nep-for New Economics Papers
on Forecasting
Issue of 2020‒07‒27
eighteen papers chosen by
Rob J Hyndman
Monash University

  1. Forecasting volatility with a stacked model based on a hybridized Artificial Neural Network By E. Ramos-P\'erez; P. J. Alonso-Gonz\'alez; J. J. N\'u\~nez-Vel\'azquez
  2. Deus ex Machina? A Framework for Macro Forecasting with Machine Learning By Marijn A. Bolhuis; Brett Rayner
  3. Where Should We Go? Internet Searches and Tourist Arrivals By Serhan Cevik
  4. From Fixed-event to Fixed-horizon Density Forecasts: Obtaining Measures of Multi-horizon Uncertainty from Survey Density Forecasts By Ganics, Gergely; Rossi, Barbara; Sekhposyan, Tatevik
  5. Real-time Probabilistic Nowcasts of UK Quarterly GDP Growth using a Mixed-Frequency Bottom-up Approach By Ana Beatriz Galvao; Marta Lopresto
  6. Forecasting the New England States’ Tax Revenues in the Time of the COVID-19 Pandemic By Bo Zhao
  7. Media Attention vs. Sentiment as Drivers of Conditional Volatility Predictions: An Application to Brexit By Massimo Guidolin; Manuela Pedio
  8. The More the Merrier? A Machine Learning Algorithm for Optimal Pooling of Panel Data By Marijn A. Bolhuis; Brett Rayner
  9. Large Time-Varying Volatility Models for Electricity Prices By Angelica Gianfreda; Francesco Ravazzolo; Luca Rossini
  10. Forecasting Urban Residential Stock Turnover Dynamics using System Dynamics and Bayesian Model Averaging By Zhou, W.; O’Neill, E.; Moncaster, A.; Reiner D.; Guthrie, P.
  11. Predicting the VIX and the Volatility Risk Premium: The Role of Short-run Funding Spreads Volatility Factors By Elena Andreou; Eric Ghysels
  12. Applying Dynamic Training-Subset Selection Methods Using Genetic Programming for Forecasting Implied Volatility By Sana Ben Hamida; Wafa Abdelmalek; Fathi Abid
  13. Real-Time Forecasting with a (Standard) Mixed-Frequency VAR During a Pandemic By Frank Schorfheide; Dongho Song
  14. The CARES ACT and Poverty in the COVID-19 Crisis: Promises and Pitfalls of the Recovery Rebates and Expanded Unemployment Benefits By Zachary Parolin; Megan Curran; Christoper Wimer
  15. Inference in Bayesian Additive Vector Autoregressive Tree Models By Florian Huber; Luca Rossini
  16. Nowcasting German GDP By Andreini, Paolo; Charlotte Senftleben-König, Charlotte; Hasenzagl, Thomas; Reichlin, Lucrezia; Strohsal, Till
  17. Forecasting Estimates of Poverty During the COVID-19 Crisis By Zachary Parolin; Christoper Wimer
  18. Theoretical approaches to forecasting regional macro-indicators By Gorshkova, Taisiya (Горшкова, Таисия); Turuntseva, Marina (Турунцева, Марина)

  1. By: E. Ramos-P\'erez; P. J. Alonso-Gonz\'alez; J. J. N\'u\~nez-Vel\'azquez
    Abstract: An appropriate calibration and forecasting of volatility and market risk are some of the main challenges faced by companies that have to manage the uncertainty inherent to their investments or funding operations such as banks, pension funds or insurance companies. This has become even more evident after the 2007-2008 Financial Crisis, when the forecasting models assessing the market risk and volatility failed. Since then, a significant number of theoretical developments and methodologies have appeared to improve the accuracy of the volatility forecasts and market risk assessments. Following this line of thinking, this paper introduces a model based on using a set of Machine Learning techniques, such as Gradient Descent Boosting, Random Forest, Support Vector Machine and Artificial Neural Network, where those algorithms are stacked to predict S&P500 volatility. The results suggest that our construction outperforms other habitual models on the ability to forecast the level of volatility, leading to a more accurate assessment of the market risk.
