Forecasting papers 2009-03-19

In this issue we have Forecasting Exchange Rate Volatility, A breakdown of GDP forecast errors for Austria, Experts´ Macroeconomics Expectations: An Evaluation of Mexican Short-Run Forecasts, Forecasting the fragility of the banking and insurance sector and many more.

  • Forecasting Exchange Rate Volatility: The Superior Performance of Conditional Combinations of Time Series and Option Implied Forecasts
    Date: 2009-01
    By: Guillermo Benavides
    Carlos Capistrán
    This paper provides empirical evidence that combinations of option implied and time series volatility forecasts that are conditional on current information are statistically superior to individual models, unconditional combinations, and hybrid forecasts. Superior forecasting performance is achieved by both, taking into account the conditional expected performance of each model given current information, and combining individual forecasts. The method used in this paper to produce conditional combinations extends the application of conditional predictive ability tests to select forecast combinations. The application is for volatility forecasts of the Mexican Peso-US Dollar exchange rate, where realized volatility calculated using intra-day data is used as a proxy for the (latent) daily volatility.
    Keywords: Composite Forecasts, Forecast Evaluation, GARCH, Implied volatility, Mexican Peso-U.S. Dollar Exchange Rate, Regime-Switching
    JEL: C22
  • Why did we fail to predict GDP during the last cycle? A breakdown of forecast errors for Austria.
    Date: 2009-02-11
    By: Martin Schneider (Oesterreichische Nationalbank, Economic Analysis Division, P.O. Box 61, A-1010 Vienna,)
    Christian Ragacs (Oesterreichische Nationalbank, Economic Analysis Division, P.O. Box 61, A-1010 Vienna,)
    This paper proposes an informal taxonomy to break down forecast errors of institutional forecasts. This breakdown is demonstrated for the forecasts of the Oesterreichische Nationalbank (OeNB) for Austrian GDP. The main result is that the largest part of the forecast errors can be explained by erroneous projections of the international environment. Data revisions also substantially contribute to the forecasting error for the forecast of the current year. Domestic exogenous variables play a minor role only. The inclusion of judgement improves the forecasting performance.
    Keywords: Forecast error taxonomy; Breakdown; Austria; Judgement; Technical forecast.
  • Experts´ Macroeconomics Expectations: An Evaluation of Mexican Short-Run Forecasts.
    Date: 2008-08
    By: Carlos Capistrán
    Gabriel López-Moctezuma
    This document analyzes inflation, exchange rate, interest rate, and GDP growth forecasts from the monthly Survey of Specialists in Economics from the Private Sector, maintained by Banco de México. The study concentrates on the mean across forecasters for the period from January 1995 to April 2008. The study evaluates the efficiency in the use of information and the relative performance using as benchmarks forecasts from time series models and from other macroeconomic variables. Inflation, interest rate, and GDP expectations seem to incorporate information in a relatively efficient manner. These forecasts appear to be better, in mean squared error terms, than the benchmark forecasts, except for the case of one-yearahead inflation. In addition, exchange rate forecasts do not seem to optimally incorporate available information and do not seem to improve upon forecasts obtained from a random walk model.
    Keywords: Predictive ability, Rational expectations, Rolling-forecasts.
    JEL: C22
  • Forecasting the fragility of the banking and insurance sector
    Date: 2009-02
    By: Kerstin Bernoth
    Andreas Pick
    This paper considers the issue of forecasting financial fragility of banks and insurances using a panel data set of performance indicators, namely distance-to-default, taking unobserved common factors into account. We show that common factors are important in the performance of banks and insurances, analyze the influences of a number of observable factors on banking and insurance performance, and evaluate the forecasts from our model. We find that taking unobserved common factors into account reduces the root mean square forecasts error of  firm specific forecasts by up to 11% and of system forecasts by up to 29% relative to a model based only on observed variables. Estimates of the factor loadings suggest that the correlation of financial institutions has been relatively stable over the forecast period.
    Keywords: Financial stability; financial linkages; banking; insurances; unobserved common factors; forecasting
    JEL: C53
  • Forecasting Large Datasets with Conditionally Heteroskedastic Dynamic Common Factors
    Date: 2009
    By: Lucia Alessi
    Matteo Barigozzi
    Marco Capasso
    We propose a new method for multivariate forecasting which combines Dynamic Factor and multivariate GARCH models. We call the model Dynamic Factor GARCH, as the information contained in large macroeconomic or financial datasets is captured by a few dynamic common factors, which we assume being conditionally heteroskedastic. After describing the estimation of the model, we present simulation results and carry out two empirical applications on financial asset returns and macroeconomic series, with a particular focus on different measures of inflation. Our proposed model outperforms the benchmarks in forecasting the conditional volatility of returns and the inflation level. Moreover, it allows to predict conditional covariances of all the time series in the panel.
