This article makes the common mistake of confusing correlation with causation. Some data points might correlate but that doesn't imply they cause each other.
I would think that what is most likely is that good income family students (who can afford summer schools) probably go to better school, which is by itself self-reinforcing.