Forecast Cycle Research Articles

Mind the Gap: What is the Best Measure of Slack in the Euro Area?

Assessing the magnitude of the output gap is critical to achieving an optimal policy mix. Unfortunately, the gap is an unobservable variable, which, in practice, has been estimated in a variety of ways, depending on the preferences of the modeler. This model selection problem leads to a substantial degree of uncertainty regarding the magnitude of the output gap, which can reduce its usefulness as a policy tool. To overcome this problem, in this paper we attempt to insert some discipline into this search by providing two metrics-inflation forecasting and business cycle dating-against which different options can be evaluated using aggregated euro-area GDP data. Our results suggest that Gali, Gertler, and Lopez-Salido's (2001) inefficiency wedge performs best in inflation forecasting and production function methodology dominates in the prediction of turning points. If, however, a unique methodology must be selected, the quadratic trend delivers the best overall results.

Mind the Gap: What is the Best Measure of Slack in the Euro Area?

Assessing the magnitude of the output gap is critical to achieving an optimal policy mix. Unfortunately, the gap is an unobservable variable, which, in practice, has been estimated in a variety of ways, depending on the preferences of the modeler. This model selection problem leads to a substantial degree of uncertainty regarding the magnitude of the output gap, which can reduce its usefulness as a policy tool. To overcome this problem, in this paper we attempt to insert some discipline into this search by providing two metrics-inflation forecasting and business cycle dating-against which different options can be evaluated using aggregated euro-area GDP data. Our results suggest that Gali, Gertler, and Lopez-Salido's (2001) inefficiency wedge performs best in inflation forecasting and production function methodology dominates in the prediction of turning points. If, however, a unique methodology must be selected, the quadratic trend delivers the best overall results.

Analyst Compensation and Forecast Bias

In a recent paper, Bernhardt et al. (2004) developed a non-parametric test for bias in forecasts by professional financial analysts that is robust to correlated information amongst analysts and information arrival over the forecasting cycle. The tests show that analysts anti-herd: Analysts systematically issue biased contrarian forecasts that overshoot the publicly-available consensus forecast in the direction of their private information. In this campanion paper, we show that for those analysts that report later in the forecast-horizon, a reward scheme that is convex in relative performance may shed some light on this strategic behavior. The pattern and magnitude of the forecast bias in the last forecast are identical to the results in Bernhardt et al., and slightly higher in some sub-samples. An analysis of daily returns around the date of earnings announcement reveals that investors do not fully unravel the bias in late forecasts.

An Updated Statistical Hurricane Intensity Prediction Scheme (SHIPS) for the Atlantic and Eastern North Pacific Basins

Abstract Updates to the Statistical Hurricane Intensity Prediction Scheme (SHIPS) for the Atlantic basin are described. SHIPS combines climatological, persistence, and synoptic predictors to forecast intensity changes using a multiple regression technique. The original version of the model was developed for the Atlantic basin and was run in near–real time at the Hurricane Research Division beginning in 1993. In 1996, the model was incorporated into the National Hurricane Center operational forecast cycle, and a version was developed for the eastern North Pacific basin. Analysis of the forecast errors for the period 1993–96 shows that SHIPS had little skill relative to forecasts based upon climatology and persistence. However, SHIPS had significant skill in both the Atlantic and east Pacific basins during the 1997 hurricane season. The regression coefficients for SHIPS were rederived after each hurricane season since 1993 so that the previous season’s forecast cases were included in the sample. Modificatio...

Using forecast sensitivity patterns to improve future forecast skill

A simple, relatively inexpensive technique has been developed for using past forecast errors to improve the future forecast skill. the method uses the forecast model and its adjoint and can be considered as a simplified 4-dimensional variational (4-D VAR) system. One-or two-day forecast errors are used to calculate a small perturbation (sensitivity perturbation) to the analyses that minimizes the forecast error. the longer forecasts started from the corrected initial conditions, although better than the original forecasts, are still significantly worse than the shorter forecasts started from the latest analysis, even though they both had access to information covering the same period. As a much less expensive alternative to 4-D VAR, the adjusted initial conditions from one or two days ago are used as a starting point for a second iteration of the regular NCEP analysis and forecast cycle until the present time (t = O) analysis is reached. Forecast experiments indicate that the new analyses result in improvements to medium-range forecast skill, and suggest that the technique can be used in operations, since it increases the cost of the regular analysis cycle by a maximum factor of about 4 to 8, depending on the length of the analysis cycle that is repeated. Several possible operational configurations are also tested. The model used in these experiments is the NCEP's operational global spectral model with 62 waves triangular truncation and 28 ő-vertical levels. an adiabatic version of the adjoint was modified to make it more consistent with the complete forecast model, including only a few simple physical parametrizations (horizontal diffusion and vertical mixing). This adjoint model was used to compute the gradient of the forecast error with respect to initial conditions.

