P. A. V. B. Swamy
From Wikipedia, the free encyclopedia
P. A. V. B. Swamy | |
|---|---|
| Born | July 3, 1936 |
| Died | January 27, 2026 (aged 89) Rancho Palos Verdes, California, U.S. |
| Alma mater | University of Wisconsin–Madison Andhra University |
| Known for | Originating random coefficient estimation |
| Scientific career | |
| Fields | Statistics Econometrics |
| Institutions | Federal Reserve System Ohio State University SUNY Buffalo |
| Doctoral advisor | Arthur Goldberger |
Paravastu Aananta Venkata Bhattandha Swamy (born July 7, 1936 – January 27, 2026)[1] was an Indian-born statistician. For fifty-six years, Swamy's research has focused on econometric issues.[2]
After earning a B.A. in economics from Andhra University, India, in 1956, both an M.A. in economics and a M.S. in statistics from the same university in 1958, Swamy attended the University of Wisconsin–Madison. He finished his Ph.D. dissertation on random coefficient estimation under the supervision of Arthur Goldberger in 1968.
Career
In 1967, Swamy joined the economics faculty at SUNY Buffalo as assistant professor, when he published his first article based on his dissertation on random coefficient estimation,[3] followed in 1970 by another on the subject,[4] and in 1971 by a much cited monograph,[5] the latter cementing his reputation as an authority on random coefficient estimation. In 1972 he became professor at Ohio State University. In 1974, he joined the Federal Reserve System, where he worked first as an Economist and then as a Senior Economist in the Division of Research and Statistics until 1995, when he joined the Office of the Comptroller of the Currency as a researcher, followed by an appointment to the Bureau of Labor Statistics in 2002, from which he retired in 2009.
As summarized by Stephen G. Hall, Nobel Laureate Lawrence Klein, George S. Tavlas, and Arnold Zellner in a special issue of the journal Economic Modelling[6] dedicated to his contributions up to 2010, Swamy's research agenda has been devoted to some of the most pressing issues in econometrics. These include (1) a prevailing ignorance of true functional forms in economic relationships, (2) the presence of unobserved variables underlying the need for error terms, (3) the difficulty of obtaining accurate estimates of a model's parameters if error terms and included variables are correlated, which they must be, given the presence of unobserved variables, and (4) the problem of errors in measurement. With the collaboration of a number of long-time colleagues, including Jatinder S. Mehta and George S. Tavlas, Stephen G. Hall, Peter Tinsley, I-Lok Chang, and Peter von zur Muehlen, Swamy's singular contribution to econometrics has been to devise a methodology, set down in numerous publications, that has enabled the profession to address these issues in a coherent and systematic manner.[6]
In 1975, and his long-time co-author Jatinder S. Mehta, introduced and proved a theorem, later dubbed the "Swamy-Mehtha Theorem" and subsequently proved by Clive Granger,[7] stating that any nonlinear functional form can be exactly represented by a model that is linear in variables but that has time-varying coefficients. The implication of this result is that, even if one does not know the correct functional form of a relationship, one can always represent this relationship as a time-varying coefficient relationship and thus estimate it.[8]
Also in 1975, Swamy and Mehta extended their methodology to include crossectional data in a paper entitled "Estimation of Linear Models with Time and Cross-Sectionally Varying Coefficients"[9]
In 1976, to render estimation of time-varying stationary stochastic coefficients models operational, Swamy and Peter A. Tinsley published a paper on linear prediction and estimation.[10] In 1985, in an examination of the probabilistic-logical foundations of econometrics, Swamy and Peter von zur Muehlen published a paper,[11] reprinted in a volume on the foundations of probability, econometrics, and economic games,[12] which developed themes that would animate much of his later work, including a paper on the nature and testability of causality.[13]
Applying standard principles of probability theory and lessons learned from random coefficient modeling, Swamy co-authored a paper with James R. Barth and Peter A. Tinsley that questioned the validity of conventional formulations of the rational expectations postulate as a violation of the axiomatic basis of modern statistical theory by confounding ‘objective’ and ‘subjective’ notions of probability.[14] In a subsequent paper, Swamy and George S. Tavlas derived conditions under which the predictions generated from time-varying coefficient models coincide with the predictions generated from the relevant economic theory thereby fulfilling the rational expectations postulate.[15]
A 1988 publication by John W. Pratt and Robert Schlaifer on the interpretation and observation of laws,[16] became a compelling leitmotif for most of Swamy's subsequent work on estimating economic models.[17][18] To implement Pratt and Schlaifer's and Robert Basmann's compelling definitions of an economic law and to solve the problem of uniqueness arising from a correlation between the error term and included variables in a regression, Swamy resorted to a novel practice of modeling regression coefficients themselves as functions of variables external to the model, called "coefficient drivers."[19][20][21]
