Why do we need to drop the beginning simulated values?

I dont understand what is the purpose of ‘drop=300’ or anynumber in some .mod file compiled by other people. Why do we need to drop the first 100 values to calculate the statistics?

Your model simulation can be seen as an ergodic Markov process. Dropping initial periods assures that the model converges to the ergodic distribution and the effect of the starting values (i.e. the steady state, which is not identical to the mean of the ergodic distribution) washes out.