Using forecast sensitivity patterns to improve future forecast skill

AbstractA simple, relatively inexpensive technique has been developed for using past forecast errors to improve the future forecast skill. the method uses the forecast model and its adjoint and can be considered as a simplified 4‐dimensional variational (4‐D VAR) system. One‐or two‐day forecast errors are used to calculate a small perturbation (sensitivity perturbation) to the analyses that minimizes the forecast error. the longer forecasts started from the corrected initial conditions, although better than the original forecasts, are still significantly worse than the shorter forecasts started from the latest analysis, even though they both had access to information covering the same period.As a much less expensive alternative to 4‐D VAR, the adjusted initial conditions from one or two days ago are used as a starting point for a second iteration of the regular NCEP analysis and forecast cycle until the present time (t = O) analysis is reached. Forecast experiments indicate that the new analyses result in improvements to medium‐range forecast skill, and suggest that the technique can be used in operations, since it increases the cost of the regular analysis cycle by a maximum factor of about 4 to 8, depending on the length of the analysis cycle that is repeated. Several possible operational configurations are also tested.The model used in these experiments is the NCEP's operational global spectral model with 62 waves triangular truncation and 28 ő‐vertical levels. an adiabatic version of the adjoint was modified to make it more consistent with the complete forecast model, including only a few simple physical parametrizations (horizontal diffusion and vertical mixing). This adjoint model was used to compute the gradient of the forecast error with respect to initial conditions.

NWP Experiments with a Gridpoint Semi-Lagrangian Semi-ImplicitGlobal Model at NCEP

Abstract A series of 5-day forecasts for both winter and summer initial conditions have been performed using a fairly high resolution version (0.9375° latitude–longitude grid with 28 vertical levels) of a gridpoint semi-Lagrangian global forecast model with second-order semi-implicit off-centering and a time step of 1800 s. Most of the physics included in the model are taken from the NCEP operational MRF model. The forecasts are evaluated by comparing the anomaly correlations computed using the NCEP operational global analyses with those for the operational MRF forecasts. It is found that the skill scores of the semi-Lagrangian model are fairly close to those of the operational MRF model despite the handicap of not having its own analyses forecast cycle. The forecasts are also used to get an estimate of the climate drift of the model. It is found that the drift is similar to but somewhat larger than that for the NCEP MRF model. In the second part of the paper, a spatially averaged Eulerian treatment of or...

The Dependence of Short-Range 500-mb Height Forecasts on the Initial Flow Regime

Abstract Forecast errors in the 500-mb geopotential height field over North America and adjacent ocean environs are calculated for the National Meteorological Center's Nested Grid Model (NGM). The eight winters 1985/86-1992/93 are examined. Errors are compared for the time-mean flow and for four recurring planetary-scale flow regimes, and the statistical significance of the differences is estimated. Overall, the NGM produces very accurate 500-mb height forecasts out to 48 h, with every forecast cycle of the study period exhibiting useful deterministic skill at 48 h when averaged over the study domain. During the first two winters of operational NGM implementation, the spatially averaged errors were noticeably greater than in subsequent winters. NGM error was essentially constant during the 1987/88-1992/93 winters. The bias in the NGM works to erode the asymmetries associated with the wintertime stationary waves. When the errors are categorized by the initial flow configuration, no significant differences ...

How to suppress a lead

According to the findings of Scholer (1994)1 the “Business Climate” developed by the ifo Institute for Economic Research, Munich, cannot be used as a leading indicator for business cycle forecast. This paper shows that this result is due to Scholer's choice of statistical methods applied: apart from generating stationary time series, Scholer's filter systematically suppresses the business cycle components and amplifies the irregular components. It impairs the basis for the causality test.

NWP-Initialized Satellite Temperature Retrievals Using Statistical Regularization and Singular Value Decomposition Methods

Abstract Several research groups have announced plans to merge satellite profile retrieval methods and numerical weather prediction methods into “interactive” satellite retrieval approaches for both weather and climate-scale endeavors. Satellite profile retrievals, produced from algorithms that depend on hydrodynamic weather prediction models for first-guess and conditioning data, may be expected to contain some influence of the numerical weather prediction (NWP) model quite distinct from any influence of the satellite measurements. Research is described in this paper in which possible adverse impacts of NWP-produced first-guess information on temperature profile retrievals appear to signal danger for interactive methodologies. Deep-layer, synoptically correlated NWP forecast errors influence satellite retrieval errors in such a way that systematic distortions of the hydrostatic and baroclinic character of the resulting fields could lead to degradations of a subsequent forecast cycle rather than improveme...