    Date: 2020–06
  2. By: Marijn A. Bolhuis; Brett Rayner
    Abstract: We develop a framework to nowcast (and forecast) economic variables with machine learning techniques. We explain how machine learning methods can address common shortcomings of traditional OLS-based models and use several machine learning models to predict real output growth with lower forecast errors than traditional models. By combining multiple machine learning models into ensembles, we lower forecast errors even further. We also identify measures of variable importance to help improve the transparency of machine learning-based forecasts. Applying the framework to Turkey reduces forecast errors by at least 30 percent relative to traditional models. The framework also better predicts economic volatility, suggesting that machine learning techniques could be an important part of the macro forecasting toolkit of many countries.
    Keywords: Production growth;Capacity utilization;Economic growth;Stock markets;Emerging markets;Forecasts,Nowcasting,Machine learning,GDP growth,Cross-validation,Random Forest,Ensemble,Turkey.,WP,forecast error,factor model,predictor,forecast,OLS
    Date: 2020–02–28
  3. By: Serhan Cevik
    Abstract: The widespread availability of internet search data is a new source of high-frequency information that can potentially improve the precision of macroeconomic forecasting, especially in areas with data constraints. This paper investigates whether travel-related online search queries enhance accuracy in the forecasting of tourist arrivals to The Bahamas from the U.S. The results indicate that the forecast model incorporating internet search data provides additional information about tourist flows over a univariate approach using the traditional autoregressive integrated moving average (ARIMA) model and multivariate models with macroeconomic indicators. The Google Trends-augmented model improves predictability of tourist arrivals by about 30 percent compared to the benchmark ARIMA model and more than 20 percent compared to the model extended only with income and relative prices.
    Keywords: Real effective exchange rates;Economic growth;Economic forecasting;Real exchange rates;Personal income;Forecasting,tourist arrivals,Google Trends,time-series models,WP,ARIMA,tourist arrival,autoregressive,forecast model,time-series
    Date: 2020–01–31
  4. By: Ganics, Gergely; Rossi, Barbara; Sekhposyan, Tatevik
    Abstract: Surveys of professional forecasters produce precise and timely point forecasts for key macroeconomic variables. However, the accompanying density forecasts are not as widely utilized, and there is no consensus about their quality. This is partly because such surveys are often conducted for "fixed events". For example, in each quarter, panelists are asked to forecast output growth and inflation for the current calendar year and the next, implying that the forecast horizon changes with each survey round. The fixed-event nature limits the usefulness of survey density predictions for policymakers and market participants, who often wish to characterize uncertainty a fixed number of periods ahead ("fixed-horizon"). Is it possible to obtain fixed-horizon density forecasts using the available fixed-event ones? We propose a density combination approach that weights fixed-event density forecasts according to a uniformity of the probability integral transform criterion, aiming at obtaining a correctly calibrated fixed-horizon density forecast. Using data from the US Survey of Professional Forecasters, we show that our combination method produces competitive density forecasts relative to widely used alternatives based on historical forecast errors or Bayesian VARs. Thus, our proposed fixed-horizon predictive densities are a new and useful tool for researchers and policymakers.
    Date: 2020–01
  5. By: Ana Beatriz Galvao; Marta Lopresto
    Abstract: We propose a nowcasting system to obtain real-time predictive intervals for the first release of UK quarterly GDP growth that can be implemented in a menu-driven econometric software. We design a bottom-up approach: forecasts for GDP components (from the output and the expenditure approaches) are inputs into the computation of probabilistic forecasts for GDP growth. For each GDP component considered, mixed-data sampling regressions are applied to extract predictive content from monthly and quarterly indicators. We find that predictions from the nowcasting system are accurate, in particular when nowcasts are computed using monthly indicators 30 days before the GDP release. The system is also able to provide well-calibrated predictive intervals.
