The proposed law does not say WHO reproduces it, merely that someone MUST be able to reproduce the results. If the EPA can point to another, independent, study which reproduces the results of the first study, it meets those qualifications.
Scientific studies often cost significant amounts of money to produce - at least, they cost significant amounts of researcher time. Unless a study is extremely controversial or you expect to get very different results, few scientists will spend the time. There is minimal new knowledge to be gained, most journals rarely publish papers on successful reproductions, and a CV that says "I did the same as Williams, the same as Jones, and the same as Mayer, and got the same results in each case" is not a career starter for a scientist, either.
Indeed, most studies that can be reproduced can be reproduced from the published papers. It's just hard work and expensive, which is why it's rarely done. Demanding reproducibility is fine, but demanding actual reproduction (as proof of reproducibility) would kill most science-based initiatives cold. Note, in particular, another law proposed by Lamar Smith that would allow NSF funding only for research that is "not duplicative of other research projects being funded by the Foundation or other Federal science agencies". Take the two together, and you have a requirement for reproduction, but deny funding to do the reproduction. Ooops - how convenient.