Stratified sampling assumes no correlations among variables while Optmization does so you need to take into account correlations and COV, that makes optmization more complex to calculate than stratified sampling. In relation to Optimization :
Yes. Less “transaction” cost to be precise. Optimization needs constant rebalancing even when there are no index changes/dividend reinvestment because of the need to match risk sensitivities over time.
Q2 2016 CFA PM mock. The main reason Stratified was preferred over the optimization in the question was investor’s desire to reduce transaction costs.