    Keywords: nowcasting, GDP growth, mixed frequency regression, forecast combination, probabilistic forecasts
    JEL: C53 E32
    Date: 2020–05
  6. By: Bo Zhao
    Abstract: State governments across the United States face the prospect of sharply declining tax revenues due to the COVID-19 pandemic. They need reliable and up-to-date revenue forecasts to make financially sound policy decisions during this public health and economic crisis. This paper proposes an objective, transparent, simple, and efficient method to forecast state tax revenues in this time of the COVID-19 pandemic. The model is based on only two input factors: the state unemployment rate and an empirically determined time trend. The predictions from the model closely track the actual values of tax revenues for the New England states over the past 25 years. Using this method, this paper forecasts state tax revenues for fiscal year 2021 and suggests large decreases in the New England states. The paper discusses policy options to address the expected declines in revenues and highlights the urgent need for more federal grants without tight strings attached.
    Keywords: COVID-19; revenue forecasting; state tax revenue; NEPPC
    JEL: C22 C53 H71
    Date: 2020–07–09
  7. By: Massimo Guidolin; Manuela Pedio
    Abstract: Using data on international, on-line media coverage and tone of the Brexit referendum, we test whether it is media coverage or tone to provide the largest forecasting performance improvements in the prediction of the conditional variance of weekly FTSE 100 stock returns. We find that versions of standard symmetric and asymmetric Generalized Autoregressive Conditional Heteroskedasticity (GARCH) models augmented to include media coverage and especially media tone scores outperform traditional GARCH models both in- and out-of-sample.
    Keywords: Attention, Sentiment, Text Mining, Forecasting, Conditional Variance, GARCH model, Brexit
    JEL: C53 C58 G17
    Date: 2020
  8. By: Marijn A. Bolhuis; Brett Rayner
    Abstract: We leverage insights from machine learning to optimize the tradeoff between bias and variance when estimating economic models using pooled datasets. Specifically, we develop a simple algorithm that estimates the similarity of economic structures across countries and selects the optimal pool of countries to maximize out-of-sample prediction accuracy of a model. We apply the new alogrithm by nowcasting output growth with a panel of 102 countries and are able to significantly improve forecast accuracy relative to alternative pools. The algortihm improves nowcast performance for advanced economies, as well as emerging market and developing economies, suggesting that machine learning techniques using pooled data could be an important macro tool for many countries.
    Keywords: Economic models;Production growth;Developing countries;Emerging markets;Data analysis;Machine learning,GDP growth,forecasts,panel data,pooling.,WP,forecast error,DGP,forecast,economic structure,output growth
    Date: 2020–02–28
  9. By: Angelica Gianfreda; Francesco Ravazzolo; Luca Rossini
    Abstract: We study the importance of time-varying volatility in modelling hourly electricity prices when fundamental drivers are included in the estimation. This allows us to contribute to the literature of large Bayesian VARs by using well-known time series models in a huge dimension for the matrix of coefficients. Based on novel Bayesian techniques, we exploit the importance of both Gaussian and non-Gaussian error terms in stochastic volatility. We find that by using regressors as fuels prices, forecasted demand and forecasted renewable energy is essential in order to properly capture the volatility of these prices. Moreover, we show that the time-varying volatility models outperform the constant volatility models in both the in-sample model- fit and the out-of-sample forecasting performance.
    Keywords: Electricity, Hourly Prices, Renewable Energy Sources, Non-Gaussian, Stochastic-Volatility, Forecasting
    Date: 2020–07
  10. By: Zhou, W.; O’Neill, E.; Moncaster, A.; Reiner D.; Guthrie, P.
    Abstract: Knowing the size of the building stock is perhaps the most basic determinant in assessing energy use in buildings. However, official statistics on urban residential stock for many countries are piecemeal at best. Previous studies estimating stock size and energy use make various debateable methodological assumptions and only produce deterministic results. We present a Bayesian approach to characterise stock turnover dynamics and estimate stock size uncertainties, applied here to the case of China. Firstly, a probabilistic dynamic building stock turnover model is developed to describe the building aging and demolition process, governed by a hazard function specified by a parametric survival model. Secondly, using five candidate parametric survival models, the building stock turnover model is simulated through Markov Chain Monte Carlo to obtain posterior distributions of model-specific parameters, estimate marginal likelihood, and make predictions of stock size. Thirdly, Bayesian Model Averaging is applied to create a model ensemble that combines model-specific posterior predictive distributions of the recent historical stock evolution pathway in proportion to posterior model probabilities. Finally, the Bayesian Model Averaging model ensemble is extended to forecast future trajectories of residential stock development through 2100. The modelling results suggest that the total stock in China will peak around 2065, at between 42.4 and 50.1 billion m 2 . This Bayesian modelling framework produces probability distributions of annual total stock, age-specific substocks, annual new buildings and annual demolition rates. This can support future analysis of policy trade-offs across embodied-versus-operational energy consumption, in the context of sector-wide decarbonisation.
