Forecasting papers 2008-11-27

In this issue we have: Accuracy, Certainty and Surprise – A Prediction Market on the Outcome of the 2002 FIFA World Cup ; Practical Issues in the Analysis of Univariate GARCH Models ; Long Memory versus Structural Breaks in Modeling and Forecasting Realized Volatility ; and more.


  • Accuracy, Certainty and Surprise – A Prediction Market on the Outcome of the 2002 FIFA World Cup
    Date: 2008-07-24
    By: Schmidt, Carsten (Sonderforschungsbereich 504, University of Mannheim)
    Strobel, Martin (Maastricht University)
    Volkland, Henning Oskar (Goldman Sachs & Co., Frankfurt)
    In this chapter, we present our empirical investigation of the forecasting accuracy of a prediction market experiment drawn on the outcome of the World Cup 2002. We analyse the predictive accuracy of 64 markets and compare to bookmakers' quotes and chance as benchmarks. We revisit the evaluation of Schmidt and Werwatz (Chapter 16) and compare our results directly to their findings. In addition, we propose a new method for testing predictive accuracy by means of a non-parametric test for the similarity of probability distributions and we evaluate the incorporation of information in market prices by comparing pre-match and half-time price data. We find a reversed favourite-longshot bias when analysing market prices before the start of the match and this bias does not disappear with the inflow of new information until half-time. Unlike the market based predictions bookmakers appear to be perfectly calibrated. Since! there were substantial deviations in outcome between the 2000 European Championship and our data, we offer possible explanations for the much worse performance of the 2002 World Cup prediction market. Consistent with Schmidt and Werwatz (Chapter 16) prediction markets do assign relatively higher probabilities to the favourite when compared to the odds-setters. Together with a long streak of surprising outcomes this fact appears most likely to be responsible for the predictive inaccuracy.

  • Practical Issues in the Analysis of Univariate GARCH Models
    Date: 2008-04
    By: Eric Zivot (Department of Economics, University of Washington)
    This paper gives a tour through the empirical analysis of univariate GARCH models for financial time series with stops along the way to discuss various practical issues associated with model specification, estimation, diagnostic evaluation and forecasting.

  • Long Memory versus Structural Breaks in Modeling and Forecasting Realized Volatility
    Date: 2008-09
    By: Kyongwook Choi (Department of Economics, The University of Seoul,)
    Wei-Choun Yu (Economics and Finance Department, Winona State University)
    Eric Zivot (Department of Economics, University of Washington)
    We explore the possibility of structural breaks in realized volatility with observed long-memory properties for the daily Deutschemark/Dollar, Yen/Dollar and Yen/Deutschemark spot exchange rate realized volatility. We find that structural breaks can partly explain the persistence of realized volatility. We propose a VAR-RV-Break model that provides superior predictive ability compared to most of the forecasting models when the future break is known. With unknown break dates and sizes, we find that the VAR-RV-I(d) long memory model, however, is a very robust forecasting method even when the true financial volatility series are generated by structural breaks.

