From Nature news.
Calcagno, in contrast, found that 3–6 years after publication, papers published on their second try are more highly cited on average than first-time papers in the same journal — regardless of whether the resubmissions moved to journals with higher or lower impact.
Calcagno and colleagues think that this reflects the influence of peer review: the input from referees and editors makes papers better, even if they get rejected at first.
Based on my experience with economics journals as an editor and author I highly doubt that. Authors pay very close attention to referees’ demands when they are asked to resubmit to the same journal because of course those same referees are going to decide on the next round. On the other hand authors pretty much ignore the advice of referees who have proven their incompetence by rejecting their paper.
Instead my hypothesis is that authors with good papers start at the top journals and expect a rejection or two (on average) before the paper finally lands somewhere reasonably good. Authors of bad papers submit them to bad journals and have them accepted right away. Drew Fudenberg suggested something similar.