Why Standard Monte Carlo Simulations Understate Retirement Risk: Evidence from Indian Financial Markets

Traditional retirement simulations assume that returns are independent and identically distributed (IID). Using three decades of Indian market data, we show that this assumption leads to a systematic understatement of retirement failure risk by 0.3–1.7 percentage points. Real deposit rates exhibit strong yearly cycles and equity–deposit correlations spike during market stress, making IID models unreliable. Dependence-preserving block bootstrap simulations correct this bias and reveal that sustainable withdrawal rates fall to roughly 3.6%, 3.2%, and 2.0% for conservative, moderate, and aggressive portfolios, respectively. We also find that common simulation sizes of fewer than 10,000 runs lack the statistical power to detect these differences. The results imply that regulators and retirement planners should adopt dependence-preserving methods and larger iteration counts when projecting long-term outcomes.