|
on Forecasting |
| By: | Konstantins Benkovskis |
| Abstract: | The conjunctural information from monthly indicators, e.g. industrial production, retail trade turnover, M3, confidence indicators, etc. could partly replace GDP data before the first official release is published. It is possible to incorporate monthly indicators into short-term forecasting models of GDP using quarterly bridge equations or state space models. In many cases monthly indicators are released with a lag, and GDP forecasts based on actual figures are available only shortly before the official release. To eliminate this drawback, missing observations of monthly indicators could be forecasted using simple univariate time-series models. To perform real-time analysis of the forecasting performance of bridge equations and state space models, a real-time database containing real GDP series with 28 vintages of quarterly real GDP was created. According to calculations, only bridge equations and state space models containing M3 monthly data perform better than the benchmark ARIMA model. Both model types using M3 provide valuable information forecast for the first and final releases of GDP. This does not mean, however, that other conjunctural indicators should not be used in forecasting, as the analysis does not take into account possible future changes in links between monthly indicators and quarterly GDP growth. |
| Keywords: | bridge equations, state space model, out-of-sample forecasting, real-time database, interpolation |
| JEL: | C22 C53 E37 |
| Date: | 2008–09–15 |
| URL: | http://d.repec.org/n?u=RePEc:ltv:wpaper:200805&r=for |
| By: | Viktors Ajevskis; Gundars Davidsons |
| Abstract: | The study aims at evaluating how useful the application of models using large panels of data in forecasting Latvia's GDP is. Two factor models have been used: the Stock-Watson factor model and the generalised dynamic factor model. The forecast findings by the two models have been compared with the results obtained by the benchmark autoregressive model. The results suggest that compared with simpler autoregressive models both the Stock-Watson factor model and the generalised dynamic factor model ensure forecast improvement, which, however, has not been statistically significant if statistical tests are used. |
| Keywords: | forecasting, factor models, large cross section |
| JEL: | C32 C33 E53 |
| Date: | 2008–04–29 |
| URL: | http://d.repec.org/n?u=RePEc:ltv:wpaper:200802&r=for |
| By: | Andrés González (Banco de la República, Bogotá and CREATES, University of Aarhus, Denmark); Kirstin Hubrich (European Central Bank, Frankfurt am Main and CREATES, University of Aarhus, Denmark); Timo Teräsvirta (CREATES, University of Aarhus, Denmark) |
| Abstract: | In this work, we make use of the shifting-mean autoregressive model which is a flexible univariate nonstationary model. It is suitable for describing characteristic features in inflation series as well as for medium-term forecasting. With this model we decompose the inflation process into a slowly moving nonstationary component and dynamic short-run fluctuations around it. We fit the model to the monthly euro area, UK and US inflation series. An important feature of our model is that it provides a way of combining the information in the sample and the a priori information about the quantity to be forecast to form a single inflation forecast. We show, both theoretically and by simulations, how this is done by using the penalised likelihood in the estimation of model parameters. In forecasting inflation, the central bank inflation target, if it exists, is a natural example of such prior information. We further illustrate the application of our method by an ex post forecasting experiment for euro area and UK inflation. We find that that taking the exogenous information into account does im- prove the forecast accuracy compared to that of a linear autoregressive benchmark model. |
| Keywords: | Nonlinear forecast, nonlinear model, nonlinear trend, penalised likelihood, structural shift, time-varying parameter |
| JEL: | C22 C52 C53 E31 E47 |
| Date: | 2009–01–28 |
| URL: | http://d.repec.org/n?u=RePEc:aah:create:2009-03&r=for |
| By: | Eliana González; Luis F. Melo; Viviana Monroy; Brayan Rojas |
| Abstract: | ABSTRACT. We use a dynamic factor model proposed by Stock and Watson [1998, 1999, 2002a,b] to forecast Colombian inflation. The model includes 92 monthly series observed over the period 1999:01-2008:06. The results show that for short-run horizons, factor model forecasts significantly outperformed the auto-regressive benchmark model in terms of the root mean squared forecast error statistic. |
