nep-for New Economics Papers
on Forecasting
Issue of 2018‒05‒21
eleven papers chosen by
Rob J Hyndman
Monash University

  1. Forecasting GDP: Do Revisions Matter? By Check, Adam J.; Nolan, Anna K.; Schipper, Tyler C.
  2. The Hidden Predictive Power of Cryptocurrencies: Evidence from US Stock Market By Kazeem Isah; Ibrahim D. Raheem
  3. How far can we forecast? Statistical tests of the predictive content By Breitung, Jörg; Knüppel, Malte
  4. The accountability imperative for quantifiying the uncertainty of emission forecasts : evidence from Mexico By Daniel Puig; Oswaldo Morales Napoles; Fatemeh Bakhtiari; Gissela Landa
  5. Developing an underlying inflation gauge for China By Amstad, Marlene; Ye, Huan; Ma, Guonan
  6. A mixture autoregressive model based on Student's $t$-distribution By Mika Meitz; Daniel Preve; Pentti Saikkonen
  7. Distributional regression forests for probabilistic precipitation forecasting in complex terrain By Lisa Schlosser; Torsten Hothorn; Reto Stauffer; Achim Zeileis
  8. Modeling the future evolution of the virtual water trade network By Andrea Fracasso; Massimo Riccabonii; Martina Sartori; Stefano Schiavo
  9. Forecasting the Need for Budget Funds in the Implementation of Reforms in the Education System By Klyachko, Tatiana; Tokareva, Galina
  10. The Medium-Term Forecast for the Development of Professional Education of the Russian Federation (by Level of Professional Education) By Belyakov, Sergei
  11. Deciphering Professional Forecasters’ Stories - Analyzing a Corpus of Textual Predictions for the German Economy By Ulrich Fritsche; Johannes Puckelwald

  1. By: Check, Adam J.; Nolan, Anna K.; Schipper, Tyler C.
    Abstract: This paper investigates the informational content of regular revisions to real GDP growth and its components. We perform a real-time forecasting exercise for the advance estimate of real GDP growth using dynamic regression models that include GDP and GDP component revisions. Echoing other work in the literature, we find little evidence that including aggregate GDP growth revisions improves forecast accuracy relative to an AR(1) baseline model; however, when we include revisions to components of GDP (i.e. C, I, G, X, and M) we find improvements in forecast accuracy. Overall, nearly 68\% of all models that contain subsets of component revisions outperform our baseline model. The "best" component-augmented model forecasts roughly 0.2 percentage points better, and a large subset of models improve RMSFE by more than 5%. Finally, we use Bayesian model comparison to demonstrate that differences in forecast performance are unlikely to be the result of statistical noise. Our results imply that component revisions, in particular to consumption, contain important information for forecasting GDP growth.
    Keywords: Data revisions, real-time data, forecasting, GDP
    JEL: C11 C53 C82 E01
    Date: 2018–04–01
  2. By: Kazeem Isah (Centre for Econometric and Allied Research, University of Ibadan); Ibrahim D. Raheem (School of Economics, University of Kent, Canterbury, UK)
    Abstract: This paper is motivated by the news that the surge in cryptocurrencies is an important candidate to in explaining the plummeting stock markets. To validate this believe, we construct a predictive model in which cryptocurrencies are identified as the predictors of US stock returns. The inherent statistical properties of cryptocurrencies such as persistence, endogeneity, and conditional heteroscedasticity are being accounted for in the Westerlund and Narayan (2015) estimator. Three salient results emanated from our estimations. First, we validated the importance of cryptocurrencies in predicting US stock prices; second, the cryptocurrencies predictive model outperforms the conventional time-series models such as Autoregressive Integrated Moving Average (ARIMA) model and the Autoregressive Fractionally Integrated Moving Average (ARFIMA); third, our results are robust to different method of forecast performance evaluation measures and different sub-sample periods. These results have important policy implications for the investors and policymakers.
    Keywords: Stock Prices, Cryptocurrency, Digital Asset Prices, Predictive Model, Forecast Evaluation
    JEL: C52 C53 G11 G14 G17
    Date: 2018–05
  3. By: Breitung, Jörg; Knüppel, Malte
    Abstract: Forecasts are useless whenever the forecast error variance fails to be smaller than the unconditional variance of the target variable. This paper develops tests for the null hypothesis that forecasts become uninformative beyond some limiting forecast horizon h. Following Diebold and Mariano (DM, 1995) we propose a test based on the comparison of the mean-squared error of the forecast and the sample variance. We show that the resulting test does not possess a limiting normal distribution and suggest two simple modifications of the DM-type test with different limiting null distributions. Furthermore, a forecast encompassing test is developed that tends to better control the size of the test. In our empirical analysis, we apply our tests to macroeconomic forecasts from the survey of Consensus Economics. Our results suggest that forecasts of macroeconomic key variables are barely informative beyond 2-4 quarters ahead.
