In this issue we have Performance of combined double seasonal univariate time series models for forecasting water demand, Forecasting with Universal Approximators and a Learning Algorithm, Forecasting bank loans loss-given-default, Forecasting and Structural Change and more.
|By:||Jorge Caiado (CEMAPRE, School of Economics and Management (ISEG), Technical University of Lisbon)|
|In this article, we examine the daily water demand forecasting performance of double seasonal univariate time series models (Holt-Winters, ARIMA and GARCH) based on multi-step ahead forecast mean squared errors. A within-week seasonal cycle and a within-year seasonal cycle are accommodated in the various model specifications to capture both seasonalities. We investigate whether combining forecasts from different methods for different origins and horizons could improve forecast accuracy. The analysis is made with daily data for water consumption in Granada, Spain.|
|Keywords:||ARIMA, Combined forecasts, Double seasonality, Exponential Smoothing, Forecasting, GARCH, Water demand|
|By:||Anders Bredahl Kock (Aarhus University and CREATES)|
|This paper applies three universal approximators for forecasting. They are the Artificial Neural Networks, the Kolmogorov- Gabor polynomials, as well as the Elliptic Basis Function Networks. Even though forecast combination has a long history in econometrics focus has not been on proving loss bounds for the combination rules applied. We apply the Weighted Average Algorithm (WAA) of Kivinen and Warmuth (1999) for which such loss bounds exist. Specifically, one can bound the worst case performance of the WAA compared to the performance of the best single model in the set of models combined from. The use of universal approximators along with a combination scheme for which explicit loss bounds exist should give a solid theoretical foundation to the way the forecasts are performed. The practical performance will be investigated by considering various monthly postwar macroeconomic data sets for the G7 as well as the Scandinavi! an countries.|
|Keywords:||Forecasting, Universal Approximators, Elliptic Basis Function Network, Forecast Combination, Weighted Average Algorithm|
Simon van Norden
|A probabilistic forecast is the estimated probability with which a future event will satisfy a particular 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 gropuing 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 pseudo-forecasts made using the data vintage available at the forecast date. We evaluate outcomes using both first-release outcome measures as well as later, thoroug! hly revised data. We find strong evidence of incorrect calibration in professional forecasts of recessions and inflation. We also present evidence of asymmetries in the performace of inflation forecasts based on real-time output gaps.|
|By:||Joao A. Bastos (CEMAPRE, School of Economics and Management (ISEG), Technical University of Lisbon)|
|With the advent of the new Basel Capital Accord, banking organizations are invited to estimate credit risk capital requirements using an internal ratings based approach. In order to be compliant with this approach, institutions must estimate the expected loss-given-default, the fraction of the credit exposure that is lost if the borrower defaults. This study evaluates the ability of a parametric fractional response regression and a nonparametric regression tree model to forecast bank loan credit losses. The out-of-sample predictive ability of these models is evaluated at several recovery horizons after the default event. The out-of-time predictive ability is also measured for a recovery horizon of one year. The performance of the models is benchmarked against recovery estimates given by a naive model in which predicted recoveries are given by historical averages.|
|Keywords:||Forecasting, bank loans, loss-given-default, fractional response regression, regression trees|
|Financial returns typically display heavy tails and some skewness, and cinditional vairance models with these features often outperform more limited models. The difference in performance may be especially important in estimating quantities that depend on tail features, including risk measures such as the expected shortfall. Here, using a recent generalization of the asymmetric Student-t distribution to allow separate parameters to control skewness and the thickness of each tail, we fit daily financial returns and forecast expected shortfall for the S&P 500 composite index; the generalized distribution is used for the standardized innovations in a nonlinear, asymmetric GARCH-type model. The results provide empirical evidence for the usefulness of the generalized distribution in improving prediction of downside market risk of financial assets.|
|The purpose of this study was to analyze structural changes that took place in the cotton industry in recent years and develop a statistical model that reflects the current drivers of U.S. cotton prices. Legislative changes authorized the U.S. Department of Agriculture to resume publishing cotton price forecasts for the first time in 79 years. In addition, systematic problems have become apparent in the forecasting models used by USDA and elsewhere, highlighting the need for an updated review of price relationships. This study concluded that a structural break in the U.S. cotton industry occurred in 1999, and that world cotton supply has become an important determinant of U.S. cotton prices. Chinaâs trade and production policy also continues to be an important factor in price determination. The model developed here forecasts changes in the U.S. upland cotton farm price based on changes in U.S. cotton supply, changes in ! U.S. stocks-to-use ratio (S/U), changes in Chinaâs net imports as a share of world consumption, selected farm policy parameters, and changes in the foreign supply of cotton.|
|Keywords:||forecasting, cotton, price, demand, trade, structural change, farm programs., Demand and Price Analysis, Q100, Q110, Q130,|
