nep-for New Economics Papers
on Forecasting
Issue of 2016‒10‒16
nine papers chosen by
Rob J Hyndman
Monash University

  1. Tricks for improving non-homogeneous regression for probabilistic precipitation forecasts: Perfect predictions, heavy tails, and link functions By Manuel Gebetsberger; Jakob W. Messner; Georg J. Mayr; Achim Zeileis
  2. Evaluating a Long-run Forecast: The World Bank Poverty Forecasts By Jin Ho Kim; Herman O. Stekler
  3. Forecasting inflation in post-oil boom years: A case for non-linear models? By Vugar Ahmadov; Shaig Adigozalov; Salman Huseynov; Fuad Mammadov; Vugar Rahimov
  4. Short term Bayesian inflation forecasting for Tunisia By Dahem, Ahlem
  5. Forecasting the Estonian rate of inflation using factor models By Nicolas Reigl
  6. Daily Currency Interventions in Emerging Markets: Incorporating Reserve Accumulation By Murat Midiliç; Michael Frömmel
  7. Forecasting Goods and Services Inflation in Sweden By Mossfeldt, Marcus; Stockhammar, Pär
  8. Adjusting for Information Content when Comparing Forecast Performance By Andersson, Michael K.; Aranki, Ted; Reslow, André
  9. Revisiting the transitional dynamics of business-cycle phases with mixed frequency data By Marie Bessec

