Agustín Maravall

Maravall spent his childhood in Paris, then moved to Madrid and finished secondary school at the Colegio Estudio. He completed a doctorate in agricultural engineering at the Technical University of Madrid, and worked for several years at the Spanish Ministry of Agriculture. With a Fulbright-Ford fellowship he came to the United States and completed a Ph.D. in economics at the University of Wisconsin–Madison. In 1975, he moved to Washington, D.C. to work as a staff economist in the Research Division of the Federal Reserve Board of Governors. In 1979, he returned to Madrid as a senior economist in the Research Department of the Bank of Spain, and in 1989 he moved to Italy as Full Professor in the Department of Economics of the European University Institute in Florence. In 1996, he returned to the Bank of Spain as Chief Economist and Head of the Time Series Analysis Unit.

Maravall has been on the editorial board of multiple professional journals, including the Journal of Business and Economic Statistics and the Journal of Econometrics. He served on the Program and Scientific Committees of many international meetings and conferences. His teaching experience includes teaching courses in more than 30 countries to participants from more than 60 countries. He was Special Advisor to the European Central Bank and Eurostat, a member of the Board of Directors of the former Institute for Advanced Studies in the Social Sciences of Madrid, and a member of the High Advisory Council for R & D of the Generalitat Valenciana.

Maravall's research focused on time series analysis, time series modeling, and time series applications to economic data. His key contribution, in collaboration with Victor Gómez was the creation of a model-based approach for jointly solving many statistical time series difficulties that affect the analysis and interpretation of economic time series. In the presence of potentially missing observations^[3] the typical (default) technique first performs automatic identification and forecasting of regression-ARIMA models (including adjustment for outliers and calendar effects). The model is then split into models for unobserved components (such as seasonal, trend, transient, and cyclical components), from which filters are generated to estimate and forecast the unobserved components.^[4][5][6] The model-based structure generates the estimators' joint distribution, from which parametric tests and inferences (such as the standard errors of all estimators and forecasts) can be obtained.^[7][8]

Two relevant features of the procedure are first, that all estimators for a given series are derived from the same (well-defined) model and will thus be internally consistent. Second, the automated procedure can be efficiently and reliably applied to very large sets of time series.^[9][10]

His research has been published in academic journals (with some reprinted in books), the Springer-Verlag monograph series Lecture Notes in Statistics and Lecture Notes in Economics and Mathematical Systems, the Bank of Spain monograph series, and in multiple book chapters.^[11]