    Keywords: building stock, lifetime distribution, System Dynamics, Bayesian Model Averaging, Markov Chain Monte Carlo, embodied energy, operational energy, China
    JEL: C11 O18 R21
    Date: 2020–06–16
  11. By: Elena Andreou; Eric Ghysels
    Abstract: This paper presents an innovative approach to extract Volatility Factors which predict the VIX, the S&P500 Realized Volatility (RV) and the Variance Risk Premium (VRP). The approach is innovative along two different dimensions, namely: (1) we extract Volatility Factors from panels of filtered volatilities - in particular large panels of univariate ARCH-type models and propose methods to estimate common Volatility Factors in the presence of estimation error and (2) we price equity volatility risk using factors which go beyond the equity class namely Volatility Factors extracted from panels of volatilities of short-run funding spreads. The role of these Volatility Factors is compared with the corresponding factors extracted from the panels of the above spreads as well as related factors proposed in the literature. Our monthly short-run funding spreads Volatility Factors provide both in- and out-of-sample predictive gains for forecasting the monthly VIX, RV as well as the equity premium, while the corresponding daily volatility factors via Mixed Data Sampling (MIDAS) models provide further improvements.
    Keywords: Factor asset pricing models; Volatility Factors; ARCH filters
    JEL: C2 C5 G1
    Date: 2020–03
  12. By: Sana Ben Hamida; Wafa Abdelmalek; Fathi Abid
    Abstract: Volatility is a key variable in option pricing, trading and hedging strategies. The purpose of this paper is to improve the accuracy of forecasting implied volatility using an extension of genetic programming (GP) by means of dynamic training-subset selection methods. These methods manipulate the training data in order to improve the out of sample patterns fitting. When applied with the static subset selection method using a single training data sample, GP could generate forecasting models which are not adapted to some out of sample fitness cases. In order to improve the predictive accuracy of generated GP patterns, dynamic subset selection methods are introduced to the GP algorithm allowing a regular change of the training sample during evolution. Four dynamic training-subset selection methods are proposed based on random, sequential or adaptive subset selection. The latest approach uses an adaptive subset weight measuring the sample difficulty according to the fitness cases errors. Using real data from SP500 index options, these techniques are compared to the static subset selection method. Based on MSE total and percentage of non fitted observations, results show that the dynamic approach improves the forecasting performance of the generated GP models, specially those obtained from the adaptive random training subset selection method applied to the whole set of training samples.
    Date: 2020–06
  13. By: Frank Schorfheide; Dongho Song
    Abstract: In this paper we resuscitate the mixed-frequency vector autoregression (MF-VAR) developed in Schorfheide and Song (2015) to generate real-time macroeconomic forecasts for the U.S. during the COVID-19 pandemic. The model combines eleven time series observed at two frequencies: quarterly and monthly. We deliberately do not modify the model specification in view of the recession induced by the COVID-19 outbreak. We find that forecasts based on a pre-crisis estimate of the VAR using data up until the end of 2019 appear to be more stable and reasonable than forecasts based on a sequence of recursive estimates that include the most recent observations. Overall, the MF-VAR outlook is quite pessimistic. The estimated MF-VAR implies that level variables are highly persistent, which means that the COVID-19 shock generates a long-lasting reduction in real activity. Regularly updated forecasts are available at
    Keywords: Bayesian inference; COVID-19; Macroeconomic Forecasting; Minnesota Prior
    JEL: C11 C32 C53
    Date: 2020–07–08
  14. By: Zachary Parolin (Columbia University); Megan Curran (Columbia University); Christoper Wimer (Columbia University)
    Abstract: In response to rapidly rising unemployment rates, Congress passed the Coronavirus Aid, Relief, and Economic Security (CARES) Act, which included nearly $500 billion in direct income transfers for families across the country. In this brief, we apply new forecasting methods to project the effect of the CARES Act’s income transfers on poverty rates using the Supplemental Poverty Measure (SPM) framework. In response to rapidly rising unemployment rates, Congress passed the Coronavirus Aid, Relief, and Economic Security (CARES) Act, which included nearly $500 billion in direct income transfers for families across the country. In this brief, we apply new forecasting methods to project the effect of the CARES Act’s income transfers on poverty rates using the Supplemental Poverty Measure (SPM) framework.
