nep-for New Economics Papers
on Forecasting
Issue of 2015‒03‒13
twelve papers chosen by
Rob J Hyndman
Monash University

  1. The Out-of-sample Performance of an Exact Median-Unbiased Estimator for the Near-Unity AR(1) Model By Medel, Carlos; Pincheira, Pablo
  2. “Effects of removing the trend and the seasonal component on the forecasting performance of artificial neural network techniques” By Oscar Claveria; Enric Monte; Salvador Torra
  3. Forecasting day-ahead electricity load using a multiple equation time series approach By A.E Clements; A.S Hurn
  4. “Regional Forecasting with Support Vector Regressions: The Case of Spain” By Oscar Claveria; Enric Monte; Salvador Torra
  5. Inflation Dynamics and the Hybrid Neo Keynesian Phillips Curve: The Case of Chile By Medel, Carlos
  6. “Multiple-input multiple-output vs. single-input single-output neural network forecasting” By Oscar Claveria; Enric Monte; Salvador Torra
  7. Fractional Integration of the Price-Dividend Ratio in a Present-Value Model of Stock Prices By Adam Goliński; João Madeira; Dooruj Rambaccussing
  8. Bayesian nonparametric calibration and combination of predictive distributions By Federico Bassetti; Roberto Casarin; Francesco Ravazzolo
  9. Predictability of the daily high and low of the S&P 500 index By Jones, Clive
  10. Forecast Accuracy of Small and Large Scale Dynamic Factor Models in Developing Economies By Germán López Espinosa
  11. “Self-organizing map analysis of agents' expectations. Different patterns of anticipation of the 2008 financial crisis” By Oscar Claveria; Enric Monte; Salvador Torra
  12. EuroMInd-D: A Density Estimate of Monthly Gross Domestic Product for the Euro Area By Tommaso Proietti; Martyna Marczak; Gianluigi Mazzi

  1. By: Medel, Carlos; Pincheira, Pablo
    Abstract: We analyse the multihorizon forecasting performance of several strategies to estimate the stationary AR(1) model in a near-unity context. We focus on the Andrews' (1993) exact median-unbiased estimator (BC), the OLS estimator, and the driftless random walk (RW). In addition, we explore the forecasting performance of pairwise combinations between these individual strategies. We do this to investigate whether the Andrews' (1993) correction of the OLS downward bias helps in reducing mean squared forecast errors. Via simulations, we find that BC forecasts typically outperform OLS forecasts. When BC is compared to the RW we obtain mixed results, favouring the latter as the persistence of the true process increases. Interestingly, we also find that the combination of BC and RW performs well when the persistence of the process is high.
    Keywords: Near-unity autoregression; median-unbiased estimation; unbiasedness; unit root model; forecasting; forecast combinations
    JEL: C22 C52 C53 C63
    Date: 2015–03–04
  2. By: Oscar Claveria (Faculty of Economics, University of Barcelona); Enric Monte (Polytechnic University of Catalunya); Salvador Torra (Faculty of Economics, University of Barcelona)
    Abstract: This study aims to analyze the effects of data pre-processing on the performance of forecasting based on neural network models. We use three different Artificial Neural Networks techniques to forecast tourist demand: a multi-layer perceptron, a radial basis function and an Elman neural network. The structure of the networks is based on a multiple-input multiple-output setting (i.e. all countries are forecasted simultaneously). We use official statistical data of inbound international tourism demand to Catalonia (Spain) and compare the forecasting accuracy of four processing methods for the input vector of the networks: levels, growth rates, seasonally adjusted levels and seasonally adjusted growth rates. When comparing the forecasting accuracy of the different inputs for each visitor market and for different forecasting horizons, we obtain significantly better forecasts with levels than with growth rates. We also find that seasonally adjusted series significantly improve the forecasting performance of the networks, which hints at the significance of deseasonalizing the time series when using neural networks with forecasting purposes. These results reveal that, when using seasonal data, neural networks performance can be significantly improved by working directly with seasonally adjusted levels.
    Keywords: artificial neural networks, forecasting, multiple-input multiple-output (MIMO), seasonality, detrending, tourism demand, multilayer perceptron, radial basis function, Elman JEL classification: L83, C53, C45, R11
    Date: 2015–01
  3. By: A.E Clements; A.S Hurn
    Keywords: Short-term load forecasting, seasonality, intra-day correlation, recursive equation system.
