|
on Forecasting |
| By: | Bidong Liu; Jiali Liu; Tao Hong |
| Abstract: | This paper introduces the concept of sister models, and proposes a sister model based load forecast combination method to enhance the point forecasting accuracy. Using the data from the Global Energy Forecasting Competition 2014, we create a case study with 4 sister forecasts from 4 sister models. The result shows that the simple average of the 4 sister forecasts yields lower error than each individual forecast by 2% to 10%. |
| Keywords: | Electric load forecasting; Forecast combination; Sister forecast |
| JEL: | C22 C32 C53 Q47 |
| Date: | 2015–02–05 |
| URL: | http://d.repec.org/n?u=RePEc:wuu:wpaper:hsc1502&r=for |
| By: | Zeng, Jing |
| Abstract: | Including disaggregate variables or using information extracted from the disaggregate variables into a forecasting model for an eco- nomic aggregate may improve the forecasting accuracy. In this paper we suggest to use boosting as a method to select the disaggregate variables which are most helpful in predicting an aggregate of interest. We compare this method with the direct forecast of the aggregate, a forecast which aggregates the disaggregate forecasts and a direct forecast which additionally uses information from factors obtained from the disaggregate components. A recursive pseudo-out-of-sample forecasting experiment for key Euro area macroeconomic variables is conducted. The results suggest that using boosting to select relevant predictors is a viable and competitive approach in forecasting an aggregate. |
| JEL: | C43 C53 C22 |
| Date: | 2014 |
| URL: | http://d.repec.org/n?u=RePEc:zbw:vfsc14:100310&r=for |
| By: | Michalski , Tomasz; Amat , Christophe |
| Abstract: | Simple exchange rate models based on economic fundamentals were shown to have a difficulty in beating the random walk when predicting the exchange rates out of sample in the modern floating era. Using methods from machine learning -- sequential adaptive ridge regression -- that prevent overfitting in-sample for better and more stable forecasting performance out-of-sample the authors show that fundamentals from the PPP, UIRP and monetary models consistently improve the accuracy of exchange rate forecasts for major currencies over the floating period era 1973-2013 and are able to beat the random walk prediction giving up to 5% improvements in terms of the RMSE at a 1 month forecast. "Classic'' fundamentals hence contain useful information about exchange rates even for short forecasting horizons -- and the Meese and Rogoff (1983) puzzle is overturned. Such conclusions cannot be obtained when rolling or recursive OLS regressions are used as is common in the literature. |
| Keywords: | exchange rates; forecasting; machine learning; purchasing power parity; uncovered interest rate parity; monetary exchange rate models |
| JEL: | C53 F31 F37 |
| Date: | 2014–08–29 |
| URL: | http://d.repec.org/n?u=RePEc:ebg:heccah:1049&r=for |
| By: | Molenaars, Tomas K.; Reinerink, Nick H.; Hemminga, Marcus A. |
| Abstract: | We define a parameter representing the relative forecast performance to compare forecasting results of different methods. By using this parameter, we analyze the performance of the dynamic Nelson-Siegel model and, for comparison, the first order autoregressive (AR(1)) model applied to a set of US bond yield data that covers a time span from November 1971 to December 2008. As a reference, we take the random walk model applied to the yield data. Our findings indicate that none of the models can convincingly beat the random walk model. Furthermore, there is no advantage in using the more advanced and complicated dynamic Nelson-Siegel model over a simple AR(1) model. |
| Keywords: | Term structure of interest rates; Yield curve modeling; Dynamic Nelson-Siegel model; Out-of-sample forecasting evaluations. |
| JEL: | C5 E4 G17 |
| Date: | 2013–07–11 |
| URL: | http://d.repec.org/n?u=RePEc:pra:mprapa:61862&r=for |
| By: | Mihaela Simionescu (Institute for Economic Forecasting, Romanian Academy, Bucharest) |
| Abstract: | The necessity of improving the forecasts accuracy grew in the context of actual economic crisis, but few researchers were interested till now in finding out some empirical strategies to improve their predictions. In this article, for the inflation rate forecasts on the horizon 2010-2012, we proved that the one-step-ahead forecasts based on updated AR(2) models could be substantially improved by generating new predictions using Monte Carlo method and bootstrap technique to simulate the models’ coefficients. In this article we introduced a new methodology of constructing the forecasts, by using the limits of the bias-corrected-accelerated bootstrap intervals for the initial data series of the variable to predict. After evaluating the accuracy of the new forecasts, we found out that all the proposed strategies improved the initial AR(2) forecasts and even the predictions of two experts in forecasting. Our own method based on the lower limits of BCA intervals generated the best forecasts. In the forecasting process based on AR models the uncertainty analysis was introduced, by calculating, under the hypothesis of normal distribution, the probability that the predicted value exceeds a critical value. |
