Forecasting papers 2009-03-01

In this issue we have: Variable Selection and Inference for Multi-period Forecasting Problems ;  Forecasting Exchange Rates with a Large Bayesian VAR ; Path Forecast Evaluation ; Pooling versus Model Selection for Nowcasting with Many Predictors ; and more.

  • Variable Selection and Inference for Multi-period Forecasting Problems
    Date: 2009-01
    By: Pesaran, M Hashem
    Pick, Andreas
    Timmermann, Allan G
    This paper conducts a broad-based comparison of iterated and direct multi-step forecasting approaches applied to both univariate and multivariate models. Theoretical results and Monte Carlo simulations suggest that iterated forecasts dominate direct forecasts when estimation error is a first-order concern, i.e. in small samples and for long forecast horizons. Conversely, direct forecasts may dominate in the presence of dynamic model misspecification. Empirical analysis of the set of 170 variables studied by Marcellino, Stock and Watson (2006) shows that multivariate information, introduced through a parsimonious factor-augmented vector autoregression approach, improves forecasting performance for many variables, particularly at short horizons.
    Keywords: factor-augmented VAR; forecast horizon; macroeconomic forecasting
    JEL: C53 E27
  • Forecasting Exchange Rates with a Large Bayesian VAR
    Date: 2008-10
    By: Carriero, Andrea
    Kapetanios, George
    Marcellino, Massimiliano
    Models based on economic theory have serious problems at forecasting exchange rates better than simple univariate driftless random walk models, especially at short horizons. Multivariate time series models suffer from the same problem. In this paper, we propose to forecast exchange rates with a large Bayesian VAR (BVAR), using a panel of 33 exchange rates vis-a-vis the US Dollar. Since exchange rates tend to co-move, the use of a large set of them can contain useful information for forecasting. In addition, we adopt a driftless random walk prior, so that cross-dynamics matter for forecasting only if there is strong evidence of them in the data. We produce forecasts for all the 33 exchange rates in the panel, and show that our model produces systematically better forecasts than a random walk for most of the countries, and at any forecast horizon, including at 1-step ahead.
    Keywords: Bayesian VAR; Exchange Rates; Forecasting
    JEL: C11 C53 F31
    Date: 2009-02
    By: Rangan Gupta (Department of Economics, University of Pretoria)
    Alain Kabundi (Department of Economics and Econometrics, University of Johannesburg)
    This paper analyzes the ability of principal component regressions and Bayesian regression methods under Gaussian and double-exponential prior in forecasting the real house price of the United States (US), based on a monthly dataset of 112 macroeconomic variables. Using an in-sample period of 1992:01 to 2000:12, Bayesian regressions are used to forecast real US house prices at the twelve-months-ahead forecast horizon over the out-of-sample period of 2001:01 to 2004:10. In terms of the Mean Square Forecast Errors (MSFEs), our results indicate that a principal component regression with only one factor is best-suited for forecasting the real US house price. Amongst the Bayesian models, the regression based on the double exponential prior outperforms the model with Gaussian assumptions.
    Keywords: Bayesian Regressions, Principal Components, Large-Cross Sections
    JEL: C11 C13 C33 C53
  • Path Forecast Evaluation
    Date: 2008-10
    By: Jordà, Òscar
    Marcellino, Massimiliano
    A path forecast refers to the sequence of forecasts 1 to H periods into the future. A summary of the range of possible paths the predicted variable may follow for a given confidence level requires construction of simultaneous confidence regions that adjust for any covariance between the elements of the path forecast. This paper shows how to construct such regions with the joint predictive density and Scheffé's (1953) S-method. In addition, the joint predictive density can be used to construct simple statistics to evaluate the local internal consistency of a forecasting exercise of a system of variables. Monte Carlo simulations demonstrate that these simultaneous confidence regions provide approximately correct coverage in situations where traditional error bands, based on the collection of marginal predictive densities for each horizon, are vastly off mark. The paper showcases these methods with an application to the mo! st recent monetary episode of interest rate hikes in the U.S. macroeconomy.
    Keywords: error bands.; path forecast; simultaneous confidence region
    JEL: C32 C52 C53
  • Pooling versus Model Selection for Nowcasting with Many Predictors: An Application to German GDP
    Date: 2009
    By: Vladimir Kuzin
    Massimiliano Marcellino
    Christian Schumacher
    This paper discusses pooling versus model selection for now- and forecasting in the presence of model uncertainty with large, unbalanced datasets. Empirically, unbalanced data is pervasive in economics and typically due to di¤erent sampling frequencies and publication delays. Two model classes suited in this context are factor models based on large datasets and mixed-data sampling (MIDAS) regressions with few predictors. The specification of these models requires several choices related to, amongst others, the factor estimation method and the number of factors, lag length and indicator selection. Thus, there are many sources of mis-specification when selecting a particular model, and an alternative could be pooling over a large set of models with di¤erent specifications. We evaluate the relative performance of pooling and model selection for now- and forecasting quarterly German GDP, a key macroeconomic indicator for t! he largest country in the euro area, with a large set of about one hundred monthly indicators. Our empirical findings provide strong support for pooling over many speci.cations rather than selecting a specific model.
    Keywords: nowcasting, forecast combination, forecast pooling, model selection, mixed-frequency data, factor models, MIDAS
    JEL: E37 C53
  • Some Issues in Modeling and Forecasting Inflation in South Africa
    Date: 2009-02
    By: Aron, Janine
    Muellbauer, John
    Inflation targeting central banks will be hampered without good models to assist them to be forward-looking. Many current inflation models fail to forecast turning points adequately, because they miss key underlying long-run influences. The world is on the cusp of a dramatic turning point in inflation. If inflation falls rapidly, such models can underestimate the speed at which interest rates should fall, damaging growth. Our forecasting models for the new measure of producer price inflation suggest methodological lessons, and build in conflicting pressures on SA inflation from exchange rate depreciation, terms of trade shocks, collapsing oil, food and other commodity prices, and other shocks. Our US and SA forecasting models for consumer price inflation underline the methodological points, and suggest the usefulness of thinking about sectoral trends. Finally, we apply the sectoral approach to understanding the monetary ! policy implications of introducing a new CPI measure in SA that uses imputed rents rather than interest rates to capture housing costs.
    Keywords: forecasting inflation; homeowner costs in the CPI; PPI inflation; South Africa
    JEL: C22 C51 C52 C53 E31 E52 E58
  • On the Statistical Identification of DSGE Models
    Date: 2009-02
    By: Consolo, Agostino
    Favero, Carlo A
    Paccagnini, Alessina
    Dynamic Stochastic General Equilibrium (DSGE) models are now considered attractive by the profession not only from the theoretical perspective but also from an empirical standpoint. As a consequence of this development, methods for diagnosing the fit of these models are being proposed and implemented. In this article we illustrate how the concept of statistical identification, that was introduced and used by Spanos(1990) to criticize traditional evaluation methods of Cowles Commission models, could be relevant for DSGE models. We conclude that the recently proposed model evaluation method, based on the DSGE-VAR(ë), might not satisfy the condition for statistical identification. However, our application also shows that the adoption of a FAVAR as a statistically identified benchmark leaves unaltered the support of the data for the DSGE model and that a DSGE-FAVAR can be an optimal forecasting model.
    Keywords: Bayesian analysis; Dynamic stochastic general equilibrium model; Factor-Augmented Vector Autoregression; Model evaluation
    JEL: C11 C52
  • Taken from the NEP-FOR mailing list edited by Rob Hyndman.