    JEL: C32 Q41 Q47
    Date: 2014–09–01
  4. By: Oscar Claveria (Faculty of Economics, University of Barcelona); Enric Monte (Polytechnic University of Catalunya); Salvador Torra (Faculty of Economics, University of Barcelona)
    Abstract: This study attempts to assess the forecasting accuracy of Support Vector Regression (SVR) with regard to other Artificial Intelligence techniques based on statistical learning. We use two different neural networks and three SVR models that differ by the type of kernel used. We focus on international tourism demand to all seventeen regions of Spain. The SVR with a Gaussian kernel shows the best forecasting performance. The best predictions are obtained for longer forecast horizons, which suggest the suitability of machine learning techniques for medium and long term forecasting.
    Keywords: Forecasting, support vector regressions, artificial neural networks, tourism demand, Spain JEL classification: C02, C22, C45, C63, E27, R11
    Date: 2015–01
  5. By: Medel, Carlos
    Abstract: It is recognised that the understanding and accurate forecasts of key macroeconomic variables are fundamental for the success of any economic policy. In the case of monetary policy, many efforts have been made towards understanding the relationship between past and expected values of inflation, resulting in the so-called Hybrid Neo-Keynesian Phillips Curve (HNKPC). In this article I investigate to which extent the HNKPC help to explain inflation dynamics as well as its out-of-sample forecast, for the case of the Chilean economy. The results show that the forward-looking component is significative and accounts from 1.58 to 0.40 times the lagged inflation coefficient. Also, I find predictive gains close to 45% (respect to a backward-looking specification) and up to 80% (respect to the random walk) when forecasting at 12-months ahead.
    Keywords: New Keynesian Phillips Curve; inflation forecast; out-of-sample comparisons; survey data; real-time dataset
    JEL: C22 C53 E31 E37 E47
    Date: 2015–03–06
  6. By: Oscar Claveria (Faculty of Economics, University of Barcelona); Enric Monte (Polytechnic University of Catalunya); Salvador Torra (Faculty of Economics, University of Barcelona)
    Abstract: This study attempts to improve the forecasting accuracy of tourism demand by using the existing common trends in tourist arrivals form all visitor markets to a specific destination in a multiple-input multiple-output (MIMO) structure. While most tourism forecasting research focuses on univariate methods, we compare the performance of three different Artificial Neural Networks in a multivariate setting that takes into account the correlations in the evolution of inbound international tourism demand to Catalonia (Spain). We find that the MIMO approach does not outperform the forecasting accuracy of the networks when applied country by country, but it significantly improves the forecasting performance for total tourist arrivals. When comparing the forecast accuracy of the different models, we find that radial basis function networks outperform multilayer-perceptron and Elman networks.
    Keywords: tourism demand, forecasting, multivariate, multiple-output, artificial neural networks JEL classification: C22, C45, C63, L83, R11
    Date: 2015–01
  7. By: Adam Goliński; João Madeira; Dooruj Rambaccussing
    Abstract: We re-examine the dynamics of returns and dividend growth within the present-value model of stock prices. We …nd that the price-dividend ratio exhibits long memory and that this induces antipersistence in returns and dividend growth. As such, traditional return forecasting regressions based on the price-dividend ratio are invalid. These …ndings suggest that expected returns and expected dividend growth should be modelled as ARFIMA processes in the present-value framework. We show this improves the models forecast ability in-sample and out-of-sample.
    Keywords: price-dividend ratio, persistence, fractional integration, return predictability, present-value model
    JEL: G12 C32 C58
    Date: 2015–02
  8. By: Federico Bassetti (University of Pavia); Roberto Casarin (University of Venice); Francesco Ravazzolo (Norges Bank (Central Bank of Norway)and BI Norwegian Business School)
    Abstract: We introduce a Bayesian approach to predictive density calibration and combination that accounts for parameter uncertainty and model set incompleteness through the use of random calibration functionals and random combination weights. Building on the work of Ranjan and Gneiting (2010) and Gneiting and Ranjan (2013), we use infinite beta mixtures for the calibration. The proposed Bayesian nonparametric approach takes advantage of the flexibility of Dirichlet process mixtures to achieve any continuous deformation of linearly combined predictive distributions. The inference procedure is based on Gibbs sampling and allows accounting for uncertainty in the number of mixture components, mixture weights, and calibration parameters. The weak posterior consistency of the Bayesian nonparametric calibration is provided under suitable conditions for unknown true density. We study the methodology in simulation examples with fat tails and multimodal densities and apply it to density forecasts of daily S&P returns and daily maximum wind speed at the Frankfurt airport.