    Keywords: Dynamic factors, multivariate GARCH, covolatility forecasting, inflation forecasting
    JEL: C52
  • Theta Model Forecasts for Financial Time Series: A Case Study in the S&P500
    Date: 2009
    By: Konstantinos Nikolopoulos
    Dimitrios Thomakos
    Fotios Petropoulos
    Vassilis Assimakopoulos
    The Theta model created a lot of interest in academic circles due to its surprisingly good performance in the M3 forecasting competition. However, this interest was not followed up by other studies, with the exception of Hyndman and Billah in 2003. In addition, the Theta model performance has not been tested on a large dataset of non-demand forecasting series, nor its properties have been examined analytically for time series that are found in finance and economics. The present study presents some empirical results on the application of the Theta model for forecasting the evolution of the S&P500 index, both as an examination of its relative performance against the standard benchmarks, and as a motivation for further theoretical work. We use weekly data over a long period of 20 years and the Theta model is used alongside the benchmark models and rolling-origin forecasts are generated. The results are interesting since they show that the Theta model has performance that is either on par or better than the benchmarks.
    Keywords: financial time series, forecasting, S&P500, Theta model.
  • The Theta Model in the Presence of a Unit Root Some new results on "optimal" theta forecasts
    Date: 2009
    By: Dimitrios Thomakos
    Konstantinos Nikolopoulos
    We significantly extend earlier work by Assimakopoulos and Nikopoloulos (2000) and Hyndman and Billah (2003) on the properties and performance of the Theta model, and potentially explain its very good performance in the M3 forecasting competition. We derive a number of new theoretical results for theta forecasts when the data generating process contains both deterministic and stochastic trends. In particular (a) we show that using the standard theta forecasts coincides with the minimum mean-squared error forecast when the innovations are uncorrelated; (b) we provide, for the first time, an optimal value for the theta parameter, which coincides with the first order autocorrelation of the innovations, and thus provide a single optimal theta line; (c) we show that the optimal linear combination of two standard theta lines coincides with the single optimal theta line of (b). Under (b) and (c) we show that the optimal theta fo recast function is identical with that of an ARIMA(1,1,0) model. Furthermore, we illustrate how the Theta model can be generalized to include local behavior in two different ways.
    Keywords: forecasting, theta model, unit roots.
  • "Modeling and Forecasting the Volatility of the Nikkei 225 Realized Volatility Using the ARFIMA-GARCH Model"
    Date: 2009-01
    By: Isao Ishida (Faculty of Economics and Graduate School of Public Policy, University of Tokyo)
    Toshiaki Watanabe (Institute of Economic Research, Hitotsubashi University)
    In this paper, we apply the ARFIMA-GARCH model to the realized volatility and the continuous sample path variations constructed from high-frequency Nikkei 225 data. While the homoskedastic ARFIMA model performs excellently in predicting the Nikkei 225 realized volatility time series and their square-root and log transformations, the residuals of the model suggest presence of strong conditional heteroskedasticity similar to the finding of Corsi et al. (2007) for the realized S&P 500 futures volatility. An ARFIMA model augmented by a GARCH(1,1) specification for the error term largely captures this and substantially improves the fit to the data. In a multi-day forecasting setting, we also find some evidence of predictable time variation in the volatility of the Nikkei 225 volatility captured by the ARFIMA-GARCH model.
  • Forecasting Errors: Yet More Problems for Identification?
    Date: 2009-02
    By: Contini, Bruno (LABORatorio R. Revelli)
    Forecasting errors pose a serious problem of identification, often neglected in empirical applications. Any attempt of estimating choice models under uncertainty may lead to severely biased results in the presence of forecasting errors even when individual expectations on future events are observed together with the standard outcome variables.
    Keywords: identification, forecasting errors, subjective probabilities
    JEL: C01
  • Neural Networks for Cross-Sectional Employment Forecasts: A Comparison of Model Specifications for Germany
    Date: 2009-02
    By: Roberto Patuelli (Institute for Economic Research (IRE), University of Lugano, Switzerland; The Rimini Centre for Economic Analysis, Italy)
    Aura Reggiani (Department of Economics, University of Bologna, Italy)
    Peter Nijkamp (Department of Spatial Economics, VU University Amsterdam, The Netherlands)
    Norbert Schanne (Institute for Employment Research (IAB), Nuremberg, Germany)
    In this paper, we present a review of various computational experiments – and consequent results – concerning Neural Network (NN) models developed for regional employment forecasting. NNs are widely used in several fields because of their flexible specification structure. Their utilization in studying/predicting economic variables, such as employment or migration, is justified by the ability of NNs of learning from data, in other words, of finding functional relationships – by means of data – among the economic variables under analysis. A series of NN experiments is presented in the paper. Using two data sets on German NUTS 3 districts (326 and 113 labour market districts in the former West and East Germany, respectively), the results emerging from the implementation of various NN models – in order to forecast variations in full-time employment – are provided and discussed In our approach, single forecasts are computed by the models for each district. Different specifications of the NN models are first tested in terms of: (a) explanatory variables; and (b) NN structures. The average statistical results of simulated out-of-sample forecasts on different periods are summarized and commented on. In addition to variable and structure specification, the choice of NN learning parameters and internal functions is also critical to the success of NNs. Comprehensive testing of these parameters is, however, limited in the literature. A sensitivity analysis is therefore carried out and discussed, in order to evaluate different combinations of NN parameters. The paper concludes with methodological and empirical remarks, as well as with suggestions for future research.
