveys. Not "what do we happen to have that could do," but "what could we create that would tell us what we want"?
From my question: Eligibility for food stamps is based on all related people living in the household, but you can report income separately for younger individuals. You can therefore have data points where a "1-member-household" has an income too high for eligibility, but the larger family unit living together does qualify. There are a lot of data points also well above the maximum benefit. Most believe that is reporting error.
Q -- At the external margin, above the income cutoff makes you eligible for a little bit of food stamp which is easier to funge as a family, so regression discontinuity may not actually get you what you're looking for. An alternate interpretation is that it's a measure of selection pattern rather than the effect of just a little bit of food stamp.
Great recommendation: look at the selection effect of people leaving SNAP instead of entering.
Discussion of the other paper on teenage pregnancy below the fold.
Lopoo (Syracuse) on labor and delivery complications in teenage mothers -- If you imagine an "optimal" trajectory of a woman going through grad school, working several years, meeting someone, eventually marrying, and only a few years after all that having kids, that puts fertility starting in mid-30s. And we've been heading there. A lot of our social programs try to delay first birth. Has evolution totally mismatched ability to bear a child and ability to raise a child well? Teenage births look worse than average, but better than first births of older mothers.
Discussant 1 (USDA) -- Is it strange that most people only had one negative pregnancy outcome, rather than multiple? Are hospitals marking only one when there are multiple? If teens go to lower quality hospitals, their better performance on first births is even stronger.
Q -- Is it right to hold other factors constant? Drug use, poor education and income, etc. If that is part of what makes a teenage mother a teenage mother, should we hold it constant? A - If I could get wealthy teenagers to have kids, I would see really dramatic differences in the effect. You don't see as much of a change when the other factors are thrown in. So they answer different questions.