Suppose that the daily geometric returns of stock (i.e. logreturns) are distributed as normal with mean $2\%$, standard deviation of returns = $4\%$, and the stock has a current value of $10 million.
Given the confidence level of $95\%$, I found the $z$-value to be $1.645$, hence worked out the
%VaR = $0.02 + 0.04 \cdot 1.645 = 0.0858$
and the
$VaR (\text{in dollars}) = V_{pf}*(1-\rm exp{(0.02 + 0.04 \cdot 1.645)}) = -0.895$
However, the answer I am supposed to be getting is $4.49\%$ or $0.449$ million respectively. May I know what I am doing wrong?