nep-for New Economics Papers
on Forecasting
Issue of 2015‒06‒20
fifteen papers chosen by
Rob J Hyndman
Monash University

  1. STR: A Seasonal-Trend Decomposition Procedure Based on Regression By Alexander Dokumentov; Rob J. Hyndman
  2. Selection of an Estimation Window in the Presence of Data Revisions and Recent Structural Breaks By Jari Hännäkäinen
  3. Forecasting daily political opinion polls using the fractionally cointegrated VAR model By Morten Ørregaard Nielsen; Sergei S. Shibaev
  4. Probabilistic time series forecasting with boosted additive models: an application to smart meter data By Souhaib Ben Taieb; Raphael Huser; Rob J. Hyndman; Marc G. Genton
  5. Short term inflation forecasting: the M.E.T.A. approach By Giacomo Sbrana; Andrea Silvestrini; Fabrizio Venditti
  6. The mortgage spread as a predictor of real-time economic activity By Jari Hännäkäinen
  7. Forecasting VARs, Model Selection, and Shrinkage By Kascha, Christian; Trenkler, Carsten
  8. Forecaster overconfidence and market survey performance By Deaves, Richard; Lei, Jin; Schroeder, Michael
  9. Zero Lower Bound and Indicator Properties of Interest Rate Spreads By Jari Hännäkäinen
  10. Multi-step forecasting in the presence of breaks By Jari Hännäkäinen
  11. Zero lower bound, unconventional monetary policy and indicator properties of interest rate spreads By Jari Hännäkäinen
  12. Nowcasting and short-term forecasting of Russian GDP with a dynamic factor model By Porshakov , Alexey; Deryugina , Elena; Ponomarenko , Alexey; Sinyakov , Andrey
  13. Bootstrap inference in regressions with estimated factors and serial correlation By Antoine Djogbenou; Sílvia Gonçalves; Benoit Perron
  14. The Taylor Rule, Wealth Effects and the Exchange Rate By Rudan Wang; Bruce Morley; Javier Ordóñez
  15. The scale of predictability By Federico M. Bandi; Benoit Perron; Andrea Tamoni; Claudio Tebaldi

  1. By: Alexander Dokumentov; Rob J. Hyndman
    Abstract: We propose new generic methods for decomposing seasonal data: STR (a Seasonal-Trend decomposition procedure based on Regression) and Robust STR. In some ways, STR is similar to Ridge Regression and Robust STR can be related to LASSO. Our new methods are much more general than any alternative time series decomposition methods. They allow for multiple seasonal and cyclic components, and multiple linear regressors with constant, flexible, seasonal and cyclic influence. Seasonal patterns (for both seasonal components and seasonal regressors) can be fractional and flexible over time; moreover they can be either strictly periodic or have a more complex topology. We also provide confidence intervals for the estimated components, and discuss how STR can be used for forecasting.
    Keywords: time series decomposition, seasonal data, Tikhonov regularisation, ridge regression, LASSO, STL, TBATS, X-12-ARIMA, BSM
    JEL: C10 C14 C22
    Date: 2015
  2. By: Jari Hännäkäinen (School of Management, University of Tampere)
    Abstract: In this paper, we analyze the forecasting performance of a set of widely used window selection methods in the presence of data revisions and recent structural breaks. Our Monte Carlo and empirical results show that the expanding window estimator often yields the most accurate forecasts after a recent break. It performs well regardless of whether the revisions are news or noise, or whether we forecast first-release or final values. We find that the differences in the forecasting accuracy are large in practice, especially when we forecast GDP deflator growth after the break of the early 1980s.
    Keywords: Recent structural break, choice of estimation window, forecasting, real-time data
    JEL: C22 C53 C82
    Date: 2013–12
  3. By: Morten Ørregaard Nielsen (Queen's University and CREATES); Sergei S. Shibaev (Queen's University)
    Abstract: We examine forecasting performance of the recent fractionally cointegrated vector autoregressive (FCVAR) model. The model is applied to daily polling data of political support in the United Kingdom for 2010-2015. We compare with popular competing models and at various forecast horizons. Our findings show that the precision of forecasts generated by the FCVAR model is better than all multivariate and univariate models in the portfolio, and the four variants of the FCVAR model considered are generally ranked as the top four models in terms of forecast accuracy. Furthermore, the FCVAR model significantly outperforms the standard cointegrated VAR (CVAR) model at all forecast horizons and the relative forecast improvement is highest at longer forecast horizons, where the root mean squared forecast error of the FCVAR model is up to 20% lower than that of the CVAR benchmark model. In an empirical application to the prediction of vote shares in the 2015 UK general election, forecasts generated by the FCVAR model leading into the election appear to provide a more informative assessment of the current state of public opinion on electoral support than that suggested by the hung government prediction of the opinion poll. Specifically, the FCVAR model projects the correct direction for the realized vote shares in the election for both the Conservative and Labour parties.
