versão revista publicada em Victor Bulmer-Thomas, John H. Coatsworth e Roberto Cortés-Conde, The Cambridge Economic History of Latin America, Volume II. The Long 20th Century, Cambridge University Press, 2006, p. 101-134

This paper is concerned with modelling time series by single hidden layer feedforward neural network models. A coherent modelling strategy based on statistical inference is presented. Variable selection is carried out using simple existing techniques. The problem of selecting the number of hidden units is solved by sequentially applying Lagrange multiplier type tests, with the aim of avoiding the estimation of unidentified models. Misspecification tests are derived for evaluating an estimated neural network model. All the tests are entirely based on auxiliary regressions and are easily implemented. A small-sample simulation experiment is carried out to show how the proposed modelling strategy works and how the misspecification tests behave in small samples. Two applications to real time series, one univariate and the other multivariate, are considered as well. Sets of one-step-ahead forecasts are constructed and forecast accuracy is compared with that of other nonlinear models applied to the same series

Publicado em Journal of Forecasting, Volume 25, Número 1, pp. 49 - 75, 2006

 publicado no Journal of Development Economics v.69, n.2, p. 393-421, 2002

publicado na Revista Brasileira de Economia, v. 57, n. 2, 2003

Publicado como um capítulo do livro de R. van der Hoven e A. Shorrocks (eds.): Growth, inequality and poverty, New York, Oxford University Press.

Publicado na revista Pesquisa e Planejamento Econômico, v. 32, n. 3, dez. 2002.

Prefácio à edição brasileira de J. M. Keynes: "As conseqüências econômicas da paz", São Paulo, Imprensa Oficial do Estado, Editora Universidade de Brasília, Instituto de Pesquisa de Relações Internacionais, 2002.

The important issue of forecasting volatilities brings the difficult task of back-testing the forecasting performance. As volatility cannot be observed directly, one has to use an observable proxy for volatility or a utility function to assess the prediction quality. This kind of procedure can easily lead to poor assessment. The goal of this paper is to compare different volatility models and different performance measures using White’s Reality Check. The Reality Check consists of a non-parametric test that checks if any of a number of concurrent methods yields forecasts significantly better than a given benchmark method. For this purpose, a Monte Carlo simulation is carried out with four different processes, one of them a Gaussian white noise and the others following GARCH specifications. Two benchmark methods are used: the naive (predicting the out-of-sample volatility by in-sample variance) and the Riskmetrics method

Publicado na Brazilian Review of Econometrics, v. 25, n. 1, mai. 2005.