Thomas Rothenberg, Distinguished Fellow 2015
Thomas Rothenberg has made fundamental contributions to econometric theory. He was a leader in the theory of simultaneous equations inference and remained a leader as econometric theory turned toward other nonlinear models.
At a time when econometricians were analyzing identification model by model, he was the first to take a more general approach. In a 1973 Econometrica paper, he showed that in a very broad class of models, local identification depends on the information matrix: if it is non- singular in a neighborhood of the true parameter value, the model is identified in that neighborhood.
Rothenberg and Leenders, in their 1964 paper in Econometrica, established the asymptotic efficiency of the linearized maximum likelihood estimators, showing the equivalence in terms of asymptotic efficiency of the full-information maximum likelihood, three stage least squares, and the linearized maximum likelihood when the covariance matrix of the structural unobserved disturbances is unknown. This equivalence result justified using the computational simpler linearized methods for estimation of simultaneous equations, but was also applicable more widely, simplifying likelihood-based estimation in nonlinear models generally.
Thomas Rothenberg has also been influential in the development of methods that allow one to rank estimators and tests that are equivalent in terms of first order asymptotics. The first-order asymptotic approximations might be too far away from their finite sample behavior. In a 1984 Econometrica paper, he developed stochastic expansions to analyze the properties of various estimators and tests, showing they are equivalent to first order but have very different higher- order properties.
Thomas Rothenberg made also significant contributions to Bayesian econometrics. In a 1963 paper, he combined a classical simultaneous equation model with a loss function representing the behavior of the policy-maker. He then analyzed the effects of different priors on the posterior parameter estimates under alternative assumptions.
With Graham Elliott and James Stock, he developed commonly used efficient tests for autoregressive unit roots models.
Thomas Rothenberg is Professor (Emeritus) of Economics at the University of California – Berkeley. He has been a Fellow of the Econometric Society since 1977.