nep-for New Economics Papers
on Forecasting
Issue of 2019‒06‒10
seven papers chosen by
Rob J Hyndman
Monash University

  1. Forecasting the Colombian Unemployment Rate Using Labour Force Flows By Francisco Lasso-Valderrama; Héctor M. Zárate-Solano
  2. Sectoral Production and Diffusion Index Forecasts for Output in Lithuania By Soroosh Soofi-Siavash; Kristina Barauskaite
  3. Short-term forecasting of the US unemployment rate By Maas, Benedikt
  4. Predicting the U.S. Stock Market Return: Evidence from the Improved Augmented Regression Method By Jurdi, Doureige; Kim, Jae
  5. Forecasting Realized Gold Volatility: Is there a Role of Geopolitical Risks? By Konstantinos Gkillas; Rangan Gupta; Christian Pierdzioch
  6. Designing Robust Monetary Policy Using Prediction Pools By Deak, S.; Levine, P.; Mirza, A.; Pearlman, J.
  7. Inducing Sparsity and Shrinkage in Time-Varying Parameter Models By Huber, Florian; Koop, Gary; Onorante, Luca

  1. By: Francisco Lasso-Valderrama (Banco de la República de Colombia); Héctor M. Zárate-Solano (Banco de la República de Colombia)
    Abstract: Accurate predictions of future magnitudes of the unemployment rate are crucial for monetary policy. This paper investigates whether the use of disaggregated household survey data improves the forecasts of the Colombian 13 cities unemployment rate. We conduct an outof-sample forecast exercise to compare the performance of a model that incorporates flows of workers across different states of the labour market to that of various macroeconomic non-structural models. The paper follows the approach proposed by Barnichon & Nekarda (2013). Our results indicate that the two-state-flow model provides substantially better forecasts of the unemployment rate over longer horizons (more than five months ahead). Additionally, when forecasts are combined, significant gains in every forecasting horizon occurs. This combined forecast shows a 23% reduction in overall RMSE. **** ABSTRACT: En este documento se evalúan los pronósticos de la tasa de desempleo urbana en Colombia utilizando varias metodologías. La primera se basa en las propiedades estadísticas de la serie de tiempo de la tasa de desempleo. La segunda considera la relación entre el crecimiento del producto y los cambios en el desempleo, conocida como la Ley de Okun. Finalmente, con base en los microdatos de las encuestas de hogares se calculan los flujos de trabajadores del mercado laboral para pronosticar la tasa de desempleo de acuerdo con Barnichon y Nekarda (2013). La evaluación de los pronósticos fuera de muestra indica que el modelo de dos estados (ocupado-desocupado) es el mejor en horizontes superiores a cinco meses. Por su parte, los modelos ARIMA y la Ley de Okun compiten en precisión en horizontes de corto plazo. Cabe destacar que la combinación de los modelos de pronóstico genera ganancias significativas en todos los horizontes, alcanzando una reducción global de 23% en la raíz del error cuadrático medio. Classification-JEL: C53, E24, E27, E3, J64
    Keywords: Forecasting, unemployment, VAR models, labour market flows, Pronósticos, desempleo, modelos VAR, flujos del mercado laboral
    Date: 2019–05
  2. By: Soroosh Soofi-Siavash (Bank of Lithuania & Kaunas University of Technology); Kristina Barauskaite (Bank of Lithuania & ISM University of Management and Economics)
    Abstract: In this paper, we develop and describe quarterly data on disaggregated sectors in Lithuania which covers the period 1998-2018. The data is useful for empirical studies requiring panels with a large number of time observations and a large number of cross-sectional units. We follow the NACE2 level of disaggregation in developing our data, thus allowing us to combine the data with world input-output tables which we in turn use to identify the hubs and the main importing and exporting sectors within the economy. The data is then used for forecasting the growth rate of GDP. There is a substantial increase in the degree of covariation among sectoral production growth rates, which is observed using a split sample around 2008. When we apply factor methods, we find that this strong covariation can be explained by a few factors which are closely correlated to the growth of the retail and wholesale sectors. For GDP growth, the forecasts we consider are the diffusion index forecasts produced using a few indexes that summarize sectoral data, and the forecasts produced using the production growth of selected hubs and importing and exporting sectors. We find that the diffusion indexes and the production growth of sectors that make heavy use of imported inputs in their production have interesting forecasting power for the growth rate of GDP in the 2006-2011 and 2012-2018 periods.
    Keywords: factor model, forecasting, input-output linkages, intersectoral networks
    JEL: E27 E37 C3 C67
    Date: 2019–05–29
  3. By: Maas, Benedikt