    Keywords: Forecast calibration, Forecast combination, Density forecast, Beta mixtures, Bayesian nonparametrics, Slice sampling.
    JEL: C13 C14 C51 C53
    Date: 2015–02–26
  9. By: Jones, Clive
    Abstract: Ratios involving the current period opening price and the high or low price of the previous period are significant predictors of the current period high or low price for many stocks and stock indexes. This is illustrated with daily trading data from the S&P 500 index. Regressions specifying these “proximity variables” have higher explanatory and predictive power than benchmark autoregressive and “no change” models. This is shown with out-of-sample comparisons of MAPE, MSE, and the proportion of time models predict the correct direction or sign of change of daily high and low stock prices. In addition, predictive models incorporating these proximity variables show time varying effects over the study period, 2000 to February 2015. This time variation looks to be more than random and probably relates to investor risk preferences and changes in the general climate of investment risk.
    Keywords: predictability of stock prices, time varying parameters, proximity variable method for predicting stock prices, accuracy of proximity variable method compared with autoregressive and benchmark forecasts
    JEL: C32 C58 G11 G17
    Date: 2015–03–01
  10. By: Germán López Espinosa (Universidad de Navarra)
    Abstract: This paper compares forecast accuracy of two Dynamic Factor Models in a context of constraints interms of data availability. Estimation technique and properties of the factor decomposition depend onthe cross section dimension of the dataset included in each model: a large dataset composed by seriesbelonging to seven broad categories or a small dataset with a few prescreened variables. Short term outof-sample forecast of GDP growth is carried out with both models reproducing the real time situationof data accessibility derived from the publication lags of the series in six Latin American countries.Results show i) the important role of the inclusion of latest released data in the forecast accuracy ofboth models, ii) the better precision of predictions based on factors with respect to autoregressivemodels and iii) identify the most adequate model for each of these six countries in different temporalhorizons.
    Keywords: Factor models, nowcast, forecast, real time, developing economies
    JEL: C32 C53 E37 O54
    Date: 2015–02
  11. By: Oscar Claveria (Faculty of Economics, University of Barcelona); Enric Monte (Polytechnic University of Catalunya); Salvador Torra (Faculty of Economics, University of Barcelona)
    Abstract: By means of Self-Organizing Maps we cluster fourteen European countries according to the most suitable way to model their agents’ expectations. Using the financial crisis of 2008 as a benchmark, we distinguish between those countries that show a progressive anticipation of the crisis and those where sudden changes in expectations occur. By mapping the trajectory of economic experts’ expectations prior to the recession we find that when there are brisk changes in expectations before impending shocks, Artificial Neural Networks are more suitable than time series models for modelling expectations. Conversely, in countries where expectations show a smooth transition towards recession, ARIMA models show the best forecasting performance. This result demonstrates the usefulness of clustering techniques for selecting the most appropriate method to model and forecast expectations according to their behaviour.
    Keywords: Business surveys; Self-Organizing Maps; Clustering; Forecasting; Neural networks; Time series models; Nonlinear models JEL classification:C02; C22; C45; C63; E27
    Date: 2015–03
  12. By: Tommaso Proietti (University of Rome “Tor Vergata" and CREATES); Martyna Marczak (University of Hohenheim); Gianluigi Mazzi (Statistical Office of the European Communities)
    Abstract: EuroMInd-D is a density estimate of monthly gross domestic product (GDP) constructed according to a bottom–up approach, pooling the density estimates of eleven GDP components, by output and expenditure type. The components density estimates are obtained from a medium-size dynamic factor model of a set of coincident time series handling mixed frequencies of observation and ragged–edged data structures. They reflect both parameter and filtering uncertainty and are obtained by implementing a bootstrap algorithm for simulating from the distribution of the maximum likelihood estimators of the model parameters, and conditional simulation filters for simulating from the predictive distribution of GDP. Both algorithms process sequentially the data as they become available in real time. The GDP density estimates for the output and expenditure approach are combined using alternative weighting schemes and evaluated with different tests based on the probability integral transform and by applying scoring rules.
    Keywords: Density Forecast Combination and Evaluation, Mixed–Frequency Data, Dynamic Factor Models, State Space Models
    JEL: C32 C52 C53 E37
    Date: 2015–02–24

This nep-for issue is ©2015 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.