nep-for New Economics Papers
on Forecasting
Issue of 2019‒12‒02
ten papers chosen by
Rob J Hyndman
Monash University

  1. A Moving Average Heterogeneous Autoregressive Model for Forecasting the Realized Volatility of the US Stock Market: Evidence from Over a Century of Data By Afees A. Salisu; Rangan Gupta; Ahamuefula E. Ogbonna
  2. Lithuanian house price index: modelling and forecasting By Laurynas Narusevicius; Tomas Ramanauskas; Laura Gudauskaitė; Tomas Reichenbachas
  3. Forecasting with instabilities: an application to DSGE models with financial frictions By Roberta Cardani; Alessia Paccagnini; Stefania Villa
  4. Modelling and forecasting wind drought By Gunnar Bårdsen; Stan Hurn; Kenneth Lindsay
  5. Predictive properties of forecast combination, ensemble methods, and Bayesian predictive synthesis By Kosaku Takanashi; Kenichiro McAlinn
  6. Forecasting Bitcoin Returns: Is there a Role for the U.S. – China Trade War? By Vasilios Plakandaras; Elie Bouri; Rangan Gupta
  7. Forecast-Hedging and Calibration By Sergiu Hart; Dean P. Foster
  8. Forecasting inflation in the euro area: countries matter! By Angela Capolongo; Claudia Pacella
  9. Modeling, Forecasting, and Nowcasting U.S. CO2 Emissions Using Many Macroeconomic Predictors By Mikkel Bennedsen; Eric Hillebrand; Siem Jan Koopman
  10. Modeling and forecasting demand for electricity in Zimbabwe using the Box-Jenkins ARIMA technique By NYONI, THABANI

  1. By: Afees A. Salisu (Department for Management of Science and Technology Development, Ton Duc Thang University, Ho Chi Minh City, Vietnam and Faculty of Business Administration, Ton Duc Thang University, Ho Chi Minh City, Vietnam); Rangan Gupta (Department of Economics, University of Pretoria, Pretoria, 0002, South Africa); Ahamuefula E. Ogbonna (Centre for Econometric & Allied Research, University of Ibadan and Department of Statistics, University of Ibadan)
    Abstract: This study forecasts the monthly realized volatility of the US stock market covering the period of February, 1885 to September, 2019 using a recently developed novel approach – a moving average heterogeneous autoregressive (MAT-HAR) model, which treats threshold as a moving average generated time varying parameter rather than as a fixed or unknown parameter. The significance of asymmetric information in realized volatility of stock market forecasting is also considered by examining the case of good and bad realized volatility. The Clark and West (2007) forecast evaluation approach is employed to evaluate the forecast performance of the proposed predictive model vis-à-vis the conventional HAR and threshold HAR (T-HAR) models. We find evidence in favour of the MAT-HAR model relative to the HAR and T-HAR models. Also observed is the significant role of asymmetry in modeling the realized volatility as good realized volatility and bad realized volatility yield dissimilar predictability results. Our results are not sensitive to the choice of sample periods and realized volatility measures.
    Keywords: Realized volatility, US stock market, Forecast evaluation, HAR models
    JEL: C22 C53 G12
    Date: 2019–11
  2. By: Laurynas Narusevicius (Bank of Lithuania); Tomas Ramanauskas; Laura Gudauskaitė (Bank of Lithuania); Tomas Reichenbachas (Bank of Lithuania)
    Abstract: Timely monitoring of the housing market developments in Lithuania is one of the key elements in the analysis framework of the macroprudential authority aiming to contribute to financial stability in Lithuania. In this paper, we addressed three important questions related to Lithuanian house prices, namely, whether house prices are under- or over valuated, which explanatory variables have the biggest impact on the growth of house prices and what the future development of the Lithuanian house price index could be. Three separate modelling and forecasting exercises were performed in order to tackle these questions. The first exercise employs the vector error correction modelling (VECM) approach to assess under- or overvaluation of the house prices. We then use an autoregressive distributed lag model (ARDL) to evaluate which explanatory variables have the biggest impact on house price growth. As the last exercise, we develop a suite of models that are used to forecast future development of the house price index. The analysis presented in this paper may be viewed as a further step towards more formalised modelling and forecasting of the Lithuanian house price index.