    Keywords: neural networks, sensitivity analysis, employment forecasts, Germany
    JEL: C45
  • The Time-Series Properties on Housing Prices: A Case Study of the Southern California Market
    Date: 2009-02
    By: Rangan Gupta (Department of Economic, University of Pretoria)
    Stephen M. Miller (College of Business, University of Las Vegas, Nevada)
    We examine the time-series relationship between housing prices in eight Southern California metropolitan statistical areas (MSAs). First, we perform cointegration tests of the housing price indexes for the MSAs, finding seven cointegrating vectors. Thus, the evidence suggests that one common trend links the housing prices in these eight MSAs, a purchasing power parity finding for the housing prices in Southern California. Second, we perform temporal Granger causality tests revealing intertwined temporal relationships. The Santa Anna MSA leads the pack in temporally causing housing prices in six of the other seven MSAs, excluding only the San Luis Obispo MSA. The Oxnard MSA experienced the largest number of temporal effects from other MSAs, six of the seven, excluding only Los Angeles. The Santa Barbara MSA proved the most isolated in that it temporally caused housing prices in only two other MSAs (Los Angels and Oxnard) a nd housing prices in the Santa Anna MSA temporally caused prices in Santa Barbara. Third, we calculate out-of-sample forecasts in each MSA, using various vector autoregressive (VAR) and vector error-correction (VEC) models, as well as Bayesian, spatial, and causality versions of these models with various priors. Different specifications provide superior forecasts in the different MSAs. Finally, we consider the ability of theses time-series models to provide accurate out-of-sample predictions of turning points in housing prices that occurred in 2006:Q4. Recursive forecasts, where the sample is updated each quarter, provide reasonably good forecasts of turning points.
    Keywords: Housing prices, Forecasting
    JEL: C32
  • Forecasting Price Relationships among U.S Tree Nuts Prices
    Date: 2009-01-15
    By: Ibrahim, Mohammed
    Florkowski, Wojciech J.
    This paper investigates a vector auto regression model, using the Johansen cointegration technique, and the autoregressive integrated moving average time series models to determine the better model for forecasting US tree nut prices over the period 1992-2006. The Johansen contegration test shows lack of long run relationship among pecan, walnut, and almond prices. As such, only autoregressive integrated moving average-type models were used in forecasting U.S. nut prices.
    Keywords: substitutability, cointegration, tree nuts, long-run equilibrium forecasting, Demand and Price Analysis, Production Economics,
  • Predicting European Union recessions in the euro era: The yield curve as a forecasting tool of economic activity
    Date: 2009-03
    By: Gogas, Periklis
    Chionis, Dionisios
    Pragkidis, Ioannis
    Several studies have established the predictive power of the yield curve, ie: the difference between long and short term bond rates, in terms of real economic activity, for the U.S. and various European countries. In this paper we use data from the European Union (EU15), ranging from 1994:Q1 to 2008:Q3. The seasonally adjusted real GDP is used to extract the long run trend and the cyclical component of the European output, while the European Central Bank's euro area government benchmark bonds of various maturities are used for the calculation of the yield spreads. We also augment the models tested with non monetary policy variables: the unemployment and a composite European stock price index constructed from the indices of the three major European stock markets of London, Frankfurt and Paris. The methodology employed in the effort to forecast recessions, is a probit model of the inverse cumulative distribution function of the standard distribution, using several formal forecasting evaluation tests. The results show that the yield curve augmented with the composite stock index has significant forecasting power in terms of the EU15 real output.
    Keywords: forecasting; yield spread; recession; probit; term structure; monetary policy; real growth.
    JEL: E43
  • Modeling the Dynamics of EU Economic Sentiment Indicators: An Interaction-Based Approach
    Date: 2009-02
    By: Jaba Ghonghadze
    Thomas Lux
    This paper estimates a simple univariate model of expectation or opinion formation in continuous time adapting a ‘canonical' stochastic model of collective opinion dynamics (Weidlich and Haag, 1983; Lux, 1995, 2007). This framework is applied to a selected data set on survey-based expectations from the rich EU business and consumer survey database for twelve European countries. The model parameters are estimated through maximum likelihood and numerical solution of the transient probability density functions for the resulting stochastic process. The model's performance is assessed with respect to its out-of-sample forecasting capacity relative to univariate time series models of the ARMA(p; q) and ARFIMA(p; d; q) varieties. These tests speak for a slight superiority of the canonical opinion dynamics model over the alternatives in the majority of cases
    Keywords: expectation formation, survey-based expectations, opinion dynamics, Fokker-Planck equation, forecasting
    JEL: E32 C53
  • Taken from the NEP-FOR mailing list edited by Rob Hyndman.