nep-for New Economics Papers
on Forecasting
Issue of 2013‒05‒11
eight papers chosen by
Rob J Hyndman
Monash University

  1. Forecasting Stock Market Volatility: A Forecast Combination Approach By Nazarian, Rafik; Gandali Alikhani, Nadiya; Naderi, Esmaeil; Amiri, Ashkan
  2. Sectoral gross value-added forecasts at the regional level: Is there any information gain? By Lehmann, Robert; Wohlrabe, Klaus
  3. Delusions of Success: Comment on Dan Lovallo and Daniel Kahneman By Bent Flyvbjerg
  4. Structural-break models under mis-specification: implications for forecasting By Boonsoo Koo; Myung Hwan Seo
  5. An application of learning machines to sales forecasting under promotions By Gianni Di Pillo; Vittorio Latorre; Stefano Lucidi; Enrico Procacci
  6. Heterogeneous Beliefs and Prediction Market Accuracy By He, Xue-Zhong; Treich, Nicolas
  7. Does Specialization Matter for Trade Imbalance at Industry Level? By E. Yong Song; Chen Zhao
  8. Predicting agricultural impacts of large-scale drought: 2012 and the case for better modeling By Joshua Elliot; Michael Glotter; Neil Best; Ken Boote; Jim Jones; Jerry Hatfield; Cynthia Rozenweig; Leonard A. Smith; Ian Foster

