In general, diffusions or solutions to stochastic differential equations do not admit explicit formulas. To obtain quantitative information, one often resorts to Monte Carlo simulation. The idea is to discretize time and approximate the value of the diffusion at discrete time steps. Except for special cases, this operation will introduce discretization error, and thus the resulting estimates are usually biased. The variance reduction techniques we have discussed so far can also be brought into play. However, they will only reduce the variance of the estimate and will not affect the bias.