  1. By: Manuel Gebetsberger; Jakob W. Messner; Georg J. Mayr; Achim Zeileis
    Abstract: Raw ensemble forecasts display large errors in predicting precipitation amounts and its forecast uncertainty, especially in mountainous regions where local effects are often not captured. Therefore, statistical post-processing is typically applied to obtain automatically corrected weather forecasts where precipitation represents one of the most challenging quantities. This study applies the non-homogenous regression framework as a start-of-the-art ensemble post-processing technique to predict a full forecast distribution and improves its forecast performance with three statistical tricks. First of all, a novel split-type approach effectively accounts for perfect ensemble predictions that can occur. Additionally, the statistical model assumes a censored logistic distribution to deal with the heavy tails of precipitation amounts. Finally, the optimization of regression coefficients for the scale parameter is investigated with suitable link-functions. These three refinements are tested for stations in the European Alps for lead-times from +24h to +48h and accumulation periods of 24 and 6 hours. Results highlight an improvement due to a combination of the three statistical tricks against the default post-processing method which does not account for perfect ensemble predictions. Probabilistic forecasts for precipitation amounts as well as the probability of precipitation events could be improved, especially for 6 hour sums.
    Keywords: non-homogeneous regression, censored logistic distribution, log-link, probabilistic precipitation forecasts, operational forecasting
    JEL: C53 C61 Q50
    Date: 2016–10
  2. By: Jin Ho Kim (The George Washington University); Herman O. Stekler (The George Washington University)
    Abstract: This paper examines an issue in long-run forecasting, evaluating a forecast for which the actual data are not yet available. In this case, we analyze the World Bank’s forecasts of the poverty headcount made in 2002, but the actual data for the terminal date will not be available for some time. The methodology requires one to infer a forecast for an intermediate date for which the data are available. We show that the long-rum projections were extremely accurate because they are consistent with the trends that are observed in the latest available data.
    Date: 2016–09
  3. By: Vugar Ahmadov (Central Bank of Azerbaijan Republic); Shaig Adigozalov (Central Bank of Azerbaijan Republic); Salman Huseynov (Central Bank of Azerbaijan Republic); Fuad Mammadov (Central Bank of Azerbaijan Republic); Vugar Rahimov (Central Bank of Azerbaijan Republic)
    Abstract: In this study, we investigate relative performance of various non-linear models against that of an autoregressive model in forecasting future inflation. We find that non-linear models have trivial forecast superiority over the univariate autoregressive model in terms of central forecast accuracy. They also perform poorly when their forecasts are measured against those of the 3 variables VAR model. In addition, we also show that non-linear models cannot beat the random walk in terms of central forecast accuracy which is in line with the previous literature on Azerbaijan during the post-oil boom years. However, we also demonstrate that non-linear models still have clear forecast advantage over both linear and random walk models in predicting forecast density.
    Keywords: Forecasting, Bayesian methods, Non-linear models
    JEL: C11 C13 C32 C53
    Date: 2016–04–06
  4. By: Dahem, Ahlem
    Abstract: In order to explain clearly inflation forecasting and the dynamic of Tunisian prices, this paper uses two econometric approaches, the Standard VAR and Bayesian VAR (BVAR), to assess three models for predicting inflation, the mark-up model, the monetary model and Phillips curve over the period 1990 Q1 – 2013 Q4. In order to compare predictions, an out-of-sample estimation was conducted. We used the structural break test of Bai & Perron (1998, 2003) and the RMSE criterion for both inflation indices: CPI and PPI. We found that the Bayesian VECM mark-up model is best suited to forecast inflation for Tunisia. Our conclusions corroborate the literature of Bayesian VAR forecasting. Our findings indicate that the models which incorporate more economic information outperform the benchmark autoregressive models (AR (1) and AR (2)). The results reveal that forecasting with the BVECM markup model leads to a reduction in forecasting error compared to the other models. The results of the study are relevant to decision-makers to predict inflation in the short- and long-terms in Tunisia and may help them adopt the appropriate strategies to contain inflation.
    Keywords: Bayesian VAR - Bayesian VECM - Inflation forecasting - Mark-up Model - Monetary Model - Phillips Curve
    JEL: C11 C51 C53 E31 E37
    Date: 2015–09–01
  5. By: Nicolas Reigl
    Abstract: The paper presents forecasts of the headline and core inflation in Estonia with factor models in a recursive pseudo out-of-sample framework. The factors are constructed with a principal component analysis and are then incorporated into vector autoregressive forecasting models. The analyses show that certain factor-augmented vector autoregressive models improve upon a simple univariate autoregressive model but the forecasting gains are small and not systematic. Models with a small number of factors extracted from a large dataset are best suited for forecasting headline inflation. In contrast models with a larger number of factors extracted from a small dataset outperform the benchmark model in the forecast of Estonian headline and, especially, core inflation
    Keywords: Factor models, factor-augmented vector autoregressive models, factor analysis, principal components, inflation forecasting, forecast evaluation, Estonia
    JEL: C32 C38 C53
    Date: 2016–10–10
  6. By: Murat Midiliç (Ghent University); Michael Frömmel (Ghent University)
    Abstract: This study considers international reserve management motivation of emerging market central banks in foreign exchange market interventions. Emerging market central banks use currency intervention as a policy tool against exchange rate movements and accumulate international reserves as an insurance against sudden-stops in capital flows. To account for both of these motivations, a model of infrequent interventions only with exchange rates is extended to include international reserves-to-gross domestic product (GDP) ratio at the daily frequency. Daily values of the ratio are forecast using the Mixed Data Sampling (MIDAS) model and exchange rate returns. The model is estimated by using the floating exchange rate regime period data of Turkey. Compared with the benchmark model, it is shown that the MIDAS model does a better job in the forecasting of the reserve-to-GDP ratio. In addition to that, there are breaks in the interventions policy in Turkey, and the extended intervention model performs better than the model only with exchange rates especially in predicting purchases of US Dollar.
    Keywords: currency intervention, international reserves, emerging markets, Turkey, mixed data sampling
    JEL: F31 E58 G15
  7. By: Mossfeldt, Marcus (National Institute of Economic Research); Stockhammar, Pär (National Institute of Economic Research)
    Abstract: In this paper, we make use of a Bayesian VAR (BVAR) model to con-duct an out-of-sample forecast exercise for goods and services inflation in Sweden. Our interest in goods and services prices stems from the fact that they make up over 70 per cent of the CPI index and that they are more directly affected by the macroeconomic development than other parts of the CPI. We find that the BVAR models generally outperform both univariate models for goods and services inflation, as well as forecasts made by the National Institute of Economic Research in Sweden. This might indicate that Faust and Wright’s (2013) rather negative conclusion that inflation models cannot beat judgmental forecasts and inflation expectations might be wrong, at least in the case of Sweden.
    Keywords: Bayesian VAR; Inflation; Out-of-sample forecasting precision
    JEL: C53 E31
    Date: 2016–10–12
  8. By: Andersson, Michael K. (Finansinspektionen); Aranki, Ted (Monetary Policy Department, Central Bank of Sweden); Reslow, André (Monetary Policy Department, Central Bank of Sweden)
    Abstract: Cross institutional forecast evaluations may be severely distorted by the fact that forecasts are made at different points in time, and thus with different amount of information. This paper proposes a method to account for these differences. The method computes the timing effect and the forecaster's ability simultaneously. Monte Carlo simulation demonstrate that evaluations that do not adjust for the differences in information content may be misleading. In addition, the method is applied on a real-world data set of 10 Swedish forecasters for the period 1999-2015. The results show that the ranking of the forecasters is affected by the proposed adjustment.
    Keywords: Forecast error; Forecast comparison; Publication time; Evaluation; Error component model; Panel data
    JEL: C23 C53 E37
    Date: 2016–08–01
  9. By: Marie Bessec (LEDa - Laboratoire d'Economie de Dauphine - Université Paris-Dauphine)
    Abstract: This paper introduces a Markov-switching model in which transition probabilities depend on higher frequency indicators and their lags through polynomial weight-ing schemes. The MSV-MIDAS model is estimated via maximum likelihood (ML) methods. The estimation relies on a slightly modified version of Hamilton's recursive filter. We use Monte Carlo simulations to assess the robustness of the estimation procedure and related test statistics. The results show that ML provides accurate estimates, but they suggest some caution in interpreting the tests of the parameters involved in the transition probabilities. We apply this new model to the detection and forecasting of business cycle turning points in the United States. We properly detect recessions by exploiting the link between GDP growth and higher frequency variables from financial and energy markets. The spread term is a particularly useful indicator to predict recessions in the United States. The empirical evidence also supports the use of functional polynomial weights in the MIDAS specification of the transition probabilities.
    Keywords: Markov-switching,mixed frequency data,business cycles
    Date: 2016–09–01

This nep-for issue is ©2016 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.