I'm trying to generate a random lognormal distribution of stock returns
where the mean return is 6.5% and the standard deviation is 20%.

Previously I've done this with a normal distribution, using
NORMINV(RAND(),mean,stddev). However, since stock returns tend to be
lognormally distributed I'd like to do a lognormal distribution. This
has turned out to be tricky. When I set up
LOGINV(RAND(),ln(mean),ln(stddev)) I get the #NUM error. I believe this
is because the log of a standard deviation expressed in percent (say,
0.20) can be negative. I tried using a larger number (say, multiplying
the mean/stddev by 100 to prevent negative logs) but I'm not sure
that's right.

Does anyone know an easy way to do this? Thanks!