    Keywords: Hypothesis Testing,Predictive Accuracy,Informativeness
    JEL: C12 C32 C53
    Date: 2018
  4. By: Daniel Puig (United Nations Environmental Programme); Oswaldo Morales Napoles; Fatemeh Bakhtiari; Gissela Landa (Observatoire français des conjonctures économiques)
    Abstract: Governmental climate change mitigation targets are typically developed with the aid of forecasts of greenhouse-gas emissions. The robustness and credibility of such forecasts depends, among other issues, on the extent to which forecasting approaches can reflect prevailing uncertainties. We apply a transparent and replicable method to quantify the uncertainty associated with projections of gross domestic product growth rates for Mexico, a key driver of greenhouse-gas emissions in the country. We use those projections to produce probabilistic forecasts of greenhouse-gas emissions for Mexico. We contrast our probabilistic forecasts with Mexico’s governmental deterministic forecasts. We show that, because they fail to reflect such key uncertainty, deterministic forecasts are ill-suited for use in target-setting processes. We argue that (i) guidelines should be agreed upon, to ensure that governmental forecasts meet certain minimum transparency and quality standards, and (ii) governments should be held accountable for the appropriateness of the forecasting approach applied to prepare governmental forecasts, especially when those forecasts are used to derive climate change mitigation targets.
    Keywords: Uncertainty; Projections; Structured expert judgment; Accountability; Emission-reduction targets; Gross domestic product growth rates
    Date: 2017–09
  5. By: Amstad, Marlene; Ye, Huan; Ma, Guonan
    Abstract: Inflation in emerging markets is often driven by large, persistent changes in food and energy prices. Core inflation measures that neglect or under-weight volatile CPI subcomponents such as food and energy risk excluding information helpful in assessing current and future inflation trends. This paper develops an underlying inflation gauge (UIG) for China, extracting the persistent part of the common component in a broad dataset of price and non-price variables. Our proposed UIG for China avoids the excess volatility reduction that plagues traditional Chinese core inflation measures. When forecasting headline CPI, the proposed UIG outperforms traditional core inflation measures over a variety of samples.
    JEL: C13 C33 E31 E37 G15 C43
    Date: 2018–04–27
  6. By: Mika Meitz; Daniel Preve; Pentti Saikkonen
    Abstract: A new mixture autoregressive model based on Student's $t$-distribution is proposed. A key feature of our model is that the conditional $t$-distributions of the component models are based on autoregressions that have multivariate $t$-distributions as their (low-dimensional) stationary distributions. That autoregressions with such stationary distributions exist is not immediate. Our formulation implies that the conditional mean of each component model is a linear function of past observations and the conditional variance is also time varying. Compared to previous mixture autoregressive models our model may therefore be useful in applications where the data exhibits rather strong conditional heteroskedasticity. Our formulation also has the theoretical advantage that conditions for stationarity and ergodicity are always met and these properties are much more straightforward to establish than is common in nonlinear autoregressive models. An empirical example employing a realized kernel series based on S&P 500 high-frequency data shows that the proposed model performs well in volatility forecasting.
    Date: 2018–05
  7. By: Lisa Schlosser; Torsten Hothorn; Reto Stauffer; Achim Zeileis
    Abstract: To obtain a probabilistic model for a dependent variable based on some set of explanatory variables, a distributional approach is often adopted where the parameters of the distribution are linked to regressors. In many classical models this only captures the location of the distribution but over the last decade there has been increasing interest in distributional regression approaches modeling all parameters including location, scale, and shape. Notably, so-called non-homogenous Gaussian regression (NGR) models both mean and variance of a Gaussian response and is particularly popular in weather forecasting. More generally, the GAMLSS framework allows to establish generalized additive models for location, scale, and shape with smooth linear or nonlinear effects. However, when variable selection is required and/or there are non-smooth dependencies or interactions (especially unknown or of high-order), it is challenging to establish a good GAMLSS. A natural alternative in these situations would be the application of regression trees or random forests but, so far, no general distributional framework is available for these. Therefore, a framework for distributional regression trees and forests is proposed that blends regression trees and random forests with classical distributions from the GAMLSS framework as well as their censored or truncated counterparts. To illustrate these novel approaches in practice, they are employed to obtain probabilistic precipitation forecasts at numerous sites in a mountainous region (Tyrol, Austria) based on a large number of numerical weather prediction quantities. It is shown that the novel distributional regression forests automatically select variables and interactions, performing on par or often even better than GAMLSS specified either through prior meteorological knowledge or a computationally more demanding boosting approach.