| Keywords: | accuracy, forecasts, Monte Carlo method, bootstrap technique, biased-corrected-accelerated bootstrap intervals |
| JEL: | C15 C53 |
| Date: | 2015–02 |
| URL: | http://d.repec.org/n?u=RePEc:rjr:wpiecf:150206&r=for |
| By: | Molenaars, Tomas K.; Reinerink, Nick H.; Hemminga, Marcus A. |
| Abstract: | The objective of our work is to analyze the forecast performance of the dynamic Nelson-Siegel yield curve model and, for comparison, the first order autoregressive (AR(1)) model applied to a set of US bond yield data that covers a large timespan from November 1971 to December 2008. As a reference we take the random walk model applied to the yield data. For our analysis, we make use of a simple parameter representing the relative forecast performance to compare forecasting results of different methods. Our findings indicate that none of the yield curve models convincingly beats the random walk model. Furthermore, our results show that deriving conclusions on basis of model testing for a limited time period is inadequate. |
| Keywords: | Term structure of interest rates; Yield curve modeling; Dynamic Nelson-Siegel model; Out-of-sample forecasting evaluations. |
| JEL: | C5 E4 G17 |
| Date: | 2015–02–05 |
| URL: | http://d.repec.org/n?u=RePEc:pra:mprapa:61917&r=for |
| By: | Wohlrabe, Klaus; Teresa, Buchen |
| Abstract: | The use of large datasets for macroeconomic forecasting has received a great deal of interest recently. Boosting is one possible method of using high-dimensional data for this purpose. It is a stage-wise additive modelling procedure, which, in a linear specification, becomes a variable selection device that iteratively adds the predictors with the largest contribution to the fit. Using data for the United States, the euro area and Germany, we assess the performance of boosting when forecasting a wide range of macroeconomic variables. Moreover, we analyse to what extent its forecasting accuracy depends on the method used for determining its key regularisation parameter, the number of iterations. We find that boosting mostly outperforms the autoregressive benchmark, and that $K$-fold cross-validation works much better as stopping criterion than the commonly used information criteria. |
| JEL: | C53 E27 C52 |
| Date: | 2014 |
| URL: | http://d.repec.org/n?u=RePEc:zbw:vfsc14:100626&r=for |
| By: | Drechsel, Katja; Giesen, Sebastian; Lindner, Axel |
| Abstract: | This study analyzes the performance of the IMF World Economic Outlook forecasts for world output and the aggregates of both the advanced economies and the emerging and developing economies. With a focus on the forecast for the current and the next year, we examine whether IMF forecasts can be improved by using leading indicators with monthly updates. Using a real-time dataset for GDP and for the indicators we nd that some simple single-indicator forecasts on the basis of data that are available at higher frequency can signi cantly outperform the IMF forecasts if the publication of the Outlook is only a few months old. |
| JEL: | C52 C53 E37 |
| Date: | 2014 |
| URL: | http://d.repec.org/n?u=RePEc:zbw:vfsc14:100393&r=for |
| By: | Andrés Ramírez Hassan; Jhonatan Cardona Jiménez; Raul Pericchi Guerra |
| Abstract: | In this paper we analyze the effect of four possible alternatives regarding the prior distributions in a linear model with autoregressive errors to predict piped water consumption: Normal-Gamma, Normal-Scaled Beta two, Studentized-Gamma and Student's t-Scaled Beta two. We show the effects of these prior distributions on the posterior distributions under different assumptions associated with the coefficient of variation of prior hyperparameters in a context where there is a conflict between the sample information and the elicited hyperparameters. We show that the posterior parameters are less affected by the prior hyperparameters when the Studentized-Gamma and Student's t-Scaled Beta two models are used. We show that the Normal-Gamma model obtains sensible outcomes in predictions when there is a small sample size. However, this property is lost when the experts overestimate the certainty of their knowledge. In the case that the experts greatly trust their beliefs, it is a good idea to use Student's t distribution as the prior distribution, because we obtain small posterior predictive errors. In addition, we find that the posterior predictive distributions using one of the versions of Student's t as prior are robust to the coefficient of variation of the prior parameters. Finally, it is shown that the Normal-Gamma model has a posterior distribution of the variance concentrated near zero when there is a high level of confidence in the experts' knowledge: this implies a narrow posterior predictive credibility interval, especially using small sample sizes. |
| Keywords: | Autoregressive model, Bayesian analysis, Forecast, Robust prior |
| JEL: | C11 C53 |
| Date: | 2014–07–23 |
| URL: | http://d.repec.org/n?u=RePEc:col:000122:012434&r=for |