| Date: | 2009–02–09 |
| URL: | http://d.repec.org/n?u=RePEc:col:000094:005273&r=for |
| By: | Konstantin A. Kholodilin; Boriss Siliverstovs |
| Abstract: | The paper evaluates the quality of the German national accounting data (GDP and its use-side components) as measured by the magnitude and dispersion of the forecast/ revision errors. It is demonstrated that government consumption series are the least reliable, whereas real GDP and real private consumption data are the most reliable. In addition, early forecasts of GDP, private consumption, and investment growth rates are shown to be systematically upward biased. Finally, early forecasts of all the variables seem to be no more accurate than naïve forecasts based on the historical mean of the final data. |
| Keywords: | Quality of statistical data, real-time data, signal-to-noise ratio, forecasts, revisions |
| JEL: | C53 C89 |
| Date: | 2009 |
| URL: | http://d.repec.org/n?u=RePEc:diw:diwwpp:dp858&r=for |
| By: | Zagaglia, Paolo (Dept. of Economics, Stockholm University) |
| Abstract: | I study the dynamics of oil futures prices in the NYMEX using a large panel dataset that includes global macroeconomic indicators, financial market indices, quantities and prices of energy products. I extract common factors from these series and estimate a Factor-Augmented Vector Autoregression for the maturity structure of oil futures prices. I find that latent factors generate information that, once combined with that of the yields, improves the forecasting performance for oil prices. Furthermore, I show that a factor correlated to purely financial developments contributes to the model performance, in addition to factors related to energy quantities and prices. |
| Keywords: | Crude Oil; Futures Markets; Factor Models |
| JEL: | C53 D51 E52 |
| Date: | 2009–02–10 |
| URL: | http://d.repec.org/n?u=RePEc:hhs:sunrpe:2009_0007&r=for |
| By: | Gerard van Welzenis |
| Abstract: | The financial crisis is still spreading and deepening and is making inroads on the real economy. Forecasting is always "work in progres" but the exceptional situation today makes it feel like shooting at a moving target from a ship on the high seas. The forecasts of the Working Group were prepared at the beginning of October, starting off from data supplied by member institutes, prepared somewhere in the third quarter. We already took the liberty during our meeting to make bigger than usual adjustments to these national trade forecasts, to get closer to the rapidly deteriorating situation in the world economy. But it is clear that our efforts fell short of what we are experiencing today. <br> Given the uncertainties, the Working Group did not think it opportune to prepare a completely new detailed trade forecast. Instead we present the data prepared early October, which already incorporate strong downward shifts in the overall outlook, but give ample attention to the current trends and risks. We also provide a model simulation estimating the possible effects on trade variables of the rapidly declining short-term growth outlook for the world economy, the weaker euro and the lower oil and other commodity prices. In the original forecast, world trade growth decelerated from 7.4% in 2007 via 5.0% in 2008 to 2.5% next year. The simulation suggests that world trade levels might actually decline in 2009. |
| Keywords: | International trade; forecast |
| Date: | 2008–12 |
| URL: | http://d.repec.org/n?u=RePEc:cpb:docmnt:176&r=for |
| By: | Athanasopoulos, George; Issler, João Victor; Guillén, Osmani Teixeira de Carvalho; Farshid, Vahid |
| Abstract: | We study the joint determination of the lag length, the dimension of the cointegrating space and the rank of the matrix of short-run parameters of a vector autoregressive (VAR) model using model selection criteria. We consider model selection criteria which have data-dependent penalties for a lack of parsimony, as well as the traditional ones. We suggest a new procedure which is a hybrid of traditional criteria and criteria with data-dependant penalties. In order to compute the fit of each model, we propose an iterative procedure to compute the maximum likelihood estimates of parameters of a VAR model with short-run and long-run restrictions. Our Monte Carlo simulations measure the improvements in forecasting accuracy that can arise from the joint determination of lag-length and rank, relative to the commonly used procedure of selecting the lag-length only and then testing for cointegration. |
| Date: | 2009–02–05 |
| URL: | http://d.repec.org/n?u=RePEc:fgv:epgewp:689&r=for |