nep-for New Economics Papers
on Forecasting
Issue of 2019‒03‒25
eight papers chosen by
Rob J Hyndman
Monash University

  1. Calibrating GDP fan charts using probit models with a comparison to the approaches of the Bank of England and Riksbank By David Turner; Thomas Chalaux
  2. Forecasting the capacity of mobile networks By Bastos, João A.
  3. Forecasting Equity Index Volatility by Measuring the Linkage among Component Stocks By Qiu, Yue; Xie, Tian; Yu, Jun; Zhou, Qiankun
  4. Estimating Dynamic Conditional Spread Densities to Optimise Daily Storage Trading of Electricity By Ekaterina Abramova; Derek Bunn
  5. Quarterly Forecasting Model for India’s Economic Growth: Bayesian Vector Autoregression Approach By Sen Gupta, Abhijit; Iyer, Tara
  6. Multimodal Deep Learning for Finance: Integrating and Forecasting International Stock Markets By Sang Il Lee; Seong Joon Yoo
  7. Prediction Markets and Poll Releases: When Are Prices Most Informative? By Alasdair Brown; James Reade; Leighton Vaughan Williams
  8. Bayesian MIDAS Penalized Regressions: Estimation, Selection, and Prediction By Matteo Mogliani

  1. By: David Turner; Thomas Chalaux
    Abstract: Fan charts were pioneered by the Bank of England and Riksbank and provide a visuallyappealing means to convey the uncertainty surrounding a forecast. This paper describes amethod for parameterising fan charts around GDP growth forecasts by which the degree ofuncertainty is based on past forecast errors, but the skew is derived from a probit modelbasedassessment of the probability of a future downturn. The probit-based fan chartsclearly out-perform the Bank of England and Riksbank approaches when applied toforecasts made immediately preceding the Global Financial Crisis. These examples alsohighlight weaknesses with the Bank of England and Riksbank approaches.The Riksbank approach implicitly assumes that forecast errors are normallydistributed, but over a long track record this is unlikely to be the case becauseforecasters are generally poor at predicting downturns, which leads to bias and skewin the pattern of forecast errors. Thus, the Riksbank fan chart is neither an accuraterepresentation of past forecast errors, nor is it a reflection of the risk assessmentunderlying the forecast.The Bank of England approach relies heavily on the judgment of the members ofthe Monetary Policy Committee to assess risks. However, even when they havecorrectly foreseen the nature of future risks, the quantitative translation of theserisks into the fan chart skew has been too timid. Perhaps one reason for this is thatthe fan chart prediction intervals based on historical forecast errors already appearquite wide so that inflating them by adding skew may appear embarrassing (at leastex ante).The approach advocated in this paper addresses these weaknesses by recognising thatforecast errors are not symmetrical: firstly, this leads to more compressed predictionintervals in the upper part of the fan chart (representing the possibility of under-prediction);and secondly, using the large forecast errors from past downturns to calibrate downwardskew clearly supports a more bold approach when there is a risk of a downturn. A weaknessof the probit model-based approach is that it will not predict atypical downturns. Forexample, in the current conjuncture it would not pick up risks associated with a ‘no deal’Brexit or a global trade war. However, a downturn triggered by atypical events may bemore severe if risk factors describing a typical business-financial cycle are also high.
    Keywords: downturn, economic forecasts, fan charts, recession, risk, uncertainty
    JEL: E58 E17 E65 E66 E01
    Date: 2019–03–08
  2. By: Bastos, João A.
    Abstract: The optimization of mobile network capacity usage is an essential operation to promote positive returns on network investments, prevent capacity bottlenecks, and deliver good end user experience. This study examines the performance of several statistical models to predict voice and data traffic in a mobile network. While no method dominates the others across all time series and prediction horizons, exponential smoothing and ARIMA models are good alternatives to forecast both voice and data traffic. This analysis shows that network managers have at their disposal a set of statistical tools to plan future capacity upgrades with the most effective solution, while optimizing their investment and maintaining good network quality.
    Keywords: Mobile networks, Forecasting, ARIMA models, Exponential smoothing, Time series
    JEL: C53 O32
    Date: 2019–03–13
  3. By: Qiu, Yue (WISE and School of Economics, Xiamen University); Xie, Tian (School of Economics, Singapore Management University); Yu, Jun (School of Economics and Lee Kong Chian School of Business, Singapore Management University); Zhou, Qiankun (Department of Economics, Louisiana State University)
    Abstract: The linkage among the realized volatilities across component stocks are important when modeling and forecasting the relevant index volatility. In this paper, the linkage is measured via an extended Common Correlated Effects (CCE) approach under a panel heterogeneous autoregression model where unobserved common factors in errors are assumed. Consistency of the CCE estimator is obtained. The common factors are extracted using the principal component analysis. Empirical studies show that realized volatility models exploiting the linkage effects lead to significantly better out-of-sample forecast performance, for example, an up to 32% increase in the pseudo R2. We also conduct various forecasting exercises on the the linkage variables that compare conventional regression methods with popular machine learning techniques.
    Keywords: Volatility Forecasting; Heterogeneous autoregression; Common correlated effect; Factor analysis; Random forest
