Economists test theories with micro data, information on individual people or households. Real-world data, however, are often biased—wages are observed only for those who work—and many decisions are discrete, like taking a job or not. Heckman devised the “Heckman correction” to remove selection bias, while McFadden formulated the “conditional logit model” to turn discrete choices into equations. These tools are now staples for evaluating labour-market programs, designing traffic systems, and tackling many other social issues.

Microeconometrics links economic theory (utility maximisation, full information) with statistical inference. Heckman formalised an unobserved participation equation with a Probit-type model; estimating participation probabilities in stage one and inserting the inverse Mills ratio (λ) in stage two corrects heteroskedastic and correlated errors caused by non-random sampling. McFadden assumed an extreme-value distribution of random utilities and showed that the probability that individual i chooses alternative j becomes exp(Vij)/Σk exp(Vik), thus deriving the multinomial logit model directly from microeconomic principles. Both frameworks have been extended to panel data and simulation-based estimation (e.g., MCMC, simulated moments). They now constitute core methodology in policy evaluation, industrial organisation, marketing research, and many other fields.

Heckman (1974, 1979) solved the identification of E[y*|x,s=1] when outcomes are observed only in a selected regime, assuming joint normality of (u, v) and supplying both FIML and two-step (Heckit) estimators. He generalised the framework to duration and programme-evaluation contexts, introducing LATE and MTE concepts that bridge experimental and non-experimental designs. McFadden (1974) derived the multinomial logit from a random-utility maximisation model with IID EV1 errors, proposed the Hausman–McFadden test for IIA, and developed nested and hierarchical GEV structures to accommodate correlated alternatives. His work on mixed logit, simulated MLE, and MSM established feasible estimation of high-dimensional integrals, paving the way for large-scale demand analysis. Collectively, these contributions are now canonical in econometric theory and structural counterfactual policy analysis.