Agustín Maravall: An interview with the International Journal of Forecasting

Agustín Maravall Herrero (Madrid, 1944) is one of the world’s authorities in seasonal adjustment and automatic forecasting of economic time series. He studied Agricultural Engineering and completed a doctorate at the Universidad Politécnica de Madrid. With a Ford-Fulbright fellowship he moved to the University of Wisconsin-Madison to obtain a Ph.D. in Economics in 1975. He worked at the Research Division of the Federal Reserve Board of Governors (the Fed) in Washington D.C. and in 1979 returned to Madrid as a Senior Economist in the Research Department of the Banco de España (BE). In the period 1989-96, he was a full professor in the Department of Economics of the European University Institute (EUI) in Florence. He returned to the BE as Chief Economist and Head of the Time Series Analysis Unit and retired in December 2014.Maravall has done outstanding research in time series and has been a pioneer in developing methodology and writing computer programs for automatic estimation and model selection, seasonal adjustment, and forecasting of time series. His programs “Time Series Regression with ARIMA noise, Missing observations and Outliers” (TRAMO) and “Signal Extraction in ARIMA Time Series” (SEATS), jointly developed with Victor Gómez, have had a large influence in applied forecasting, including adjusting series for seasonality and possibly other undesirable effects, such as outliers, or missing observations, and have been used in many economic institutions around the world. He has been very active in promoting the automatic analysis of time series, teaching short courses in many countries. Also, he has stimulated research in this field being on the editorial board of the Journal of Business and Economic Statistics and the Journal of Econometrics. He has been a Special Advisor to the European Central Bank (ECB) and Eurostat in time series analysis. His research contributions have been recognized as Fellow of the Journal of Econometrics, 1995; Fellow of the American Statistical Association, 2000; Julius Shiskin Award for Economic Statistics, 2004, and the highest prizes for Economic Research in Spain: The Rey Jaime I Prize in Economics, 2005 and the Rey Juan Carlos Prize in Economics, 2014.     

A nonlinear long memory model, with an application to US unemployment

Two important empirical features of US unemployment are that shocks to the series seem rather persistent and that it seems to rise faster during recessions than that it falls during expansions. To jointly capture these features of long memory and nonlinearity, we put forward a new time series model and evaluate its empirical performance. We find that the model describes the data rather well and that it outperforms related competitive models on various measures of fit.     

Non-parametric, unconditional quantile estimation for efficiency analysis with an application to Federal Reserve check processing operations

This paper examines the technical efficiency of US Federal Reserve check processing offices over 1980–2003. We extend results from Park et al. [Park, B., Simar, L., Weiner, C., 2000. FDH efficiency scores from a stochastic point of view. Econometric Theory 16, 855–877] and Daouia and Simar [Daouia, A., Simar, L., 2007. Nonparametric efficiency analysis: a multivariate conditional quantile approach. Journal of Econometrics 140, 375–400] to develop an unconditional, hyperbolic, α-quantile estimator of efficiency. Our new estimator is fully non-parametric and robust with respect to outliers; when used to estimate distance to quantiles lying close to the full frontier, it is strongly consistent and converges at rate root-n, thus avoiding the curse of dimensionality that plagues data envelopment analysis (DEA) estimators. Our methods could be used by policymakers to compare inefficiency levels across offices or by managers of individual offices to identify peer offices.     

Combination of long term and short term forecasts, with application to tourism demand forecasting

Forecast combination is a well-established and well-tested approach for improving the forecasting accuracy. One beneficial strategy is to use constituent forecasts that have diverse information. In this paper we consider the idea of diversity being accomplished by using different time aggregations. For example, we could create a yearly time series from a monthly time series and produce forecasts for both, then combine the forecasts. These forecasts would each be tracking the dynamics of different time scales, and would therefore add diverse types of information. A comparison of several forecast combination methods, performed in the context of this setup, shows that this is indeed a beneficial strategy and generally provides a forecasting performance that is better than the performances of the individual forecasts that are combined.As a case study, we consider the problem of forecasting monthly tourism numbers for inbound tourism to Egypt. Specifically, we consider 33 individual source countries, as well as the aggregate. The novel combination strategy also produces a generally improved forecasting accuracy.     

Forecasting with vector autoregressive (VAR) models subject to business cycle restrictions

In the last decade VAR models have become a widely-used tool for forecasting macroeconomic time series. To improve the out-of-sample forecasting accuracy of these models, Bayesian random-walk prior restrictions are often imposed on VAR model parameters. This paper focuses on whether placing an alternative type of restriction on the parameters of unrestricted VAR models improves the out-of-sample forecasting performance of these models. The type of restriction analyzed here is based on the business cycle characteristics of U.S. macroeconomic data, and in particular, requires that the dynamic behavior of the restricted VAR model mimic the business cycle characteristics of historical data. The question posed in this paper is: would a VAR model, estimated subject to the restriction that the cyclical characteristics of simulated data from the model “match up” with the business cycle characteristics of U.S. data, generate more accurate out-of-sample forecasts than unrestricted or Bayesian VAR models?     

A new methodology to measure efficiencies of inputs (outputs) of decision making units in Data Envelopment Analysis with application to agriculture

This paper aims at developing a new methodology to measure and decompose global DMU efficiency into efficiency of inputs (or outputs). The basic idea rests on the fact that global DMU's efficiency score might be misleading when managers proceed to reallocate their inputs or redefine their outputs. Literature provides a basic measure for global DMU's efficiency score. A revised model was developed for measuring efficiencies of global DMUs and their inputs (or outputs) efficiency components, based on a hypothesis of virtual DMUs. The present paper suggests a method for measuring global DMU efficiency simultaneously with its efficiencies of inputs components, that we call Input decomposition DEA model (ID-DEA), and its efficiencies of outputs components, that we call output decomposition DEA model (OD-DEA). These twin models differ from Supper efficiency model (SE-DEA) and Common Set Weights model (CSW-DEA). The twin models (ID-DEA, OD-DEA) were applied to agricultural farms, and the results gave different efficiency scores of inputs (or outputs), and at the same time, global DMU's efficiency score was given by the Charnes, Cooper and Rhodes (Charnes et al., 1978) [1], CCR78 model. The rationale of our new hypothesis and model is the fact that managers don't have the same information level about all inputs and outputs that constraint them to manage resources by the (global) efficiency scores. Then each input/output has a different reality depending on the manager's decision in relationship to information available at the time of decision. This paper decomposes global DMU's efficiency into input (or output) components' efficiencies. Each component will have its score instead of a global DMU score. These findings would improve management decision making about reallocating inputs and redefining outputs. Concerning policy implications of the DEA twin models, they help policy makers to assess, ameliorate and reorient their strategies and execute programs towards enhancing the best practices and minimising losses.     

Estimation of variance of housing prices using spatial conditional heteroskedasticity (SARCH) model with an application to Boston housing price data

In this paper we investigate housing price volatility within a spatial econometrics setting. We propose an extended spatial regression model of the real estate market that includes the effects of both conditional heteroskedasticity and spatial autocorrelation. Our suggested model has features similar to those of autoregressive conditional heteroskedasticity (ARCH) in the time-series context. We utilize the spatial ARCH (SARCH) model to analyze Boston housing price data used by Harrison and Rubinfeld (1978) and Gilley and Pace (1996). We show that measuring the variability of housing prices is an important issue and our SARCH model captures the conditional spatial variability of Boston housing prices. We argue that there is a different source of spatial variation, which is independent of traditional housing and neighborhood characteristics, and is captured by the SARCH model.