Gary King

Overview

Gary King is a political scientist and statistician at Harvard University's Institute for Quantitative Social Science (IQSS). His research spans causal inference, statistical methodology, and the application of formal quantitative methods to social science and public policy problems. In the context of this wiki, he is relevant for his collaboration with Samir Soneji on formal mortality forecasting for Social Security solvency.

Key Contributions / Features

• Pedroza and King (2002) — "Revisiting Demographic Methods." Reformulated Lee-Carter as a state-space model and proved (a) the SVD step is MLE under Gaussian errors, (b) standard LC forecasts are technically wrong (ignore last residual), (c) LC point forecasts ≈ per-age random walks. Bayesian Gibbs-sampler implementation still fails out-of-sample, confirming that the constant $\beta_a$ assumption is the binding structural limitation. See Lee-Carter Model.
• Soneji and King (2012) — "Statistical Security for Social Security." Demography 49(3): 1037–1060. Co-developed the Bayesian hierarchical mortality forecasting model (building on Girosi and King 2008) that incorporates demographic smoothness priors and risk-factor covariates. Provides open-source software and replication materials for both the SSA's existing methodology and the alternative approach. See SSA Mortality Forecasting.

Related

• Claudia Pedroza
• Samir Soneji
• Office of the Chief Actuary
• SSA Mortality Forecasting

Sources

• Pedroza and King 2002 — Revisiting Demographic Methods
• Soneji and King 2012 — Statistical Security for Social Security