    Keywords: poverty, COVID-19, social policy, SPM
    Date: 2020–06
  15. By: Florian Huber; Luca Rossini
    Abstract: Vector autoregressive (VAR) models assume linearity between the endogenous variables and their lags. This linearity assumption might be overly restrictive and could have a deleterious impact on forecasting accuracy. As a solution, we propose combining VAR with Bayesian additive regression tree (BART) models. The resulting Bayesian additive vector autoregressive tree (BAVART) model is capable of capturing arbitrary non-linear relations between the endogenous variables and the covariates without much input from the researcher. Since controlling for heteroscedasticity is key for producing precise density forecasts, our model allows for stochastic volatility in the errors. Using synthetic and real data, we demonstrate the advantages of our methods. For Eurozone data, we show that our nonparametric approach improves upon commonly used forecasting models and that it produces impulse responses to an uncertainty shock that are consistent with established findings in the literature.
    Date: 2020–06
  16. By: Andreini, Paolo; Charlotte Senftleben-König, Charlotte; Hasenzagl, Thomas; Reichlin, Lucrezia; Strohsal, Till
    Abstract: This paper develops a nowcasting model for the German economy. The model outperforms a number of alternatives and produces forecasts not only for GDP but also for other key variables. We show that the inclusion of foreign variables improves the model's performance, while financial variables do not. Additionally, a comprehensive model averaging exercise reveals that factor extraction in a single model delivers slightly better results than averaging across models. Finally, we estimate a "news" index for the German economy constructed as a weighted average of the nowcast errors related to each variable included in the model.
    Date: 2020–01
  17. By: Zachary Parolin (Columbia University); Christoper Wimer (Columbia University)
    Abstract: To what extent will the COVID-19 pandemic increase levels of poverty in the United States? Official estimates of poverty in the United States are presented on an annual basis and with a considerable lag. In this brief, we apply a novel method for forecasting poverty rates in the United States using the Supplemental Poverty Measure (SPM) framework with a goal of providing projections of poverty rates throughout the COVID-19 crisis.
    Keywords: poverty, COVID-19, social policy, SPM
    Date: 2020–04
  18. By: Gorshkova, Taisiya (Горшкова, Таисия) (The Russian Presidential Academy of National Economy and Public Administration); Turuntseva, Marina (Турунцева, Марина) (The Russian Presidential Academy of National Economy and Public Administration)
    Abstract: The work is devoted to the analysis of existing theoretical models for forecasting regional macro-indicators and the study of the possibility of forecasting Russian data based on the selected theoretical approaches. A comparative analysis of theoretical approaches to modeling regional data is carried out. The approaches considered include diversification indices based on various economic theory, analysis of the possibility of using a composite welfare index as a proxy variable for the economic situation, and application of dynamic and non-linear models to regional data. The study was conducted on data on a set of macro indicators (CPI, GRP per capita, unemployment rate, average per capita income, etc.) in all regions of Russia, as well as for regions united by federal districts and by clusters determined on the basis of theoretical approaches. On the Russian data, various diversification indices were analyzed, and ensembles of neural networks and vector autoregressions were constructed, including taking into account the spatial dependence between the indicators.
    Date: 2020–03

This nep-for issue is ©2020 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.