    Abstract: This paper aims to assess whether Google search data is useful when predicting the US unemployment rate among other more traditional predictor variables. A weekly Google index is derived from the keyword “unemployment” and is used in diffusion index variants along with the weekly number of initial claims and monthly estimated latent factors. The unemployment rate forecasts are generated using MIDAS regression models that take into account the actual frequencies of the predictor variables. The forecasts are made in real-time and the forecasts of the best forecasting models exceed, for the most part, the root mean squared forecast error of two benchmarks. However, as the forecasting horizon increases, the forecasting performance of the best diffusion index variants decreases over time, which suggests that the forecasting methods proposed in this paper are most useful in the short-term.
    Keywords: Forecasting, Unemployment rate, MIDAS, Google Trends
    JEL: C32 C53 E32
    Date: 2019–04–16
  4. By: Jurdi, Doureige; Kim, Jae
    Abstract: We examine whether the stock market return is predictable from a range of financial indicators and macroeconomic variables, using monthly U.S. data from 1926 to 2012. We adopt the improved augmented regression method for parameter estimation, statistical inference, and out-of-sample forecasting. By employing moving sub-sample windows, we evaluate the time-variation of predictability free from data snooping bias and report changes in predictability dynamics over time. Although we may find statistically significant in-sample predictability from time to time, the associated effect size estimates are fairly small in most cases. We also find weak predictability of the stock market return from multistep ahead (out-of-sample) forecasts. In addition, we find that mean-variance investors realize sporadic economic gains in utility based on predictive regression forecasts relative to naive model historic average forecasts
    Keywords: Bias-correction; Financial ratios; Forecasting; Return predictability; Utility gains
    JEL: G17
    Date: 2019–05–20
  5. By: Konstantinos Gkillas (Department of Business Administration, University of Patras-University Campus, Rio, P.O. Box 1391, 26500 Patras, Greece); Rangan Gupta (Department of Economics, University of Pretoria, Pretoria, South Africa); Christian Pierdzioch (Department of Economics, Helmut Schmidt University, Holstenhofweg 85, P.O.B. 700822, 22008 Hamburg, Germany)
    Abstract: We use a quantile-regression heterogeneous autoregressive realized volatility (QR-HARRV) model to study whether geopolitical risks have predictive value in sample and out-of-sample for realized gold-returns volatility estimated from intradaily data. We consider overall geopolitical risks along with a decomposition into actual risks (i.e., acts) and threats, and we control for overall the impact of economic policy uncertainty (EPU). We find that, after controlling for EPU, the components of geopolitical risks have predictive power for realized volatility mainly at a longer forecast horizon when we account for the potential asymmetry of the loss function a forecaster uses to evaluate forecasts.
    Keywords: Gold-price returns, Realized volatility, Geopolitical risks, Forecasting
    Date: 2019–05
  6. By: Deak, S.; Levine, P.; Mirza, A.; Pearlman, J.
    Abstract: How should a forward-looking policy maker conduct monetary policy when she has a finite set of models at her disposal, none of which are believed to be the true data generating process? In our approach, the policy makerfirst assigns weights to models based on relative forecasting performance rather than in-sample fit, consistent with her forward-looking objective. These weights are then used to solve a policy design prob-lem that selects the optimized Taylor-type interest-rate rule that is robust to model uncertainty across a set of well-established DSGE models with and without financial frictions. We find that the choice of weights has a significant impact on the robust optimized rule which is more inertial and aggressive than either the non-robust single model counterparts or the optimal robust rule based on backward-looking weights asin the common alternative Bayesian Model Averaging. Importantly, we show that a price-level rule has excellent welfare and robustness properties, and therefore should be viewed as a key instrument for policy makers facing uncertainty over the nature offinancial frictions.
    JEL: H
    Date: 2019
  7. By: Huber, Florian (University of Salzburg); Koop, Gary (University of Strathclyde); Onorante, Luca (European Central Bank)
    Abstract: Time-varying parameter (TVP) models have the potential to be over-parameterized, particularly when the number of variables in the model is large. Global-local priors are increasingly used to induce shrink- age in such models. But the estimates produced by these priors can still have appreciable uncertainty. Sparsification has the potential to remove this uncertainty and improve forecasts. In this paper, we develop computationally simple methods which both shrink and sparsify TVP models. In a simulated data exercise we show the benefits of our shrink-then-sparsify approach in a variety of sparse and dense TVP regressions. In a macroeconomic forecast exercise, we find our approach to substantially improve forecast performance relative to shrinkage alone.
    Keywords: Sparsity; shrinkage; hierarchical priors; time varying parameter regression
    JEL: C11 C30 D31 E30
    Date: 2019–05–26

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.