Understanding causal links between economic policy and variables like GDP or inflation is challenging because macroeconomists cannot run lab experiments. Sargent assumed that households, firms, and policymakers form ‘rational expectations’ about the future and developed procedures to estimate structural parameters that stay constant even when policy rules change. Sims, on the other hand, used vector autoregression (VAR) to identify independent policy shocks from historical data and to show their time paths through impulse-response functions. His approach helps us decide whether changes in interest rates cause inflation to move or merely react to it. Together, their methods allow central banks and governments to measure and compare the effects of interest-rate moves or fiscal spending in a quantitative way.

Sargent’s contribution lies in embedding rational expectations into dynamic stochastic general equilibrium (DSGE) models and providing a ‘structural estimation’ framework that withstands the Lucas critique while allowing policy counter-factuals. Using tools such as the Kalman filter, he empirically studied money demand during episodes of hyperinflation and the gradual disinflation of the 1970s. Sims, in his landmark 1980 paper “Macroeconomics and Reality,” criticized large simultaneous-equation models and introduced VAR as a flexible, minimally-theoretical way to capture dynamic interdependence among endogenous variables. He developed identification strategies—long-run and short-run restrictions, Cholesky ordering, and external instruments—to recover policy shocks and measure their propagation and multipliers. Modern DSGE-VAR hybrids and factor-augmented VARs that inform monetary policy trace their intellectual roots to these two scholars.

Sargent formalized forward-looking difference equations arising from time-varying policy rules in linear Ricardian-Lucas DSGE models, casting them in state-space form amenable to ML and GMM estimation. His framework underpins learning algorithms such as E-stability and adaptive learning that are central to analyses of fiscal feedback and debt sustainability. Sims overlaid a structural matrix on the finite-order VAR’s Wold representation, reducing the identification problem to selecting orthogonalized innovations. Long-run restrictions (à la Blanchard-Quah) and external-instrument approaches (e.g., Montero-Musso) are descendants of his scheme. He further imported Bayesian estimation and MCMC sampling, codifying prior design—including the Minnesota and Sims-Zha priors—for high-dimensional systems. These techniques constitute core infrastructure in policy institutions’ toolkits, from monetary-policy transparency reports to the IMF’s DSGE-SVAR suites.