For a two person household in 2016 the smallest possible monthly benefit was 16 USD while the largest was above 350 USD (\citealt{Lauffer2017}, p.103).  If this treatment intensity was exploited, one might be able to derive the ATE per monthly dollar benefit on health outcomes. Going beyond the ATE, it might be possible to derive a marginal treatment effect under reasonable assumptions that could inform the debate whether food stamp awards should be more generous.

While the identification and estimation are very clean, there are some estimates that were not shown, that would have been interesting to see. Firstly, the CPS data was only used for the analysis without measurement error. Given the substantial underreporting in the CPS \cite{C.1997}  it would have been interesting to see how the ATE intervals would have been after accounting for this degree of underreporting. Secondly, the results including measurement error for the health outcomes were only shown with the two strictest assumptions. Since these were the outcomes were the MRT assumption was the least innocuous it would have been interesting to see all results here as well.