    Keywords: parametric models, regression trees, random forests, recursive partitioning, probabilistic forecasting, GAMLSS
    Date: 2018–08
  8. By: Andrea Fracasso (Department of Economics [Università di Trento]); Massimo Riccabonii (School for advanced studies Lucca); Martina Sartori; Stefano Schiavo (Observatoire français des conjonctures économiques)
    Abstract: The paper investigates how the topological features of the virtual water (VW) network and the size of the associated VW flows are likely to change over time, under different socio-economic and climate scenarios. We combine two alternative models of network formation –a stochastic and a fitness model, used to describe the structure of VW flows- with a gravity model of trade to predict the intensity of each bilateral flow. This combined approach is superior to existing methodologies in its ability to replicate the observed features of VW trade. The insights from the models are used to forecast future VW flows in 2020 and 2050, under different climatic scenarios, and compare them with future water availability. Results suggest that the current trend of VW exports is not sustainable for all countries. Moreover, our approach highlights that some VW importers might be exposed to “imported water stress” as they rely heavily on imports from countries whose water use is unsustainable.
    Keywords: Virtual water trade; Complex networks; Fitness model; Agricultural production; Preferential attachment; Gravity model; Water stress
    JEL: F14 F18 Q25 Q56
    Date: 2017–12
  9. By: Klyachko, Tatiana (Russian Presidential Academy of National Economy and Public Administration (RANEPA)); Tokareva, Galina (Russian Presidential Academy of National Economy and Public Administration (RANEPA))
    Abstract: This paper builts a system of forecasts for the budget financing of the education system and its levels, as well as tasks to be solved in this area. Three scenarios for the development of education up to 2024 are presented, the consequences and risks of their implementation are determined. It is shown that the main beneficiaries in obtaining budget funds in the forecast period are pre-school and general education. Secondary vocational education loses in budget financing in almost all variants of the forecast. With regard to the risks of the development of the education system, it's primarily related to the underfunding of educational programs for teachers, without which it is impossible either to update the content of education, or to include new educational technologies in the educational process, as well as a weak growth in real terms of budget expenditures for applied scientific research in the field of education, since it is scientific research that should determine the direction of reforming the various levels of education and New mechanisms for managing education in changing socio-economic conditions, to set an adequate update of the content of general and professional education.
    Date: 2018–04
  10. By: Belyakov, Sergei (Russian Presidential Academy of National Economy and Public Administration (RANEPA))
    Abstract: The development of medium-term forecasts of the economy and finance of the education system, conducted by the Center for Continuous Education Economics of the RANEPA in previous years, made it possible to determine a number of requirements for this work. In particular, the initial data for the development of forecasts should be data from official statistics, as well as reports on the implementation of the consolidated budget of the Russian Federation in terms of education expenditure. In addition, for the development of forecasts, some ratios of normative nature are necessary: funding standards, permissible ratios of the number of children attending pre-school educational institutions, the number of places in them, the ratio of the number of children and pedagogical workers, the established (certain) ratios between the number of students in organizations of general school) education and teachers in them and some others. The possibility of changing these ratios allows the development of various variants of forecasts, which creates an information basis for the adoption of appropriate managerial decisions.
    Date: 2018–04
  11. By: Ulrich Fritsche (Universität Hamburg (University of Hamburg)); Johannes Puckelwald (Universität Hamburg (University of Hamburg))
    Abstract: We analyze a corpus of 564 business cycle forecast reports for the German economy. The dataset covers nine institutions and 27 years. From the entire reports we select the parts that refer exclusively to the forecast of the German economy. Sentiment and frequency analysis confirm that the mode of the textual expressions varies with the business cycle in line with the hypothesis of adaptive expectations. A calculated 'uncertainty index' based on the occurrence of modal words matches with the economic policy uncertainty index by Baker et al. (2016). The latent Dirichlet allocation (LDA) model and the structural topic model (STM) indicate that topics are significantly state- and time-dependent and different across institutions. Positive or negative forecast 'surprises' experienced in the previous year have an impact on the content of topics.
    Keywords: Sentiment analysis, text analysis, uncertainty, business cycle forecast, forecast error, expectation, adaptive expectation, latent Dirichlet allocation, structural topic model
    JEL: E32 E37 C49
    Date: 2018–05

This nep-for issue is ©2018 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.