Transcript Open - The Scottish Government

Benchmarking quarterly GDP
to other official statistics.
Andrew Mortimer
Short-term Economic Indicators Branch
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
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
Lagged by a number of
years
• 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
Example: Hotels and catering
120.0
100.0
80.0
60.0
constant price
40.0
20.0
0.0
1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
Example: Hotels and catering
120.0
100.0
80.0
constant price
60.0
current price
40.0
20.0
0.0
1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
Example: Hotels and catering
120.0
100.0
80.0
constant price
60.0
current price
Implied deflator
40.0
20.0
0.0
1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
Example: Hotels and catering
120.0
100.0
80.0
Regional accounts (index)
60.0
current price
40.0
20.0
0.0
1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
Example: Hotels and catering
1.10
1.05
1.00
0.95
diffs (annual)
0.90
0.85
0.80
0.75
1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
Example: Hotels and catering
1.10
1.05
1.00
0.95
diffs (annual)
splined diffs
0.90
0.85
0.80
0.75
1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
Example: Hotels and catering
120.8
100.8
80.8
current price
60.8
benchmarked current price
40.8
20.8
0.8
1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
Example: Hotels and catering
120.8
100.8
80.8
current price
60.8
benchmarked current price
Implied deflator
40.8
20.8
0.8
1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
Example: Hotels and catering
120.8
100.8
80.8
constant price
60.8
benchmarked constant price
40.8
20.8
0.8
1 2 3 4 1 2 3 4 1 2 3 4 12 3 4 1 2 3 4 1 2 3 4 1 23 4 1 2 3 4 1 2 3 4 1 2 3
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
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
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
Timing
• Assuming that we pursue benchmarking…
• 3 choices (not mutually exclusive)
– Constrain constant price estimates to doubledeflated 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?).
Discussion
1. the need for benchmarking;
2. what source(s) to benchmark to;
3. Timing.