Benchmarking quarterly GDP to other official statistics.

1
Benchmarking quarterly GDP to other official
statistics.

• Andrew Mortimer
• Short-term Economic Indicators Branch

2
Purpose

• To demonstrate a method of constraining quarterly
  short-term indicators to other published
  statistics
• To seek the views of SESCG on
• the need for benchmarking
• what source(s) to benchmark to
• timing

3
Comparison of Quarterly GDP against other sources

• Differences between estimates are inevitable
• Actual (directly measured) values of sectoral GVA
  are available from
• Regional Accounts
• SABS (ABI)
• Scottish Supply-use tables
• Quarterly GVA uses completely different data
  sources
• Uses proxy measures as a short-term indicator of
  GVA in constant prices.
• Is currently the only official measure of real
  growth in the economy and consequently is used
  for long-run analysis

Lagged by a number of years
4
Example Hotels and catering
5
Example Hotels and catering
6
Example Hotels and catering
7
Example Hotels and catering
8
Example Hotels and catering
9
Example Hotels and catering
10
Example Hotels and catering
11
Example Hotels and catering
12
Example Hotels and catering
13
Benefits

• Consistency between official statistics is
  desirable
• Long-run growth measured using actual GVA
• Increased scrutiny of Regional Accounts / other
  official statistics
• Focus on quarterly GDP shifts, appropriately, to
  its intended use i.e. a short-term indicator of
  growth

14
Problems

• Choice of gold standard
• Depending on the source constrained to,
  potentially large revisions on an annual basis
• Scottish Government are involved in the quality
  assurance of regional accounts and have been
  instrumental in highlighting problems in the
  statistics but QA is performed using aggregates,
  not source data loss of control
• Whether or not to anticipate and adjust for
  differences what happens if systematic bias is
  indentified and then cancelled out by new
  observations (or vice-versa)?
• Some component series used in quarterly GDP are
  expressed in constant prices at the outset with
  no established deflator that could be used to
  inflate them

15
Timing

• Assuming that we pursue benchmarking
• 3 choices (not mutually exclusive)
• Constrain constant price estimates to
  double-deflated constant price I-O tables (2-3
  years)
• Constrain to time-series of current price I-O
  tables (1 year)
• Constrain to 32 industry-level regional accounts
  (immediate?).

16
Discussion

1. the need for benchmarking
2. what source(s) to benchmark to
3. Timing.