Business cycle forecasting

AbstractBusiness cycle forecasting involves several different methodological problems. Some of these are discussed in the current issue of this journal and are introduced in this paper. The forecasting approach itself often focuses on turning points in the business cycle and a number of papers in this issue examine this particular aspect of business cycle forecasting. For example, a Bayesian technique for detecting changing random slopes in leading composite indexes is discussed. Business survey data are often used in the process of forecasting business cycles. A Kalman filtering procedure is suggested to make business survey information useful in predicting changes in industrial production. Other data issues of relevance to this topic that are discussed include the preliminary data and revision problem. Methodology for using high‐frequency data and for converting high‐frequency to low‐frequency data is also presented. A number of the papers discuss the analysis of dynamic structures, such as the existence of time‐varying dynamics, and the use of vector‐autoregressive (VAR) models. Finally, a few comments are made on general structural variability aspects, related to business cycle forecasting.

Forecasting cyclical turning points with an index of leading indicators: A probabilistic approach

AbstractIn this paper the econometrics of latent variables in conjunction with leading economic indicators are used to predict turning points of the US industrial production variable for various forecasting horizons. The results reported here show that leading indicators used in regression models with a dichotomous dependent variable that marks periods of economic expansion and contraction forecast business cycle turning points accurately. Leading indicators produce the most reliable forecasts of turning points five months prior to cyclical changes and do not give any false signals. Moreover, the weights assigned to leading indicators for producing the index vary as the forecast horizon changes.

Characteristics of Surface Cyclone Forecasts in the Aviation Run of the Global Spectral Model

The performance of the Aviation Run (AVN) of the National Meteorological Center's (NMC) Global Spectral Model (GSM) in predicting surface cyclones was examined during the autumn of 1990 through the winter of 1992, a period during which the resolution of the model was increased with the implementation of the triangular truncation 126 (T126) GSM, and the GSM analysis scheme was changed. The results indicate that the finer resolution T126 GSM produces stronger and deeper cyclones than the old T80 GSM. These results also revealed that the errors in AVN position forecasts of surface cyclones were smaller than those found in the NMC Nested Grid Model (NGM). The geographical distribution of the pressure errors were similar to those found in the NGM over eastern North America and the adjacent western Atlantic Ocean. Negative pressure errors, indicative of overdeepening of surface cyclones, dominated the mountainous regions of western North America. Positive pressure errors, indicative of underdeepening of surface cyclones, dominated most of the western Atlantic. The AVN tended to underpredict the 1000-500-mb thickness over surface cyclones, especially during the first 36 h of the forecast cycle. This cold bias decreased with forecast length and in the T80 version of the AVN became a warm bias at the later forecast periods during several months. The skill in the AVN, measured by examining the sign of the forecast and observed 12-h pressure changes, the number of nonobserved and nonforecast cyclones, and skill indices revealed that the AVN is superior to the NGM in predicting the development and life cycle of surface cyclones. The AVN is able to forecast the sign of the 12-h pressure change greater than 80% of the time for the first 36 h of the forecast cycle and 74% of the time at 72 h. The results indicate that the T126 AVN performs at a skill comparable to the skill of the 48 h NGM cyclone forecasts. These results imply that the T126 AVN forecasts are accurate enough to provide guidance for basic weather forecasts to three days as has been done for the two-day forecasts for the past 25–30 years.

Recent Maximum Temperature Anomalies at Albany, New York: Fact or Fiction

Abstract An analysis has been conducted of a suspected daily maximum temperature (DMT) bias introduced by the replacement of the National Weather Service (NWS) HO-63 bygrothermograph with a modernized HO-83 instrument at Albany, New York, on 6 February 1985. The analysis involves an assessment of the bias before and after the changeover and has four components: 1) a bias comparison with seven surrounding NWS Cooperative Observer Network (COOP) stations; 2) a comparison of the DMT with the highest reported hourly temperature during the summer; 3) a comparison of 1800 and 2100 UTC model output statistics (MOS) operational temperature forecasts from the 0000 UTC forecast cycle during the summer with the observed temperatures; and 4) a comparison of the reported maximum surface temperature with the observed 850-mb temperature at the time of the surface maximum temperature in July for days with at least 75% of possible sunshine. The results show that the reported DMT at Albany has increased by ∼0.5°C relative ...

