New forecasting papers 2008-05-19

In this issue we have: Volatility Forecasting Using Explanatory Variables and Focused Selection Criteria ; Forecasting Business Cycles in a Small Open Economy: A Dynamic Factor Model for Singapore ; Estimating Fundamental Cross-Section Dispersion from Fixed Event Forecasts ; Forecast Evaluation of Explanatory Models of Financial Return Variability ; and more.

  • Volatility Forecasting Using Explanatory Variables and Focused Selection Criteria
    Date: 2007-05
    By: Christian T. Brownlees (Università degli Studi di Firenze, Dipartimento di Statistica)
    Giampiero Gallo (Università degli Studi di Firenze, Dipartimento di Statistica "G. Parenti")
    This paper assesses the performance of volatility forecasting using focused selection and combination strategies to include relevant explanatory variables in the forecasting model. The focused selection/combination strategies consist of picking up the model that minimizes the estimated risk (e.g. MSE) of a given smooth function of the parameters of interest to the forecaster. The proposed focused methods are compared with other strategies, including the well established AIC and BIC. The methodology is applied to a daily recursive 1–step ahead value–at–risk (VaR) forecasting exercise of 4 widely traded New York Stock Exchange stocks. Results show that VaR forecasts can significantly be improved upon using focused forecast strategies for the selection of relevant exogenous information. The set of explanatory variables that helps improving prediction is stock dependent. Traditional information criteria do not appe! ar to be helpful in suggesting the inclusion of explanatory variables that actually improve prediction significantly. In line with recent theoretical findings, the predictive performance of the BIC appears to be modest.
    Keywords: Forecasting, Shrinkage Estimation, FIC, MEM, GARCH, ACD
    JEL: C22 C51 C53
  • Forecasting Business Cycles in a Small Open Economy: A Dynamic Factor Model for Singapore
    Date: 2008-02
    By: Hwee Kwan Chow (School of Economics and Social Sciences, Singapore Management University, Singapore)
    Keen Meng Choy (Department of Economics, Nanyang Technological University, Singapore)
    We apply multivariate statistical methods to a large dataset of Singapore's macroeconomic variables and global economic indicators with the objective of forecasting business cycles in a small open economy. The empirical results suggest that three common factors are present in the time series at the quarterly frequency, which can be interpreted as world, regional and domestic economic cycles. This leads us to estimate a factor-augmented vector autoregressive (FAVAR) model for the purpose of optimally forecasting real economic activity in Singapore. By taking explicit account of the common factor dynamics, we find that iterative forecasts generated by this model are significantly more accurate than direct multi-step predictions based on the identified factors as well as forecasts from univariate and vector autoregressions.
    Keywords: business cycles; principal components; dynamic factor model; factor-augmented VAR; forecasting; Singapore
  • Estimating Fundamental Cross-Section Dispersion from Fixed Event Forecasts
    Date: 2008
    By: Jonas Dovern
    Ulrich Fritsche
    A couple of recent papers have shifted the focus towards disagreement of professional forecasters. When dealing with survey data that is sampled at a frequency higher than annual and that includes only fixed event forecasts, e.g. expectation of average annual growth rates measures of disagreement across forecasters naturally are distorted by a component that mainly reflects the time varying forecast horizon. We use data from the Survey of Professional Forecasters, which reports both fixed event and fixed horizon forecasts, to evaluate different methods for extracting the "fundamental" component of disagreement. Based on the paper's results we suggest two methods to estimate dispersion measures from panels of fixed event forecasts: a moving average transformation of the underlying forecasts and estimation with constant forecast-horizon- effects. Both models are easy to handle and deliver equally well performing res! ults, which show a surprisingly high correlation (up to 0:94) with the true dispersion.
    Keywords: Survey data, dispersion, disagreement, fixed event forecasts
    JEL: C22 C32 E37
  • Forecast Evaluation of Explanatory Models of Financial Return Variability
    Date: 2008
    By: Sucarrat, Genaro
    A practice that has become widespread is that of comparing forecasts of financial return variability obtained from discrete time models against high frequency estimates based on continuous time theory. In explanatory financial return variability modelling this raises several methodological and practical issues, which suggests an alternative framework is needed. The contribution of this study is twofold. First, the finite sample properties of operational and practical procedures for the forecast evaluation of explanatory discrete time models of financial return variability are studied. Second, with basis in the simulation results a simple framework is proposed and illustrated.
    Keywords: Return variability forecasting, financial volatility, explanatory modelling
    JEL: C52 C53 F31 F37
  • Non-linear predictability in stock and bond returns: when and where is it exploitable?
    Date: 2008
    By: Massimo Guidolin
    Stuart Hyde
    David McMillan
    Sadayuki Ono
    We systematically examine the comparative predictive performance of a number of alternative linear and non-linear models for stock and bond returns in the G7 countries. Besides Markov switching, threshold autoregressive (TAR), and smooth transition autoregressive (STAR) regime switching (predictive) regression models, we also estimate univariate models in which conditional heteroskedasticity is captured through GARCH, TARCH and EGARCH models and ARCH-in mean effects appear in the conditional mean. Although we fail to find a consistent winner/out-performer across all countries and asset markets, it turns out that capturing non-linear effects is of extreme importance to improve forecasting performance. U.S. and U.K. asset return data are "special" in the sense that good predictive performance seems to loudly ask for models that capture non linear dynamics, especially of the Markov switching type. Although occasional! ly also stock and bond return forecasts for other G7 countries appear to benefit from non-linear modeling (especially of TAR and STAR type), data from France, Germany, and Italy express interesting predictive results on the basis of simpler benchmarks. U.S. and U.K. data are also the only two data sets in which we find statistically significant differences between forecasting models. Results appear to be remarkably stable over time, and robust to the specification of the loss function used in statistical evaluations as well as to the methodology employed to perform pairwise comparisons.