    Keywords: forecasting, fractional cointegration, opinion poll data, vector autoregressive model
    Date: 2015–06
  4. By: Souhaib Ben Taieb; Raphael Huser; Rob J. Hyndman; Marc G. Genton
    Abstract: A large body of the forecasting literature so far has been focused on forecasting the conditional mean of future observations. However, there is an increasing need for generating the entire conditional distribution of future observations in order to effectively quantify the uncertainty in time series data. We present two different methods for probabilistic time series forecasting that allow the inclusion of a possibly large set of exogenous variables. One method is based on forecasting both the conditional mean and variance of the future distribution using a traditional regression approach. The other directly computes multiple quantiles of the future distribution using quantile regression. We propose an implementation for the two methods based on boosted additive models, which enjoy many useful properties including accuracy, flexibility, interpretability and automatic variable selection. We conduct extensive experiments using electricity smart meter data, on both aggregated and disaggregated scales, to compare the two forecasting methods for the challenging problem of forecasting the distribution of future electricity consumption. The empirical results demonstrate that the mean and variance forecasting provides better forecasts for aggregated demand, while the flexibility of the quantile regression approach is more suitable for disaggregated demand. These results are particularly useful since more energy data will become available at the disaggregated level in the future.
    Keywords: Additive models, boosting, density forecasting, energy forecasting, probabilistic forecasting
    JEL: Q47 C14 C22
    Date: 2015
  5. By: Giacomo Sbrana (NEOMA Business School); Andrea Silvestrini (Bank of Italy); Fabrizio Venditti (Bank of Italy)
    Abstract: Forecasting inflation is an important and challenging task. In this paper we assume that the core inflation components evolve as a multivariate local level process. This model, which is theoretically attractive for modelling inflation dynamics, has been used only to a limited extent to date owing to computational complications with the conventional multivariate maximum likelihood estimator, especially when the system is large. We propose the use of a method called “Moments Estimation Through Aggregation” (M.E.T.A.), which reduces computational costs significantly and delivers prompt and accurate parameter estimates, as we show in a Monte Carlo exercise. In an application to euro-area inflation we find that our forecasts compare well with those generated by alternative univariate constant and time-varying parameter models as well as with those of professional forecasters and vector autoregressions.
    Keywords: inflation, forecasting, aggregation, state space models
    JEL: C32 C53 E31 E37
    Date: 2015–06
  6. By: Jari Hännäkäinen (School of Management, University of Tampere)
    Abstract: We analyze the predictive content of the mortgage spread for U.S. economic activity. We find that the spread contains predictive power for real GDP and industrial production. Furthermore, it outperforms the term spread and Gilchrist–Zakrajsek spread in a real-time forecasting exercise. However, the predictive ability of the mortgage spread varies over time.
    Keywords: mortgage spread, forecasting, real-time data
    JEL: C53 E37 E44
    Date: 2014–09
  7. By: Kascha, Christian; Trenkler, Carsten
    Abstract: This paper provides an empirical comparison of various selection and penalized regression approaches for forecasting with vector autoregressive systems. In particular, we investigate the effect of the system size as well as the effect of various prior specification choices on the relative and overall forecasting performance of the methods. The data set is a typical macroeconomic quarterly data set for the US. We find that these specification choices are crucial for most methods. Conditional on certain choices, the variation across different approaches is relatively small. There are only a few methods which are not competitive under any scenario. For single series, we find that increasing the system size can be helpful - depending on the employed shrinkage method.
    Keywords: VAR Models , Forecasting , Model Selection , Shrinkage
    JEL: C32 C53 E47
    Date: 2015
  8. By: Deaves, Richard; Lei, Jin; Schroeder, Michael
    Abstract: We document using the ZEW panel of German stock market forecasters that weak forecasters tend to be overconfident in the sense that they provide extreme forecasts and their confidence intervals are less likely to contain eventual realizations. Moderate filters based on forecast accuracy over short rolling windows are somewhat successful in improving predictability. While poor performance can be due to various factors, a filter based on a prior tendency to provide extreme forecasts also improves predictability.
    Keywords: Overconfidence,Forecasting Performance,Stock Market
    JEL: G02 G17
    Date: 2015
  9. By: Jari Hännäkäinen (School of Management, University of Tampere)
    Abstract: This paper examines the predictive power of interest rate spreads when the zero lower bound restriction for monetary policy is binding. We show that this restriction has a major eect on the predictive content of some interest rate spreads. Most importantly, we nd that the term spread outperforms the AR benchmark in real-time forecasting exercise when the short-term rate is at the zero lower bound, but not otherwise. On the other hand, our results indicate that the dierence between the 30-year mortgage rate and ten-year Treasury bond rate is a robust predictor of future economic activity.
    Keywords: business fl uctuations, forecasting, interest rate spreads, monetary policy, zero lower bound
    JEL: C53 E32 E44 E52 E58
    Date: 2013–10
  10. By: Jari Hännäkäinen (School of Management, University of Tampere)
    Abstract: This paper analyzes the relative performance of multi-step forecasting methods in the presence of breaks and data revisions. Our Monte Carlo simulations indicate that the type and the timing of the break affect the relative accuracy of the methods. The iterated method typically performs the best in unstable environments, especially if the parameters are subject to small breaks. This result holds regardless of whether data revisions add news or reduce noise. Empirical analysis of real-time U.S. output and inflation series shows that the alternative multi-step methods only episodically improve upon the iterated method.
