Lagged variables are used in time series analysis to account for the influence of past values on current observations, helping to model temporal dependencies. They are crucial in econometrics and forecasting, allowing analysts to capture dynamic relationships over time and improve predictive accuracy.