Page 279 - Special Topic Session (STS) - Volume 3

STS544 Baoline C. et al.
advance estimates of detailed PCE services with the objective of reducing
revisions when quarterly source data become available. We measure
improvements in accuracy in terms of reduction in the root mean squared
revision (RMSR) for each detailed component.
The plan for the paper is as follows: Section 2 describes the current
extrapolation method and the proposed bridge equation and bridging with
factor frameworks for nowcasting the advance estimate. Section 3 describes
the application and the strategy for estimation and nowcasting. Section 4
reports estimation results. Section 5 discusses further research and concludes
the paper.

2. Current and Alternative Methods for Compiling Advance Estimate of
Detailed PCE Services
Currently, the U.S. national accounts compile advance estimate of PCE
services using a simple extrapolation method which takes two steps: 1)
extrapolating monthly estimates from the previous month for each of the
three months in the quarter using monthly indicators; and 2) computing the
advance quarterly estimate as the quarterly averages of the three monthly
estimates. The current extrapolation method uses information on the monthly
indicators for the current quarter. However, it does not utilize information on
the longer-term dynamics of the PCE services, nor does it utilize information
on the longer-term dynamics of the monthly indicators. To reduce revisions in
the advance estimate of PCE services, we need to allow lagged information on
the quarterly PCE services and the current and lagged information on the
relevant monthly indicators to be included in the estimation. Since compilation
of advance estimate of PCE services is equivalent to a nowcasting problem, we
consider two widely-used nowcasting techniques in this study, the bridge
equation framework and the bridging with factors model.
The bridge equation approach was first developed by Klein et al. (1989)
and has since been further developed and implemented in many studies
(Kitchen and Manaco, 2003; and Higgins, 2014). It has been described as a
tracking model which tracks quarterly growth in real GDP by tracking the
arrival of new information in real time.
In this study, we express variables in growth rates. Let denote the

quarterly growth rate of the advance estimate of a detailed PCE services
component in quarter t; and let ̅ = (̅ 1,, … , ̅ ) denote the quarterly growth
,

rates of the quarterly averages of s monthly indicator variables. For each
detailed PCE service component, the general bridge equation framework can
be expressed as

268 | I S I W S C 2 0 1 9