Initial Weather Regimes as Predictors of Numerical 30-Day Mean Forecast Accuracy

Thirty 30-day mean 500-mb-height anomaly forecasts generated by the NCAR Community Climate Model (CCM) for the year 1978 are examined in order to determine if the forecast accuracy can be estimated with the initial conditions. The initial weather regimes were defined in such a way that the regimes could discriminate between the best and the worst 30-day mean forecasts run from the initial fields in this data set. On the basis of the CCM experiments, it is suggested that the accuracy of numerical 30-day mean forecasts may depend upon the accuracy with which the cyclones and their interactions with the planetary scale are predicted early in the forecast cycle, and that this accuracy may depend upon the initial conditions.

Evaluation of the National Meteorological Center's Medium Range Forecast Model Precipitation Forecasts

Precipitation forecasts by the National Meteorological Center's medium range forecast (MRF) model are evaluated for the period 1 March 1987 to 30 March 1989. During 1988, the MRF model's tendency to forecast too much precipitation was reduced. However, there is now an amplified diurnal forecast cycle, which may be spurious. Another potentially serious error is that the MRF model tends to have more trace precipitation events than are observed.

Simplification of the Kalman filter for meteorological data assimilation

AbstractWe propose a new statistical method of data assimilation that is based on a simplification of the Kalman filter equations. The forecast error covariance evolution is approximated simply by advecting the mass‐error covariance field, deriving the remaining covariances geostrophically, and accounting for external model‐error forcing only at the end of each forecast cycle. This greatly reduces the cost of computation of the forecast error covariance, which is the central and most expensive aspect of the Kalman filter algorithm. In simulations with a linear, one‐dimensional shallow‐water model and data generated artificially, the performance of the simplified filter is compared with that of the Kalman filter and the optimal interpolation (OI) method. These experiments are designed to isolate the effect of simplifying the evolution of the forecast error covariance. The simplified filter produces analyses that are nearly optimal, and represents a significant improvement over OI.

A Real-Time, Mesoscale Numerical Weather-Prediction System Used for Research, Teaching, and Public Service at The Pennsylvania State University

A mesoscale modeling system is being applied on an experimental basis at The Pennsylvania State University (Penn State) for production of real-time, high resolution, numerical weather forecasts for the northeastern United States. The initial model experimentation is being supported by Penn State. It is believed to be the first time that a real-time, three-dimensional mesoscale model has been run routinely at an American university, although mesoscale models have been run in real time in government laboratories. A version of the Penn State/NCAR mesoscale model is employed, using a two-way interacting nested grid with a fine-grid increment of 30 km, a coarse-grid increment of 90 km, and 15 computational levels. The forecast cycle is initiated automatically by the Department of Meteorology's Digital Equipment Corporation VAX 8350 system when all the required 0000 UTC surface and upper-air National Weather Service (NWS) data have been received, quality checked, and archived. Lateral boundary conditions are ex...

Preliminary Experiments Using GALE Observations at the National Meteorological Center

Analyses and forecasts for the first 2 weeks of the Genesis of Atlantic Lows Experiment (GALE) are described. These fields were produced using the National Meteorological Center (NMC) Regional Analysis and Forecast System (RAFS). Two sets of analyses and forecasts were constructed: one using the NMC operational database only (Level IIIa), and one using the NMC data merged with high-density observations taken during GALE (Level IIIb). During the first 14 days of GALE, supplemental data were collected throughout two Intensive Observing Periods (IOPs). Comparisons of the Level IIIa and IIIb analyses over the GALE observing region in the southeastern United States indicated a worsening of the geopotential height analysis at operational NWS rawinsonde sites using the supplemental IIIb data. This was caused by inconsistencies in the height measurements at the high-density GALE rawinsonde sites. Such patterns were not observed in the wind and temperature analyses. During IOP No. 1, the Level IIIa and IIIb Nested Grid Model (NGM) forecasts were nearly identical. For IOP No. 2, one forecast cycle saw an improvement in the Level IIIb forecasts due to offshore GALE dropwindsonde data, while another was improved by the inclusion of late-arriving rawinsonde data in the IIIb analysis. The inland, high-density GALE soundings, however, had a negligible impact on NGM forecasts during the entire 12-day period.

Short-term systematic errors in global forecasts: their estimation and removal

We have found spatial and temporal coherencies in the departures of short-range-forecasts from observed and analysed geopotential fields, which can be used diagnostically and as the basis for adjustment of forecast products. We identify characteristics of the differences between forecast and observed, and forecast and analysed values, and we demonstrate the significant impact on forecast accuracy which may be gained through the routine estimation and removal of these discrepancies. The removal of these discrepancies in the data assimilation step of the forecast cycle was found to have little impact, in contrast with the effect of direct correction of forecast products in conjunction with post processing. DOI: 10.1034/j.1600-0870.1990.00001.x