    Keywords: House price index, fundamental value, time series models, forecasting, forecast combination
    JEL: C22 C32 C53 E37 R30
    Date: 2019–11–19
  3. By: Roberta Cardani (European Commission); Alessia Paccagnini (University College Dublin); Stefania Villa (Bank of Italy)
    Abstract: We assess the importance of parameter instabilities from a forecasting standpoint in a set of medium-scale DSGE models with and without financial frictions using US real-time data. We find that failing to update DSGE model parameter estimates with new data arrival deteriorates point forecasts due to the estimated parameters variation. We also find that the presence of financial frictions helps to better forecast GDP and inflation.
    Keywords: Bayesian estimation, forecasting, financial frictions, parameter instabilities
    JEL: C11 C13 C32 E37
    Date: 2019–10
  4. By: Gunnar Bårdsen (Department of Economics, Norwegian University of Science and Technology); Stan Hurn (School of Economics and Finance, QUT, Australia); Kenneth Lindsay (Department of Mathematics, University of Glasgow, Scotland)
    Abstract: The paper examines several simple dynamic probit models in terms of their usefulness in forecasting wind drought, defined as 5 or more hours of wind speed less than 3.5 m/sec during the busiest periods of the day for the demand for electricity. Dynamic probit models work well in terms of their ability to forecast and are robust by comparison with an approach based on modelling counts. There seems little advantage to moving to modelling counts unless there is added advantage to market participants in knowing the actual prediction for the number of hours of low wind. Future research should focus on the problem of identifying the first day in a series of days with slow winds, and the first day of reasonable wind after a spell of drought. Both the probit and count models could be improved in this regard.
    JEL: C22 G00
    Date: 2019–11–15
  5. By: Kosaku Takanashi; Kenichiro McAlinn
    Abstract: This paper studies the theoretical predictive properties of classes of forecast combination methods. The study is motivated by the recently developed Bayesian framework for synthesizing predictive densities: Bayesian predictive synthesis. A novel strategy based on continuous time stochastic processes is proposed and developed, where the combined predictive error processes are expressed as stochastic differential equations, evaluated using Ito's lemma. We show that a subclass of synthesis functions under Bayesian predictive synthesis, which we categorize as non-linear synthesis, entails an extra term that "corrects" the bias from misspecification and dependence in the predictive error process, effectively improving forecasts. Theoretical properties are examined and shown that this subclass improves the expected squared forecast error over any and all linear combination, averaging, and ensemble of forecasts, under mild conditions. We discuss the conditions for which this subclass outperforms others, and its implications for developing forecast combination methods. A finite sample simulation study is presented to illustrate our results.
    Date: 2019–11
  6. By: Vasilios Plakandaras (Department of Economics, Democritus University of Thrace, University Campus, Komotini, Greece); Elie Bouri (USEK Business School, Holy Spirit University of Kaslik, Jounieh, Lebanon); Rangan Gupta (Department of Economics, University of Pretoria, Pretoria, 0002, South Africa)
    Abstract: Previous studies provide evidence that trade related uncertainty tends to predict an increase in Bitcoin returns. In this paper, we extend the related literature by examining whether the information on the U.S. – China trade war can be used to forecast the future path of Bitcoin returns controlling for various explanatory variables. We apply ordinary least square (OLS) regression, support vector regression (SVR), and the least absolute shrinkage and selection operator (LASSO) techniques that stem from the field of machine learning, and find weak evidence of the role of the trade war in forecasting Bitcoin returns. Given that out-of-sample tests are more reliable than in-sample tests, our results tend to suggest that future Bitcoin returns are unaffected by trade related uncertainties, and investors can use Bitcoin as a safe haven in this context.
