Nicholas M. Kiefer

Nicholas M. Kiefer is the Ta-Chung Liu Professor in Economics and Statistical Science at Cornell University. Kiefer works primarily in econometrics and statistics with applications in financial economics, credit scoring, risk management, operational risk, anti-money laundering and fair lending issues in banking.

Professor Kiefer's interest in combining theoretical economics and statistics has led to structural empirical models in a number of fields.  His early work on program evaluation proposed methods of sorting out employment effects from wage effects of training in determining earnings.  This was followed in joint work with G. Neumann by the first structural estimation of job search models of the labor market.  This led to research on specification and estimation of equilibrium search models, in work with Neumann, B.J. Christensen and others. In finance, Kiefer has estimated models of market microstructure, with special emphasis on the roles of information and liquidity in determining price adjustments.  The PIN statistic invented in this work (joint with D. Easley and M. O'Hara), an estimate of the probability of an informed trade, is now in wide use in empirical finance. Derivative pricing and the incorporation of options data in estimating financial models is the topic of joint work with Christensen. Theoretical work on the optimal accumulation and valuation of information (joint with Easley, Nyarko, and Bala) showed that full information accumulation is not always optimal and provided guidance for optimal experimentation. Current work in finance includes specification and evaluation of credit scoring models, estimation of default probabilities for financial institutions and portfolio risk measurement and management. 

In econometrics, Kiefer's work has dealt with heterogeneity, in the form of regime shifts in switching regressions, individual effects in nonlinear models, and linear and nonlinear panel data models.  Most of the work is highly structural and likelihood based, although recent work with Christensen treats estimation of options models by simulation from the "risk-neutral measure," followed by method of moments.  Other work developed models for duration data, with special attention to unemployment duration and more recently time to default. Work with Christensen focuses on separate inference and the geometry of estimation.  Other recent work with T, Vogelsang and H. Bunzel focuses on robust testing in time series models, introducing a novel method of approximating the distributions of test statistics (now referred to as “fixed-b” or “KVB” asymptotics). Recent work with H. Choi extends these results to nonnested testing.  Other work with Choi emphasizes the geometric approach to bias correction and high-order asymptotics.  Recent work on inference about small probabilities has been applied to default probability estimation. New work with J. Racine links Bayesian methods and kernel based estimation methods.

Kiefer is a Fellow of the Econometric Society, and a past recipient of the John Simon Guggenheim Memorial Fellowship. His point of view is reflected in the book with B.J. Christensen Economic Modeling and Inference