|
on Forecasting |
| By: | Bidong Liu; Jakub Nowotarski; Tao Hong; Rafal Weron |
| Abstract: | Majority of the load forecasting literature has been on point forecasting, which provides the expected value for each step throughout the forecast horizon. In the smart grid era, the electricity demand is more active and less predictable than ever before. As a result, probabilistic load forecasting, which provides additional information on the variability and uncertainty of future load values, is becoming of great importance to power systems planning and operations. This paper proposes a practical methodology to generate probabilistic load forecasts by performing Quantile Regression Averaging (QRA) on a set of sister point forecasts. There are two major benefits of the proposed approach: 1) it can leverage the development in the point load forecasting literature over the past several decades; and 2) it does not rely so much on high quality expert forecasts, which are rarely achievable in load forecasting practice. To demonstrate the effectiveness of the proposed approach and make the results reproducible to the load forecasting community, we construct a case study using the publicly available data from the Global Energy Forecasting Competition 2014. Comparing with the benchmark methods that utilize the variability of a selected individual forecast, the proposed approach leads to dominantly better performance as measured by the pinball loss function and the Winkler score. |
| Keywords: | Electric load forecasting; Forecast combination; Pinball loss function; Probabilistic forecasting; Prediction interval; Quantile regression; Sister forecast; Winkler score |
| JEL: | C22 C32 C53 Q47 |
| Date: | 2015–02–03 |
| URL: | http://d.repec.org/n?u=RePEc:wuu:wpaper:hsc1501&r=for |
| By: | Baumeister, Christiane; Kilian, Lutz; Lee, Thomas K |
| Abstract: | Although there is much interest in the future retail price of gasoline among consumers, industry analysts, and policymakers, it is widely believed that changes in the price of gasoline are essentially unforecastable given publicly available information. We explore a range of new forecasting approaches for the retail price of gasoline and compare their accuracy with the no-change forecast. Our key finding is that substantial reductions in the mean-squared prediction error (MSPE) of gasoline price forecasts are feasible in real time at horizons up to two years, as are substantial increases in directional accuracy. The most accurate individual model is a VAR(1) model for real retail gasoline and Brent crude oil prices. Even greater reductions in MSPEs are possible by constructing a pooled forecast that assigns equal weight to five of the most successful forecasting models. Pooled forecasts have lower MSPE than the EIA gasoline price forecasts and the gasoline price expectations in the Michigan Survey of Consumers. We also show that as much as 39% of the decline in gas prices between June and December 2014 was predictable. |
| Keywords: | Brent; Expert forecasts; Forecast combination; Oil market; Real-time data; Retail gasoline price; Survey expectations; WTI |
| JEL: | C53 Q43 |
| Date: | 2015–01 |
| URL: | http://d.repec.org/n?u=RePEc:cpr:ceprdp:10362&r=for |
| By: | Oscar Claveria (Department of Econometrics. University of Barcelona); Enric Monte (Department of Signal Theory and Communications. Polytechnic University of Catalunya.); Salvador Torra (Department of Econometrics & Riskcenter-IREA. Universitat de Barcelona) |
| Abstract: | This study attempts to assess the forecasting accuracy of Support Vector Regression (SVR) with regard to other Artificial Intelligence techniques based on statistical learning. We use two different neural networks and three SVR models that differ by the type of kernel used. We focus on international tourism demand to all seventeen regions of Spain. The SVR with a Gaussian kernel shows the best forecasting performance. The best predictions are obtained for longer forecast horizons, which suggest the suitability of machine learning techniques for medium and long term forecasting. |
| Keywords: | Forecasting, support vector regressions, artificial neural networks, tourism demand, Spain JEL classification: C02, C22, C45, C63, E27, R11 |
| Date: | 2015–01 |
| URL: | http://d.repec.org/n?u=RePEc:aqr:wpaper:201506&r=for |
| By: | Anton Bekkerman (Montana State University) |
| Abstract: | While nearly instantaneous commodity futures price information provides price forecasts for national markets, many market participants are interested in forecasting local cash prices. Expected basis estimates are often used to convert futures prices into local price forecasts. This study considers basis patterns in the northern U.S. hard red spring and hard red winter wheat markets. Using data on basis values across 215 grain handling facilities, we empirically test the forecasting capabilities of numerous basis models. Contrary to basis models developed for other U.S. regions, recent futures prices, protein content, and harvest information are more important for accurate basis forecasts than historical basis averages. |
| Keywords: | basis, forecast, protein, spring wheat, winter wheat |
| JEL: | Q13 Q11 L11 |
| Date: | 2014–09–11 |
| URL: | http://d.repec.org/n?u=RePEc:mnu:wpaper:5001&r=for |
| By: | Barnes, Michelle L. (Federal Reserve Bank of Boston); Olivei, Giovanni P. (Federal Reserve Bank of Boston) |
