nep-for New Economics Papers
on Forecasting
Issue of 2021‒08‒30
ten papers chosen by
Rob J Hyndman
Monash University

  1. On the evaluation of hierarchical forecasts By George Athanasopoulos; Nikolaos Kourentzes
  2. Forecasting in the Absence of Precedent By Paul Ho
  3. Time Series Forecasting Using a Mixture of Stationary and Nonstationary Predictors By Sium Bodha Hannadige; Jiti Gao; Mervyn J Silvapulle; Param Silvapulle
  4. Dimensionality Reduction and State Space Systems: Forecasting the US Treasury Yields Using Frequentist and Bayesian VARs By Sudiksha Joshi
  5. Forecasting student enrollment using time series models and recurrent neural networks By Parvez, Rezwanul; Ali Meerza, Syed Imran; Hasan Khan Chowdhury, Nazea
  6. Nowcasting Colombian Economic Activity: DFM and Factor-MIDAS approaches By Franky Juliano Galeano-Ramírez; Nicolás Martínez-Cortés; Carlos D. Rojas-Martínez
  7. Loss-Based Variational Bayes Prediction By David T. Frazier; Ruben Loaiza-Maya; Gael M. Martin; Bonsoo Koo
  8. Forecasting Urban Residential Stock Turnover Dynamics using System Dynamics and Bayesian Model Averaging By Wei Zhou; Eoghan O’Neill; Alice Moncaster; David Reiner; Peter Guthrie
  9. Estimating the costs of energy transition scenarios using probabilistic forecasting methods By Farmer, J. Doyne; Way, Rupert; Mealy, Penny
  10. The Economics of Walking About and Predicting Unemployment By David G. Blanchflower; Alex Bryson