|Over the last few years, the prices of the main agricultural raw materials have been highly volatile. The situation is unprecedented, both in the magnitude of the upward and downward volatility observed, and in the number of agricultural commodities affected. Various factors are contributing to these contrasting shifts: the role of emerging countries, changing dietary habits, an increase in energy demand related to the boom in biofuels, adverse weather conditions and speculation. In this paper we try to capture long-term relationships between crop prices and crude oil price using a partial equilibrium and times series method. The study finds little empirical evidence that the crude oil price have a significant influence on the variation of major vegetable crops prices|
|Keywords:||Partial equilibrium modeling, Forecasting cointegration, Demand and Price Analysis, Q11, Q13, Q42,|
Lars Peter Stentoft
|While stochastic volatility models improve on the option pricing error when compared to the Black-Scholes-Merton model, mispricings remain. This paper uses mixed normal heteroskedasticity models to price options. Our model allows for significant negative skewness and time varying higher order moments of the risk neutral distribution. Parameter inference using Gibbs sampling is explained and we detail how to compute risk neutral predictive densities taking into account parameter uncertainty. When forecasting out-of-sample options on the S&P 500 index, substantial improvements are found compared to a benchmark model in terms of dollar losses and the ability to explain the smirk in implied volatilities. <P>Les modèles à volatilité stochastique apportent des améliorations en ce qui a trait à l'erreur d'établissement des prix des options comparativement au modèle de Black-Scholes-Merton. Toutefois, la fix! ation incorrecte des prix persiste. Le présent document a recours à des modèles mixtes avec hétéroscédasticité normale pour fixer les prix des options. Notre modèle permet de tenir compte de l'asymétrie négative importante et des moments d'ordre élevé variant dans le temps liés à la distribution du risque nul. Nous expliquons l'inférence des paramètres selon l'échantillonnage de Gibbs et détaillons la façon de traiter les densités prédictives de risque neutre en prenant en considération l'incertitude des paramètres. Dans le cas des prévisions concernant les options hors-échantillonnage sur l'indice S&P 500, nous constatons des améliorations importantes, par rapport à un modèle de référence, en termes de pertes exprimées en dollars et de capacité d'expliquer l'ironie des volatilités implicites.|
|Keywords:||Bayesian inference, option pricing, finite mixture models, out-of-sample prediction, GARCH models, Inférence bayésienne, fixation du prix des options, modèles à mélanges finis, prédiction hors-échantillon, modèles GARCH.|
|By:||Glaser, Markus (Sonderforschungsbereich 504)
Langer, Thomas (Westfälischen Wilhelms-Universität Münster Lehrstuhl für BWL, insbesondere Finanzierung)
Weber, Martin (Lehrstuhl für ABWL, Finanzwirtschaft, insb. Bankbetriebslehre)
|In this study, we analyze whether volatility forecasts (judgmental confidence intervals) are influenced by the specific elicitation mode (i.e. whether forecasters have to state future price levels or directly future returns as upper and lower bounds). We present questionnaire responses of about 250 students from two German universities. Participants were asked to state median forecasts as well as confidence intervals for seven stock market time series. Using a between subject design, one half of the subjects was asked to state future price levels, the other group was directly asked for returns. Consistent with prior research we find that subjects underestimate the volatility of stock returns, indicating overconfidence. As a new insight, we find that the strength of the overconfidence effect in stock market forecasts is highly significantly affected by the fact whether subjects provide price or return forecasts. Volatilit! y estimates are lower (and the overconfidence bias is thus stronger) when subjects are asked for returns compared to price forecasts.|
|Previously, linear trends were revealed in the differences between the headline CPI and the price indices for various subcategories of the CPI in the United States. These trends can be continuous, as observed with the price index for medical care, or piecewise with turning points between trends with opposite signs. Similar features are found for the PPI and its components. The presence of sustainable trends in the differences allows prediction of prices for various commodities at time horizons of several years. In addition, it is possible to time the start of transition to the next trend. Accordingly, the trends reduce the uncertainty in forecasting prices for major commodities and also for their small components. The usage of trends in the PPI could bring substantial benefits to producers (planning) and stock market participants (timely investment).|
|Keywords:||PPI; CPI; US; prediction|
|Tools and approaches are provided for nonlinear time series modelling in econometrics. A wide range of topics is covered, including probabilistic properties, statistical inference and computational methods. The focus is on the applications but the ideas of the mathematical arguments are also provided. Techniques and concepts are illustrated by various examples, Monte Carlo experiments and a real application.|
|Keywords:||Consistency and asymptotic normality; MCMC algorithms; Mixing; Nonlinear modelling; Stationarity; Time-series forecasting.|
Horan, Richard D.
Wolf, Christopher A.
|Bovine tuberculosis (bTB) in cattle has caused significant economic losses to livestock producers and has proven difficult to eradicate. It is suspected that cattle movement across different farms and regions is one of the key factors of bTB transmission in the United States. Prior attempts to model the epidemiology of bTB infection within cattle to predict disease transmission have not adequately captured the behavioral aspects of trade. A better understanding of livestock trade patterns would help in predicting disease transmission and the associated economic effects. In this paper, we develop a gravity model of livestock trade and link it to an epidemiological model of bTB transmission, with the goal being that this information could lead to improved disease surveillance and management. Our findings suggest that feedbacks between jointly determined disease dynamics and trade system matter and should be considered toge! ther for efficient disease management.|
|Keywords:||Bovine tuberculosis, Gravity model, disease management, Resource /Energy Economics and Policy,|
Taken from the NEP-FOR mailing list edited by Rob Hyndman.