| Abstract: | The widely studied Reuters/Michigan Index of Consumer Sentiment is constructed from the answers to five questions from the more comprehensive Reuters/Michigan Surveys of Consumers. Yet little work has been done on what predictive power the information taken from this more thorough compilation of consumer attitudes and expectations may have for forecasting consumption expenditures. The authors construct a limited set of real-time summary measures for 42 questions selected from these broader Surveys corresponding to three broad economic determinants of consumption—income and wealth, prices, and interest rates, and then use regression analysis to evaluate and test the ability of these summary measures to predict future changes in real consumer expenditures, even when controlling for current and future fundamentals. They explain a nontrivial portion of consumption and other real activity forecast errors from professional forecasts. This is consistent with these measures' ability to predict consumption even when conditioning on a broader set of fundamentals as well as professional forecasters' judgmental forecast adjustments. |
| JEL: | E21 E27 E52 E66 |
| Date: | 2014–10–30 |
| URL: | http://d.repec.org/n?u=RePEc:fip:fedbwp:14-10&r=for |
| By: | Bentour, El Mostafa |
| Abstract: | This paper ranks economic forecasts performances for two structural models against a benchmark of time series models, VAR and ARIMA, according to a set of statistical measures calculated for the main economic aggregates. The period of analysis covers twenty years for annual data (1985-2004) and 28 quarters for quarterly models (1998:1-2004:4). Furthermore, models are tested to see whether predictions contain additional information more than the one showed by a random walk process (Fair-Shiller, 1987). Results show a net supremacy of VAR models over structural models and have significant contribution to information than the one contained in the random walk process. |
| Keywords: | Random Walk, Structural models, Theil Criterion, VAR models |
| JEL: | C18 C32 C53 |
| Date: | 2015–01–15 |
| URL: | http://d.repec.org/n?u=RePEc:pra:mprapa:61502&r=for |
| By: | Flodberg, Caroline (National Institute of Economic Research); Österholm, Pär (National Institute of Economic Research) |
| Abstract: | In this paper, we study revisions of Swedish national accounts data. Three aspects of the revisions are considered: volatility, unbiasedness and forecast efficiency. Our results indicate that the properties of the revisions are more problematic for the production side than for the expenditure side. The high volatility of the revisions on the production side indicates that it, based on the initial data release, generally is difficult to make clear cut statements concerning production in different industries within the business sector; it is also likely to make forecasting more difficult. Concerning unbiasedness, there appears to be shortcomings for a number of variables, including GDP; this finding implies that it could be possible to improve the production of the Swedish national accounts data. |
| Keywords: | Real-time data; Volatility; Unbiasedness; Forecast efficiency |
| JEL: | E01 |
| Date: | 2015–02–04 |
| URL: | http://d.repec.org/n?u=RePEc:hhs:nierwp:0136&r=for |
| By: | Nikolay Klemashev; Alexander Shananin |
| Abstract: | This paper is devoted to revealed preference theory and its applications to testing economic data for consistency with utility maximization hypothesis, construction of index numbers, and forecasting. The quantitative measures of inconsistency of economic data with utility maximization behavior are also discussed. The structure of the paper is based on comparison between the two tests of revealed preference theory - generalized axiom of revealed preference (GARP) and homothetic axiom of revealed prefernce (HARP). We do this comparison both theoretically and empirically. In particular we assess empirically the power of these tests for consistency with maximization behavior and the size of forecasting sets based on them. For the forecasting problem we show that when using HARP there is an effective way of building the forecasting set since this set is given by the solution of the system of linear inequalities. The paper also touches upon the question of testing a set of Engel curves rather than finite set of observations for consistency with utility maximization behavior and shows that this question has effective solution when we require the rationalizing utility function to be positively homogeneous. |
| Date: | 2015–01 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:1501.05771&r=for |
| By: | Mohammadi, Ali (CESIS - Centre of Excellence for Science and Innovation Studies, Royal Institute of Technology); Basir, Nada O. (Faculty of Business and IT, University of Ontario Institute of Technology); Beyhaghi, Mehdi (College of Business,University of Texas at San Antonio) |
| Abstract: | In this paper, we study how R&D investment affect financial analyst’s earnings forecasts and how intellectual capital endowments moderate this effect. We argue that high information asymmetry and uncertainty associated with R&D investment increase a financial analysts’ earnings forecast error. Patents can remedy this relationship by signaling the ability of a firm in transforming research investments into new and valuable knowledge. Using a panel of 2,253 publicly listed U.S firms, we find that higher R & D intensity is positively correlated with financial analysts’ earnings forecast error. The endowment of intellectual capital (i.e. patents) moderates this relationship negatively. However we do not find any moderating effect for the value of patents measured as forward citations. |
| Keywords: | R&D intensity; Analyst forecasts; Patent; information asymmetry; uncertainty; Capital market |
| JEL: | G24 O32 O34 |
| Date: | 2015–02–03 |
| URL: | http://d.repec.org/n?u=RePEc:hhs:cesisp:0397&r=for |