    Keywords: Group of Seven countries ; Financial markets
  • Comparison of Volatility Measures: a Risk Management Perspective
    Date: 2008-02
    By: Christian T. Brownlees (Università degli Studi di Firenze, Dipartimento di Statistica)
    Giampiero Gallo (Università degli Studi di Firenze, Dipartimento di Statistica "G. Parenti")
    In this paper we address the issue of forecasting Value-at-Risk (VaR) using different volatility measures: realized volatility, bipower realized volatility, two scales realized volatility, realized kernel as well as the daily range. We propose a dynamic model with a flexible trend specification bonded with a penalized maximum likelihood estimation strategy: the P-Spline Multiplicative Error Model. Exploiting UHFD volatility measures, VaR predictive ability is considerably improved upon relative to a baseline GARCH but not so relative to the range; there are relevant gains from modeling volatility trends and using realized kernels that are robust to dependent microstructure noise.
    Keywords: Volatility Measures, VaR Forecasting, GARCH, MEM, P-Spline.
    JEL: C22 C51 C52 C53
  • A Model for Multivariate Non-negative Valued Processes in Financial Econometrics
    Date: 2007-12
    By: Fabrizio Cipollini (Università degli Studi di Firenze, Dipartimento di Statistica "G. Parenti")
    Robert F. Engle (Department of Finance, Stern School of Business, New York University)
    Giampiero M. Gallo (Università degli Studi di Firenze, Dipartimento di Statistica "G. Parenti")
    The Multiplicative Error Model introduced by Engle (2002) for non-negative valued processes is specified as the product of a (conditionally autoregressive) scale factor and an innovation process with positive support. In this paper we propose a multivariate extension of such a model, by taking into consideration the possibility that the vector innovation process be contemporaneously correlated. The estimation procedure is hindered by the lack of probability density functions for multivariate non-negative valued random variables. We suggest the use of copula functions to jointly estimate the parameters of the scale factors and of the correlations of the innovation processes. We illustrate the feasibility of the procedure and the gains over the equation by equation approach using a four variable fully interdependent model with different volatility measures.
    Keywords: Volatility, Copula functions, Forecasting, GARCH, MEM.
    JEL: C22 C51 C52 C53
  • Non-stationarity and meta-distribution.
    Date: 2008-03
    By: Dominique Guegan (Centre d'Economie de la Sorbonne et Paris School of Economics)
    In this paper we deal with the problem of non-stationarity encountered in a lot of data sets, mainly in financial and economics domains, coming from the presence of multiple seasonnalities, jumps, volatility, distorsion, aggregation, etc. Existence of non-stationarity involves spurious behaviors in estimated statistics as soon as we work with finite samples. We illustrate this fact using Markov switching processes, Stopbreak models and SETAR processes. Thus, working with a theoretical framework based on the existence of an invariant measure for a whole sample is not satisfactory. Empirically alternative strategies have been developed introducing dynamics inside modelling mainly through the parameter with the use of rolling windows. A specific framework has not yet been proposed to study such non-invariant data sets. The question is difficult. Here, we address a discussion on this topic proposing the concept of met! a-distribution which can be used to improve risk management strategies or forecasts.
    Keywords: Non-stationarity, switching processes, SETAR processes, jumps, forecast, risk management, copula, probability distribution function.
    JEL: C32 C51 G12
  • A non-parametric method to nowcast the Euro Area IPI.
    Date: 2008-04
    By: Laurent Ferrara (Banque de France et Centre d'Economie de la Sorbonne)
    Thomas Raffinot (CPR Asset Management)
    Non-parametric methods have been empirically proved to be of great interest in the statistical literature in order to forecast stationary time series, but very few applications have been proposed in the econometrics literature. In this paper, our aim is to test whether non-parametric statistical procedures based on a Kernel method can improve classical linear models in order to nowcast the Euro area manufacturing industrial production index (IPI) by using business surveys released by the European Commission. Moreover, we consider the methodology based on bootstrap replications to estimate the confidence interval of the nowcasts.
    Keywords: Non-parametric, kernel, nowcasting, bootstrap, Euro area IPI.
    JEL: C22 C51 E66
  • Business surveys modelling with seasonal-cyclical long memory models.
    Date: 2008-05
    By: Laurent Ferrara (Banque de France et Centre d'Economie de la Sorbonne)
    Dominique Guegan (Centre d'Economie de la Sorbonne et Paris School of Economics)
    Business surveys are an important element in the analysis of the short-term economic situation because of the timeliness and nature of the information they convey. Especially, surveys are often involved in econometric models in order to provide an early assessment of the current state of the economy, which is of great interest for policy-makers. In this paper, we focus on non-seasonally adjusted business surveys released by the European Commission. We introduce an innovative way for modelling those series taking the persistence of the seasonal roots into account through seasonal-cyclical long memory models. We empirically prove that such models produce more accurate forecasts than classical seasonal linear models.
    Keywords: Euro area, nowcasting, business surveys, seasonal, long memory.
    JEL: C22 C53 E32
  • Taken from the NEP-FOR mailing list edited by Rob Hyndman.