Page 278 - Special Topic Session (STS) - Volume 3
P. 278
STS544 Baoline C. et al.
GDP, with PCE services accounting for roughly two thirds of PCE. Because the
data used to compile PCE goods and services come from different sources,
compilation of PCE goods and services present different challenges to the
national accounts. As delays in the quarterly source data affect the first two
estimates of PCE services similarly, in this study, we focus on the first or the
advance estimate of detailed PCE services.
There are four major reasons for revisions in the advance estimate of PCE
services: 1) source data used to compile the three estimates are from
difference sources and become available at different frequencies; 2) there are
an insufficient number of relevant monthly indicators available for all detailed
components; 3) there is lack of close correlation between the monthly and
quarterly indicators; and 4) information included in the current extrapolation
method is insufficient to reflect the longer-term dynamics of the quarterly PCE
service component series and the monthly indicators.
Figure 1: Revision in advance estimate of selected PCE services components
Given delays in the quarterly source data available for compiling early
estimates, reducing revisions in the early estimates hinges on producing them
more accurately using all available information for the quarter that has just
ended. This amounts to a nowcasting problem, which is defined as the
prediction of the present, the very near future and the very recent past
(Bańbura, Giannone, Modugno and Reichlin, 2013). The basic principle of
nowcasting is to exploit information published early and possibly at higher
frequencies than the target variable in order to obtain a more accurate early
estimate before official estimate based on quarterly source data becomes
available.
The objective of this study is to introduce two nowcasting approaches to
compile advance estimates of detailed PCE services—the bridge equation (GB)
framework and the bridging with factors (BF) model. These approaches are
capable of exploiting information on the longer-term dynamics of PCE services
as well as the short-term movements in the monthly indicators driven by the
changes in the economic conditions. We apply these approaches to compiling
267 | I S I W S C 2 0 1 9