a) will a leptokurtic distribution report a higher or Lower VaR?

b) Will a negative skewness distribution report a higher or lower VaR?

Notes said: With a Leptokurtic (fat tails) distribution, VaR will underestimate the loss and its associated probability.

Shouldnt it be overestimated? Since a Leptokurtic distibution tends to be have a lower standard deviation, thus VaR based on standard nommal distribution will be farther out in the tail. Shouldnt this be over estimated loss? Since VaR is an absolute value of loss?

Anyone can help explain this?