  • The Calibration of Probabilistic Economic Forecasts
    Date: 2008-11-01
    By: John Galbraith
    Simon van Norden
    A probabilistic forecast is the estimated probability with which a future event will satisfy a specified criterion. One interesting feature of such forecasts is their calibration, or the match between predicted probabilities and actual outcome probabilities. Calibration has been evaluated in the past by grouping probability forecasts into discrete categories. Here we show that we can do so without discrete groupings; the kernel estimators that we use produce efficiency gains and smooth estimated curves relating predicted and actual probabilities. We use such estimates to evaluate the empirical evidence on calibration error in a number of economic applications including recession and inflation prediction, using both forecasts made and stored in real time and pseudoforecasts made using the data vintage available at the forecast date. We evaluate outcomes using both first-release outcome measures as well as later, th! oroughly-revised data. We find strong evidence of incorrect calibration in professional forecasts of recessions and inflation. We also present evidence of asymmetries in the performance of inflation forecasts based on real-time output gaps. <P>Une prévision probabiliste représente la probabilité qu'un événement futur satisfasse une condition donnée. Un des aspects intéressants de ces prévisions est leur calibration, c'est-à-dire l'appariement entre les probabilités prédites et les probabilités réalisées. Dans le passé, la calibration a été évaluée en regroupant des probabilités de prévisions en catégories distinctes. Nous proposons d'utiliser des estimateurs à noyaux, qui sont plus efficaces et qui estiment une relation lisse entre les probabilités prédites et réalisées. Nous nous servons de ces estimations pour évaluer l'importance empirique des erreurs de calibration dans plusieurs pratiques économiques, telles que la prévisio! n de récessions et de l'inflation. Pour ce faire, nous utilisons de s prévisions historiques, ainsi que des pseudoprévisions effectuées à l'aide de données telles qu'elles étaient au moment de la prévision. Nous analysons les résultats en utilisant autant des estimations préliminaires que des estimations tardives, ces dernières incorporant parfois des révisions importantes. Nous trouvons une forte évidence empirique d'une calibration erronée des prévisions professionnelles de récession et d'inflation. Nous présentons aussi une évidence d'asymétries dans la performance des prévisions d'inflation basées sur des estimations des écarts de la production en temps réel.
    Keywords: calibration, probability forecast, real-time data, inflation, recession, calibration, probabilités de prévisions, données « en temps réel », inflation, récession

  • Comparing the New Keynesian Phillips Curve with Time Series Models to Forecast Inflation
    Date: 2008-09-30
    By: Fabio Rumler (Oesterreichische Nationalbank, Economic Analysis Division, P.O. Box 61, A-1010 Vienna,)
    Maria Teresa Valderrama (Oesterreichische Nationalbank, Economic Analysis Division, P.O. Box 61, A-1010 Vienna,)
    The New Keynesian Phillips Curve, as a structural model of inflation dynamics, has mostly been used to explain past inflation developments, but has hardly been used for forecasting purposes. We propose a method of forecasting inflation based on the present-value formulation of the hybrid New Keynesian Phillips Curve. To evaluate the forecasting performance of this model we compare it with forecasts generated from time series models at different forecast horizons. As state-of-the-art time series models used in inflation forecasting we employ a Bayesian VAR, a traditional VAR and a simple autoregressive model. We find that the New Keynesian Phillips Curve delivers relatively more accurate forecasts compared to the other models for longer forecast horizons (more than 3 months) while they are outperformed by the time series models only for the very short forecast horizon. This is consistent with the finding in the lit! erature that structural models are able to outperform time series models only for longer horizons.
    Keywords: New Keynesian Phillips Curve, Inflation Forecasting, Forecast Evaluation, Bayesian VAR
    JEL: E31 C32 C53

  • Predictive Densities for Shire Level Wheat Yield in Western Australia
    Date: 2008
    By: William E Griffiths
    Lisa S Newton
    Christopher J O'Donnell
    Wheat yield in Western Australia (WA) depends critically on rainfall during three periods – germination, growing and flowering. The degree of uncertainty attached to a wheat-yield prediction depends on whether the prediction is made before or after the rainfall in each period has been realised. Bayesian predictive densities that reflect the different levels of uncertainty in wheat-yield predictions made at four different points in time are derived for five shires in Western Australia. The framework used for prediction is a linear regression model with stochastic regressors and inequality restrictions on the coefficients. An algorithm is developed that can be used more generally for obtaining Bayesian predictive densities in linear and nonlinear models with inequality constraints, and with or without stochastic regressors.
    Keywords: Bayesian forecasting; inequality restrictions; random regressors.

