Real-Time Representations of the Output Gap

Abstract

Methods are described for the appropriate use of data obtained and analysed in real time to represent the output gap. The methods employ cointegrating VAR techniques to model real time measures and realisations of output series jointly. The model is used to mitigate the impact of data revisions; to generate appropriate forecasts that can deliver economically-meaningful output trends and that can take into account the end-of-sample problems encountered in measuring these trends; and to calculate probability forecasts that convey in a clear way the uncertainties associated with the gap measures. The methods are applied to data for the US 1965q4-2004q4 and the improvements over standard methods are illustrated.

Improving the Reliability of Real-Time Hodrick-Prescott Filtering Using Survey Forecasts

“…Namely, we can expand the information set supplied to the filter by use of forecasts, a solution that has been proved successful in previous literature Maravall, 1999, 2001;Mise et al, 2005;Garratt et al, 2008). Under the same circunstances as in the original formulation, (1), suppose now we have h = 1, .., H forecasts of y T +h at our disposal 2 .…”

“…The most prominent attempts involve expanding the information set supplied to the filter in the temporal dimension, using either model-based forecasts (Kaiser and Maravall, 1999;Gomez, 2001;Kaiser and Maravall, 2001;Mise et al, 2005), or variables observed at a higher frequency (Aruoba et al, 2009); in the covariates dimension with the employment of model-based multivariate representations (Planas and Rossi, 2004; Valle e Azevedo et al, 2006;Altissimo et al, 2010;Valle e Azevedo, 2011;Marcellino and Musso, 2011), or jointly in both of these dimensions (Garratt et al, 2008;Clements and Galvão, 2012).…”

Decision-making in hard times: What is a recession, why do we care and how do we know when we are in one?

“…Figure 5a describes the probability of the event HP (and other) filters in defining gaps and Orphanides and van Norden (2002) showed that these are typically very vulnerable to real-time analysis. However, Garratt et al (2008) showed that the method described above, applying smoothing techniques to forecast-augmented series, can considerably improve the precision of the estimates of an output gap at the end-of-sample. This is, of course, extremely important in realtime decision-making.…”

A State Space Approach to Extracting the Signal From Uncertain Data

“…Subsequent papers have explored a variety of approaches to dealing with the uncertainty surrounding more mature estimates. Some, such as Patterson (1994) and Garratt, Lee, Mise, and Shields (2005), increase the number of releases in the model so that estimates are not assumed to become 'true' for two or three years. In the case of the United Kingdom's National Accounts, however, revisions have been applied to even more mature estimates.…”

“…One natural question is whether data-9 users should consider these previous releases as competing measures of the truth -that is, using y t+n−j t alongside y t+n t as measures of y t . In contrast with the treatment in much of the antecedent literature, we decide to exclude earlier releases from the set of measures used to estimate 'true' activity, see, for example, Garratt, Lee, Mise, and Shields (2005). The reason for using only the latest release is pragmatic.…”