    Keywords: Structural breaks, multi-step forecasting, intercept correction, real-time data
    JEL: C22 C53 C82
    Date: 2014–05
  11. By: Jari Hännäkäinen (School of Management, University of Tampere)
    Abstract: This paper re-examines the out-of-sample predictive power of interest rate spreads when the short-term nominal rates have been stuck at the zero lower bound and the Fed has used unconventional monetary policy. Our results suggest that the predictive power of some interest rate spreads have changed since the beginning of this period. In particular, the term spread has been a useful leading indicator since December 2008, but not before that. Credit spreads generally perform poorly in the zero lower bound and unconventional monetary policy period. However, the mortgage spread has been a robust predictor of economic activity over the 2003–2014 period.
    Keywords: business fluctuations, forecasting, interest rate spreads, monetary policy, zero lower bound, real-time data
    JEL: C53 E32 E44 E52 E58
    Date: 2014–06
  12. By: Porshakov , Alexey (BOFIT); Deryugina , Elena (BOFIT); Ponomarenko , Alexey (BOFIT); Sinyakov , Andrey (BOFIT)
    Abstract: Real-time assessment of quarterly GDP growth rates is crucial for evaluation of economy’s current perspectives given the fact that respective data is normally subject to substantial publication delays by national statistical agencies. Large information sets of real-time indicators which could be used to approximate GDP growth rates in the quarter of interest are in practice characterized by unbalanced data, mixed frequencies, systematic data revisions, as well as a more general curse of dimensionality problem. The latter issues could, however, be practically resolved by means of dynamic factor modeling that has recently been recognized as a helpful tool to evaluate current economic conditions by means of higher frequency indicators. Our major results show that the performance of dynamic factor models in predicting Russian GDP dynamics appears to be superior as compared to other common alternative specifications. At the same time, we empirically show that the arrival of new data seems to consistently improve DFM’s predictive accuracy throughout sequential nowcast vintages. We also introduce the analysis of nowcast evolution resulting from the gradual expansion of the dataset of explanatory variables, as well as the framework for estimating contributions of different blocks of predictors into now-casts of Russian GDP.
    Keywords: GDP nowcast; dynamic factor models; principal components; Kalman filter; nowcast evolution
    JEL: C53 C82 E17
    Date: 2015–05–28
  13. By: Antoine Djogbenou; Sílvia Gonçalves; Benoit Perron
    Abstract: This paper considers bootstrap inference in a factor-augmented regression context where the errors could potentially be serially correlated. This generalizes results in Gonçalves and Perron (2013) and makes the bootstrap applicable to forecasting contexts where the forecast horizon is greater than one. We propose and justify two residual-based approaches, a block wild bootstrap (BWB) and a dependent wild bootstrap (DWB). Our simulations document improvement in coverage rates of confidence intervals for the coefficients when using BWB or DWB relative to both asymptotic theory and the wild bootstrap when serial correlation is present in the regression errors.
    Keywords: Factor model, bootstrap, serial correlation, forecast.,
    Date: 2015–05–29
  14. By: Rudan Wang (Department of Economics, University of Bath, UK); Bruce Morley (Department of Economics, University of Bath, UK); Javier Ordóñez (Department of Economics, Universitat Jaume I, Castellón, Spain)
    Abstract: The aim of this study is to develop models of the Taylor rule and a Taylor rule based exchange rate model incorporating wealth effects, as represented by both asset prices and asset wealth. In addition these wealth effects are further divided into stock market and housing wealth. Using data for Australia, Sweden, UK and the US, the Taylor model is estimated and then used to forecast out-of-sample. The results suggest that the effects of the asset prices and wealth on the Taylor rule are mixed and depend on the country and the form the wealth takes. The outof-sample forecast performance of both the wealth augmented Taylor rule model and Taylor rule exchange rate model are then compared with the conventional Taylor Rule model and a random walk and overall the wealth augmented models outperform the conventional model and random walk in these countries.
    Keywords: exchange rate, wealth effect, forecast
    JEL: F30 F37
    Date: 2015
  15. By: Federico M. Bandi; Benoit Perron; Andrea Tamoni; Claudio Tebaldi
    Abstract: Stock return predictive relations found to be elusive when using raw data may hold true for different layers in the cascade of economic shocks. Consistent with this logic, we model stock market returns and their predictors as aggregates of uncorrelated components (details) operating over different scales and introduce a notion of scale-specific predictability, i.e., predictability on the details. We study and formalize the link between scale-specific predictability and aggregation. Using both direct extraction of the details and aggregation, we provide strong evidence of risk compensations in long-run stock market returns - as well as of an unusually clear link between macroeconomic uncertainty and uncertainty in financial markets - at frequencies lower than the business cycle. The reported tent-shaped behavior in long-run predictability is shown to be a theoretical implication of our proposed modelling approach.
    Keywords: : long run, predictability, aggregation, risk-return trade-off, Fisher hypothesis,
    JEL: C22 E32 E44 G12 G17
    Date: 2015–05–29

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.