The carte-blanche principle implies that food stamp recipients would be better off if they were
given cash instead of an equivalent amount in food stamps. I estimate the cash-equivalent value
of food stamps and the lowest price a recipient would accept to sell her “extra” food stamps on
the underground market. I estimate that between 20 and 30 percent of food stamp recipients
spend less on food than their food stamp benefit amount if they receive cash instead of stamps,
and therefore would be better off with cash. Using a theoretical model I present and data from
experiments conducted in two states, I estimate that on average “distorted” food stamp recipients
value their total benefits at 80 percent of their face value. Aggregating over recipients, the
annual deadweight loss associated with the food stamp program is one-half billion dollars. Food
diary data indicate that providing cash instead of stamps causes some distorted recipients to
decrease their food spending – especially on soda and juice – but has no negative consequence
for nutrition. As predicted by theory, inframarginal food stamp recipients do not alter their
behavior if they are given cash instead of food stamps. Although paying in-kind benefits results
in some deadweight loss, it is thought that an underground market for the excess stamps will be
created to alleviate some of the loss. I present new survey evidence indicating that stamps trade
for only about 65 percent of their face value on the underground market.

This paper provides a long-term follow-up of students who participated in the Tennessee STAR experiment. The Tennessee STAR experiment randomly assigned l 1,600 elementary school
students and their teachers to a small class, regular-size class or regular-size class with a teacher-
aide. The experiment began with the wave of students who entered kindergarten in 1985, and
lasted for four years. After third grade, all students returned to regular-size classes. We analyze
the effect of past attendance in a small class on standardized test scores through the eighth grade,
on whether students took the ACT or SAT college entrance exam, and on how they performed on
the ACT or SAT exam. The results suggest that attending a small class in the early grades is
associated with somewhat higher performance on standardized tests, and an increase in the
likelihood that students take a college-entrance exam, especially among minority students. Most
significantly, being assigned to a small class appears to have narrowed the black-white gap in
college-test taking by 54 percent.