  1. By: Nazarian, Rafik; Gandali Alikhani, Nadiya; Naderi, Esmaeil; Amiri, Ashkan
    Abstract: Recently, with the development of financial markets and due to the importance of these markets and their close relationship with other macroeconomic variables, using advanced mathematical models with complicated structures for forecasting these markets has become very popular. Besides, neural network models have gained a special position compared to other advanced models due to their high accuracy in forecasting different variables. Therefore, the main purpose of this study was to forecast the volatilities of TSE index by regressive models with long memory feature, feed forward neural network and hybrid models (based on forecast combination approach) using daily data. The results were indicative of the fact that based on the criteria for assessing forecasting error, i.e., MSE and RMSE, although forecasting errors of the feed forward neural network model were less than ARFIMA-FIGARCH model, the accuracy of the hybrid model of neural network and best GARCH was higher than each one of these models.
    Keywords: Stock Return, Long Memory, Neural Network, Hybrid Models.
    JEL: C14 C22 C45 C53
    Date: 2013–03–15
  2. By: Lehmann, Robert; Wohlrabe, Klaus
    Abstract: In this paper, we ask whether it is possible to forecast gross-value added (GVA) and its sectoral sub-components at the regional level. We are probably the first who evaluate sectoral forecasts at the regional level using a huge data set at quarterly frequency to investigate this issue. With an autoregressive distributed lag model we forecast total and sectoral GVA for one of the German states (Saxony) with more than 300 indicators from different regional levels (international, national and regional) and additionally make usage of different pooling strategies. Our results show that we are able to increase forecast accuracy of GVA for every sector and for all forecast horizons compared to an autoregressive process. Finally, we show that sectoral forecasts contain more information in the short term (one quarter), whereas direct forecasts of total GVA are preferable in the medium (two and three quarters) and long term (four quarters).
    Keywords: regional forecasting; gross value added; leading indicators; forecast combination; disaggregated forecasts
    JEL: C32 C52 C53 E37 R11
    Date: 2013–05–06
  3. By: Bent Flyvbjerg
    Abstract: Dan Lovallo and Daniel Kahneman must be commended for their clear identification of causes and cures to the planning fallacy in "Delusions of Success: How Optimism Undermines Executives' Decisions" (HBR July 2003). Their look at overoptimism, anchoring, competitor neglect, and the outside view in forecasting is highly useful to executives and forecasters. However, Lovallo and Kahneman underrate one source of bias in forecasting - the deliberate "cooking" of forecasts to get ventures started.
    Date: 2013–04
  4. By: Boonsoo Koo; Myung Hwan Seo
    Abstract: This paper revisits the least squares estimator of the linear regression with a structural break. We view the model as an approximation to the true data generating process whose exact nature is unknown but perhaps changing over time either continuously or with some jumps. This view is widely held in the forecasting literature and under this view, the time series dependence property of all the observed variables is unstable as well. We establish that the rate of convergence of the estimator to a properly defined limit is much slower than the standard super consistent rate, even slower than the square root of the sample size T and as slow as the cube root of T. We also provide an asymptotic distribution of the estimator and that of the Gaussian quasi likelihood ratio statistic for a certain class of true data generating process. We relate our finding to current forecast combination methods and bagging and propose a new averaging scheme. The performance of various contemporary forecasting methods is compared to ours using a number of macroeconomic data.
    Keywords: structural breaks, forecasting, mis-specification, cube-root asymptotics, bagging
    Date: 2013
  5. By: Gianni Di Pillo (Department of Computer, Control and Management Engineering, Universita' degli Studi di Roma "La Sapienza"); Vittorio Latorre (Department of Computer, Control and Management Engineering, Universita' degli Studi di Roma "La Sapienza"); Stefano Lucidi (Department of Computer, Control and Management Engineering, Universita' degli Studi di Roma "La Sapienza"); Enrico Procacci (ACT Solutions)
    Abstract: This paper deals with sales forecasting in retail stores of large distribution. For severalyears statistical methods such as ARIMA and Exponential Smoothing have been usedto this aim. However the statistical methods could fail if high irregularity of sales arepresent, as happens in case of promotions, because they are not well suited to modelthe nonlinear behaviors of the sales process. In the last years new methods basedon Learning Machines are being employed for forecasting problems. These methodsrealize universal approximators of non linear functions, thus resulting more able tomodel complex nonlinear phenomena. The paper proposes an assessment of the use ofLearning Machines for sales forecasting under promotions, and a comparison with thestatistical methods, making reference to two real world cases. The learning machineshave been trained using several conguration of input attributes, to point out theimportance of a suitable inputs selection.
    Keywords: Learning Machines; Neural networks; Radial basis functions; Support vector machines; Sales forecasting; Promotion policies; Nonlinear optimization
    Date: 2013–04
  6. By: He, Xue-Zhong; Treich, Nicolas
    Abstract: We consider a prediction market in which traders have heterogeneous prior beliefs in probabilities. In the two-state case, we derive necessary and sufficient conditions so that the prediction market is accurate in the sense that the equilibrium state price equals the mean probabilities of traders' beliefs. We also provide a necessary and sufficient condition for the well documented favorite-longshot bias. In an extension to many states, we revisit the results of Varian (1985) on the relationship between equilibrium state price and belief heterogeneity.
    Keywords: Prediction market, heterogeneous beliefs, risk aversion, favorite-longshot bias, complete markets, and asset prices.
    Date: 2012–08–20
  7. By: E. Yong Song (Department of Economics, Sogang University); Chen Zhao (Department of Economics, Hong Kong University of Science and Technology, Clear Water Bay, Hong Kong)
    Abstract: This paper investigates the source of bilateral trade imbalance at industry level. We build a simple model based on gravity theory and derive the prediction that the bilateral trade balance in an industry is increasing in the difference between trading partners in the output share of the industry. We test this prediction and find that the difference in industry share is highly significant in predicting both the sign and the magnitude of trade balance at industry level. We also find that FTAs tend to enlarge trade imbalance at industry level. However, the overall predictive power of the model is rather limited, suggesting that factors other than production specialization are important in determining trade balance at industry level. Another finding of the paper is that the influence of the difference in industry share on trade balance increases as we move to industries that produce more homogeneous products. This finding calls into question monopolistic competition as the main driver of gravity in international trade.
    Keywords: trade imbalance, gravity theory, specialization, output share, homogeneous products
    Date: 2012–08
  8. By: Joshua Elliot; Michael Glotter; Neil Best; Ken Boote; Jim Jones; Jerry Hatfield; Cynthia Rozenweig; Leonard A. Smith; Ian Foster
    Abstract: Not available.
    Date: 2013–04

This nep-for issue is ©2013 by Rob J Hyndman. It is provided as is without any express or implied warranty. It may be freely redistributed in whole or in part for any purpose. If distributed in part, please include this notice.
General information on the NEP project can be found at For comments please write to the director of NEP, Marco Novarese at <>. Put “NEP” in the subject, otherwise your mail may be rejected.
NEP’s infrastructure is sponsored by the School of Economics and Finance of Massey University in New Zealand.