There is no reason to doubt the basic validity of the indicator for skill under utilisation.
There is undoubtedly some noise in this variable, as in every measure obtained through survey research, However, it seems implausible that respondents who are working below their own level to the extent that this has a strong negative effect on their wages would fail to report that their skills were not being fully utilised, if this was in fact the case.