    Keywords: Bitcoin, forecasting, machine learning, U.S. – China trade war
    JEL: C53 G11 G17
    Date: 2019–11
  7. By: Sergiu Hart; Dean P. Foster
    Abstract: Calibration means that for each forecast x the average of the realized actions in the periods in which the forecast was x is, in the long run, close to x. Calibration can always be guaranteed (Foster and Vohra 1998), but it requires the forecasting procedure to be stochastic. By contrast, smooth calibration, which combines in a continuous manner nearby forecasts, can be guaranteed by a deterministic procedure (Foster and Hart 2018). In the present paper we develop the concept of forecast-hedging, which consists of choosing the forecasts in such a way that, no matter what the realized action will be, the expected forecasting track record can only improve. This approach integrates the existing calibration results by obtaining them all from the same simple basic argument, and at the same time differentiates between them according to the forecast-hedging tools that are used: deterministic and fixed point-based vs. stochastic and minimax-based. Additional benefits are new calibration procedures in the one-dimensional case that are simpler than all known such procedures, and a short proof for deterministic smooth calibration, in contrast to the complicated existing proof.
    Date: 2019–11
  8. By: Angela Capolongo (ECARES, Université Libre de Bruxelles); Claudia Pacella (Bank of Italy)
    Abstract: We construct a Bayesian vector autoregressive model with three layers of information: the key drivers of inflation, cross-country dynamic interactions, and country-specific variables. The model provides good forecasting accuracy with respect to the popular benchmarks used in the literature. We perform a step-by-step analysis to shed light on which layer of information is more crucial for accurately forecasting euro area inflation. Our empirical analysis reveals the importance of including the key drivers of inflation and taking into account the multi-country dimension of the euro area. The results show that the complete model performs better overall in forecasting inflation excluding energy and unprocessed food, while a model based only on aggregate euro area variables works better for headline inflation.
    Keywords: inflation, forecasting, euro area, Bayesian estimation
    JEL: C32 C53 E31 E37
    Date: 2019–06
  9. By: Mikkel Bennedsen (Aarhus University and CREATES); Eric Hillebrand (Aarhus University and CREATES); Siem Jan Koopman (Vrije Universiteit Amsterdam and CREATES)
    Abstract: We propose a structural augmented dynamic factor model for U.S. CO2 emissions. Variable selection techniques applied to a large set of annual macroeconomic time series indicate that CO2 emissions are best explained by industrial production indices covering manufacturing and residential utilities sectors. We employ a dynamic factor structure to explain, forecast, and nowcast the industrial production indices and thus, by way of the structural equation, emissions. We show that our model has good in-sample properties and out-of-sample performance in comparison with univariate and multivariate competitor models. Based on data through September 2019, our model nowcasts a reduction of about 2.6% in U.S. CO2 emissions in 2019 compared to 2018 as the result of a reduction in industrial production in residential utilities.
    Keywords: CO2 emissions, macroeconomic variables, dynamic factor model, variable selection, forecasting, nowcasting
    JEL: C01 C13 C32 C51 C52 C53 C55 C82 Q43 Q47
    Date: 2019–11–27
    Abstract: This study, which is the first of its kind in Zimbabwe, uses annual time series data on electricity demand in Zimbabwe from 1971 to 2014, to model and forecast the demand for electricity using the Box-Jenkins ARIMA framework. The study is guided by three objectives and these are: to analyze electricity consumption trends in Zimbabwe over the study period, to develop a reliable electricity demand forecasting model for Zimbabwe based on the Box-Jenkins ARIMA technique and last but not least, to project electricity demand in Zimbabwe over the next decade (2015 – 2025). Diagnostic tests indicate that X is an I (1) variable. Based on Theil’s U, the study presents the ARIMA (1, 1, 6) model, the diagnostic tests further show that this model is stable and hence suitable for forecasting electricity demand in Zimbabwe. The selected optimal model, the ARIMA (1, 1, 6) model proves beyond any reasonable doubt that in the next 10 years (2015 – 2025), demand for electricity in Zimbabwe will continue to fall. Amongst other policy recommendations, the study advocates for the liberalization of the electricity power sector in Zimbabwe in order to pave way for more efficient private investment whose potential is envisaged to adequately meet the existing demand for electricity.
    Keywords: ARIMA; electricity consumption; electricity demand; energy; forecasting; Zimbabwe
    JEL: P28 P48 Q41 Q43 Q47
    Date: 2019–11–05

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.