  1. By: George Athanasopoulos; Nikolaos Kourentzes
    Abstract: The aim of this paper is to provide a thinking road-map and a practical guide to researchers and practitioners working on hierarchical forecasting problems. Evaluating the performance of hierarchical forecasts comes with new challenges stemming from both the structure of the hierarchy and the application context. We discuss several relevant dimensions for researchers and analysts: the scale and units of the time series, the issue of sparsity, the forecast horizon, the importance of multiple evaluation windows and the multiple objective decision context. We conclude with a series of practical recommendations.
    Keywords: Aggregation, coherence, hierarchical time series, reconciliation
    JEL: C18 C53 C55
    Date: 2021
  2. By: Paul Ho
    Abstract: We survey approaches to macroeconomic forecasting during the COVID-19 pandemic. Due to the unprecedented nature of the episode, there was greater dependence on information outside the econometric model, captured through either adjustments to the model or additional data. The transparency and flexibility of assumptions were especially important for interpreting real-time forecasts and updating forecasts as new data were observed. With data available at the time of writing, we show how various assumptions were violated and how these systematically biased forecasts.
    Keywords: Macroeconomic Forecasting; COVID-1
    Date: 2021–06–16
  3. By: Sium Bodha Hannadige; Jiti Gao; Mervyn J Silvapulle; Param Silvapulle
    Abstract: We develop a method for constructing prediction intervals for a nonstationary variable, such as GDP. The method uses a factor augmented regression [FAR] model. The predictors in the model includes a small number of factors generated to extract most of the information in a set of panel data on a large number of macroeconomic variables considered to be potential predictors. The novelty of this paper is that it provides a method and justification for a mixture of stationary and nonstationary factors as predictors in the FAR model; we refer to this as mixture-FAR method. This method is important because typically such a large set of panel data, for example the FRED-MD, is likely to contain a mixture of stationary and nonstationary variables. In our simulation study, we observed that the proposed mixture-FAR method performed better than its competitor that requires all the predictors to be nonstationary; the MSE of prediction was at least 33% lower for mixture-FAR. Using the data in FRED-QD for the US, we evaluated the aforementioned methods for forecasting the nonstationary variables, GDP and Industrial Production. We observed that the mixture-FAR method performed better than its competitors.
    Keywords: gross domestic product, high dimensional data, industrial production, macroeconomic forecasting, panel data
    JEL: C22 C33 C38 C53
    Date: 2021
  4. By: Sudiksha Joshi
    Abstract: Using a state-space system, I forecasted the US Treasury yields by employing frequentist and Bayesian methods after first decomposing the yields of varying maturities into its unobserved term structure factors. Then, I exploited the structure of the state-space model to forecast the Treasury yields and compared the forecast performance of each model using mean squared forecast error. Among the frequentist methods, I applied the two-step Diebold-Li, two-step principal components, and one-step Kalman filter approaches. Likewise, I imposed the five different priors in Bayesian VARs: Diffuse, Minnesota, natural conjugate, the independent normal inverse: Wishart, and the stochastic search variable selection priors. After forecasting the Treasury yields for 9 different forecast horizons, I found that the BVAR with Minnesota prior generally minimizes the loss function. I augmented the above BVARs by including macroeconomic variables and constructed impulse response functions with a recursive ordering identification scheme. Finally, I fitted a sign-restricted BVAR with dummy observations.
    Date: 2021–08
  5. By: Parvez, Rezwanul; Ali Meerza, Syed Imran; Hasan Khan Chowdhury, Nazea
    Keywords: Teaching/Communication/Extension/Profession, Community/Rural/Urban Development, Institutional and Behavioral Economics
    Date: 2021–08
  6. By: Franky Juliano Galeano-Ramírez; Nicolás Martínez-Cortés; Carlos D. Rojas-Martínez
    Abstract: Economic policy decision-making requires constantly assessing the state of economic activity. However, this is not an easy task: official figures have significant lags, and the timely information is usually partial and has diffierent frequencies. This paper applies two types of short-term forecasting methodologies (Factor-MIDAS and DFM) for Colombian economic activity involving information with mixed frequencies. We present a heuristic process to select relevant variables, and we evaluate the proposed models' fits by comparing them with traditional forecasting methodologies. Overall, DFM and Factor-MIDAS forecasts are better than those generated by conventional methodologies, especially as the flow of information increases. In times of COVID-19, the model with the best relative fit was the DFM. **** RESUMEN: La toma de decisiones de política económica requiere evaluar constantemente el estado de la actividad económica. Sin embargo, ello no es una tarea fácil: las cifras oficiales tienen rezagos importantes y la información más oportuna suele ser parcial y tener frecuencias dispares. Este artículo aplica dos tipos de metodologías de pronóstico de corto plazo (Factor-MIDAS y DFM) para la actividad económica colombiana involucrando información con frecuencias mixtas. Se propone un proceso heurístico para la selección de variables relevantes y se evalúa el ajuste de los modelos comparándolo respecto a metodologías usuales de proyección. En general, los pronósticos de los modelos Factor-MIDAS y del DFM superan los generados por metodologías tradicionales, con resultados más precisos en la medida que aumenta el flujo de información. En tiempos del COVID-19, el modelo con el mejor ajuste relativo fue el DFM.
    Keywords: Colombian economic activity, nowcast, forecast, mixed frequency factor models, actividad económica colombiana, nowcast, pronóstico, modelos de frecuencia mixta con factores
    JEL: C53 E27 E52
    Date: 2021–08
  7. By: David T. Frazier; Ruben Loaiza-Maya; Gael M. Martin; Bonsoo Koo
    Abstract: We propose a new method for Bayesian prediction that caters for models with a large number of parameters and is robust to model misspecification. Given a class of high-dimensional (but parametric) predictive models, this new approach constructs a posterior predictive using a variational approximation to a loss-based, or Gibbs, posterior that is directly focused on predictive accuracy. The theoretical behavior of the new prediction approach is analyzed and a form of optimality demonstrated. Applications to both simulated and empirical data using high-dimensional Bayesian neural network and autoregressive mixture models demonstrate that the approach provides more accurate results than various alternatives, including misspecified likelihood-based predictions
    Keywords: loss-based Bayesian forecasting, variational inference, Gibbs posteriors, proper scoring rules, Bayesian neural networks, M4 forecasting competition
    JEL: C11 C53 C58
    Date: 2021
  8. By: Wei Zhou (Department of Engineering, University of Cambridge); Eoghan O’Neill (Faculty of Economics, University of Cambridge); Alice Moncaster (Department of Engineering, University of Cambridge); David Reiner (EPRG, CJBS, University of Cambridge); Peter Guthrie (Department of Engineering, University of Cambridge)
    Keywords: building stock, lifetime distribution, System Dynamics, Bayesian Model Averaging, Markov Chain Monte Carlo, embodied energy, operational energy, China
    JEL: C11 O18 R21
    Date: 2020–06
  9. By: Farmer, J. Doyne; Way, Rupert; Mealy, Penny
    Abstract: We evaluate the cost of four different scenarios for the global energy system from 2020 to 2070 using an empirically validated technology forecasting method based on an expansive historical dataset. A no-transition scenario that maintains the current energy mix provides a benchmark. Under a rapid transition scenario, solar photovoltaics and wind are quickly deployed using batteries for short-term storage. Hydrogen-based fuels are used for long-term storage and non-electrifiable applications. Energy prices become lower than historical averages after 2030 and considerably lower after 2050. This yields an expected net present saving at any sensible discount rate; at 4% for example, we predict savings of $5.6 trillion. In contrast, a slower transition is more expensive, while a nuclear scenario is substantially more expensive.
    Date: 2020–12
  10. By: David G. Blanchflower (Bruce V. Rauner ’78 Professor of Economics, Dartmouth College, Hanover, NH 03755-3514. Adam Smith School of Business, University of Glasgow and NBER); Alex Bryson (Professor of Quantitative Social Science, UCL Social Research Institute, University College London, 20 Bedford Way, London WC1H 0AL)
    Abstract: Unemployment is notoriously difficult to predict. In previous studies, once country fixed effects are added to panel estimates, few variables predict changes in unemployment rates. Using panel data for 29 European countries - Austria; Belgium; Bulgaria; Croatia; Cyprus; Czechia; Denmark; Estonia; Finland; France; Germany; Greece; Hungary; Ireland; Italy; Latvia; Lithuania; Luxembourg; Malta; Netherlands; Poland; Portugal; Romania; Slovakia; Slovenia; Spain; Sweden; Turkey and the UK - over 439 months between January 1985 and July 2021 in an unbalanced country*month panel of just over 10000 observations, we predict changes in the unemployment rate 12 months in advance based on individuals’ fears of unemployment, their perceptions of the economic situation and their own household financial situation. Fear of unemployment predicts subsequent changes in unemployment 12 months later in the presence of country fixed effects and lagged unemployment. Individuals’ perceptions of the economic situation in the country and their own household finances also predict unemployment 12 months later. Business sentiment (industry fear of unemployment) is also predictive of unemployment 12 months later. The findings underscore the importance of the “economics of walking about”. The implication is that these social survey data are informative in predicting economic downturns and should be used more extensively in forecasting. We also generate a 29 country-level annual panel on life satisfaction from 1985-2020 from the World Database of Happiness and show that the consumer level fear of unemployment variable lowers wellbeing over and above the negative impact of the unemployment rate itself. Qualitative survey metrics were able to predict the Great Recession and the economic slowdown in Europe just prior to the COVID pandemic.
    Keywords: unemployment, fear, sentiment, social attitudes, life satisfaction, recession, COVID
    JEL: J60 J64 J68
    Date: 2021–08–01

This nep-for issue is ©2021 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.