    JEL: C31 C32 G12 G17
    Date: 2019–03–02
  4. By: Ekaterina Abramova; Derek Bunn
    Abstract: This paper formulates dynamic density functions, based upon skewed-t and similar representations, to model and forecast electricity price spreads between different hours of the day. This supports an optimal day ahead storage and discharge schedule, and thereby facilitates a bidding strategy for a merchant arbitrage facility into the day-ahead auctions for wholesale electricity. The four latent moments of the density functions are dynamic and conditional upon exogenous drivers, thereby permitting the mean, variance, skewness and kurtosis of the densities to respond hourly to such factors as weather and demand forecasts. The best specification for each spread is selected based on the Pinball Loss function, following the closed form analytical solutions of the cumulative density functions. Those analytical properties also allow the calculation of risk associated with the spread arbitrages. From these spread densities, the optimal daily operation of a battery storage facility is determined.
    Date: 2019–03
  5. By: Sen Gupta, Abhijit (Asian Development Bank); Iyer, Tara (Asian Development Bank)
    Abstract: This study develops a framework to forecast India’s gross domestic product growth on a quarterly frequency from 2004 to 2018. The models, which are based on real and monetary sector descriptions of the Indian economy, are estimated using Bayesian vector autoregression (BVAR) techniques. The real sector groups of variables include domestic aggregate demand indicators and foreign variables, while the monetary sector groups specify the underlying inflationary process in terms of the consumer price index (CPI) versus the wholesale price index given India’s recent monetary policy regime switch to CPI inflation targeting. The predictive ability of over 3,000 BVAR models is assessed through a set of forecast evaluation statistics and compared with the forecasting accuracy of alternate econometric models including unrestricted and structural VARs. Key findings include that capital flows to India and CPI inflation have high informational content for India’s GDP growth. The results of this study provide suggestive evidence that quarterly BVAR models of Indian growth have high predictive ability.
    Keywords: Bayesian vector autoregressions; GDP growth; India; time series forecasting
    JEL: C11 C32 C53 F43
    Date: 2019–03–14
  6. By: Sang Il Lee; Seong Joon Yoo
    Abstract: Stock prices are influenced by numerous factors. We present a method to combine these factors and we validate the method by taking the international stock market as a case study. In today's increasingly international economy, return and volatility spillover effects across international equity markets are major macroeconomic drivers of stock dynamics. Thus, foreign market information is one of the most important factors in forecasting domestic stock prices. However, the cross-correlation between domestic and foreign markets is so complex that it would be extremely difficult to express it explicitly with a dynamical equation. In this study, we develop stock return prediction models that can jointly consider international markets, using multimodal deep learning. Our contributions are three-fold: (1) we visualize the transfer information between South Korea and US stock markets using scatter plots; (2) we incorporate the information into stock prediction using multimodal deep learning; (3) we conclusively show that both early and late fusion models achieve a significant performance boost in comparison with single modality models. Our study indicates that considering international stock markets jointly can improve prediction accuracy, and deep neural networks are very effective for such tasks.
    Date: 2019–03
  7. By: Alasdair Brown (University of East Anglia); James Reade (Department of Economics, University of Reading); Leighton Vaughan Williams (Nottingham Business School)
    Abstract: Prediction markets are a popular platform for eliciting incentivised crowd predictions. In this paper, we examine variation in the information contained in prediction market prices by studying Intrade prices on U.S. elections around the release of opinion polls. We find that poll releases stimulate an immediate uptick in trading activity. However, much of this activity involves relatively inexperienced traders and, as a result, price efficiency declines in the immediate aftermath of a poll release. It is not until more experienced traders enter the market in the following ours that price efficiency recovers. More generally, this suggests that information releases do not necessarily improve prediction market forecasts, but may instead attract noise traders who temporarily reduce price efficiency.
    Keywords: prediction markets, opinion polls, price efficiency, information efficiency
    JEL: C53 G14 D82 D83
    Date: 2018–03–20
  8. By: Matteo Mogliani
    Abstract: We propose a new approach to mixed-frequency regressions in a high-dimensional environment that resorts to Group Lasso penalization and Bayesian techniques for estimation and inference. To improve the sparse recovery ability of the model, we also consider a Group Lasso with a spike-and-slab prior. Penalty hyper-parameters governing the model shrinkage are automatically tuned via an adaptive MCMC algorithm. Simulations show that the proposed models have good selection and forecasting performance, even when the design matrix presents high cross-correlation. When applied to U.S. GDP data, the results suggest that financial variables may have some, although limited, short-term predictive content.
    Date: 2019–03

This nep-for issue is ©2019 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.