  • Can Exchange Rates Forecast Commodity Prices?
    Date: 2008-02
    By: Yu-chin Chen (University of Washington)
    Kenneth Rogoff (Harvard University)
    Barbara Rossi (Duke University)
    This paper demonstrates that "commodity currency" exchange rates have remarkably robust power in predicting future global commodity prices, both in-sample and out-of-sample. A critical element of our in-sample approach is to allow for structural breaks, endemic to empirical exchange rate models, by implementing the approach of Rossi (2005b). Aside from its practical implications, our forecasting results provide perhaps the most convincing evidence to date that the exchange rate depends on the present value of identifiable exogenous fundamentals. We also find that the reverse relationship holds; that is, that commodity prices Granger-cause exchange rates. However, consistent with the vast post-Meese-Rogoff (1983a,b) literature on forecasting exchange rates, we find that the reverse forecasting regression does not survive out-of-sample testing. We argue, however, that it is quite plausible that exchange rates wi! ll be better predictors of exogenous commodity prices than vice-versa, because the exchange rate is fundamentally forward looking. Therefore, following Campbell and Shiller (1987) and Engel and West (2005), the exchange rate is likely to embody important information about future commodity price movements well beyond what econometricians can capture with simple time series models. In contrast, prices for most commodities are extremely sensitive to small shocks to current demand and supply, and are therefore likely to be less forward looking. J.E.L. Codes: C52, C53, F31, F47. Key words: Exchange rates, forecasting, commodity prices, random walk. Acknowledgements. We would like to thank C. Burnside, C. Engel, M. McCracken, R. Startz, V. Stavreklava, A. Tarozzi, M. Yogo and seminar participants at the University of Washington for comments. We are also grateful to various staff members of the Reserve Bank of Australia, the Bank of Canada, the Reserve Bank of New Zealand, and the! IMF for helpful discussions and for providing some of the data used i n this paper.

  • What does a financial system say about future economic growth?
    Date: 2008-09-12
    By: Grabowski, Szymon
    In many research studies it is argued that it is possible to extract useful information about future economic growth from the performance of financial markets. However, this study goes further and shows that it is not only possible to use expectations derived from financial markets to forecast future economic growth, but that data about the financial system can be used for this purpose as well. The research is conducted for the Polish emerging economy on the basis of monthly data. The results suggest that, based purely on the data from the financial system, it is possible to construct reasonable measures that can, even for an emerging economy, effectively forecast future real economic activity. The outcomes are proved by two various econometric methods, namely, by a time series analysis and by a probit model. All presented models are tested in-sample and out-of-sample.
    Keywords: CCAPM; economic growth; financial markets; term spreads; expectations; forecasting
    JEL: E43 G12 E44

  • Tactical and Strategic Sales Management for Intelligent Agents Guided By Economic Regimes
    Date: 2008-10-20
    By: Ketter, W.
    Collins, J.
    Gini, M.
    Gupta, A.
    Schrater, P. (Erasmus Research Institute of Management (ERIM), RSM Erasmus University)
    We present a computational approach that autonomous software agents can adopt to make tactical decisions, such as product pricing, and strategic decisions, such as product mix and production planning, to maximize profit in markets with supply and demand uncertainties. Using a combination of machine learning and optimization techniques, the agent is able to characterize economic regimes, which are historical microeconomic conditions reflecting situations such as over-supply and scarcity. We assume an agent is capable of using real-time observable information to identify the current dominant market condition and we show how it can forecast regime changes over a planning horizon. We demonstrate how the agent can then use regime characterization to predict prices, price trends, and the probability of receiving a customer order in a dynamic supply chain environment. We validate our methods by presenting experimental re! sults from a testbed derived from the Trading Agent Competition for Supply Chain Management (TAC SCM). The results show that our agent outperforms traditional short- and long-term predictive methodologies (such as exponential smoothing) significantly, resulting in accurate prediction of customer order probabilities, and competitive market prices. This, in turn, has the potential to produce higher profits. We also demonstrate the versatility of our computational approach by applying the methodology to prediction of stock price trends.
    Keywords: agent-mediated electronic commerce;dynamic pricing;machine learning;rational decision making;market forecasting
  • Taken from the NEP-FOR mailing list edited by Rob Hyndman.