Transcript Open

GDP Timescales
Flash Estimates
Richard Morrison
GDP Branch
Short Term
Economic Indicators
Flash Estimates: "…the earliest picture of
the economy according to national
accounts concepts, which is produced and
published as soon as possible after the
end of the quarter, using a more
incomplete set of information than that
used for traditional quarterly accounts."
In Practice,
UK produce three initial estimates
• M1: Preliminary Estimate:
– t+25 days About 44% data
• M2: Output, Income and Expenditure:
– t+55 days About 67% data
• M3: UK Quarterly National Accounts:
– T+85 days
About 80%
• Blue Book, IO tables etc. later…
Scottish GVA Series.
• Currently published at t+112 days
– Index of Construction
• 2004 weight: 65 / 1000
– Index of Services
• 2004 weight: 742 / 1000
– Index of Production
• 2004 weight: 178 / 1000
– Agriculture, Forestry and Fishing
• 2004 weight 15 / 1000
Construction
• Data source
“Output in the Construction Sector”
Published by ONS at t+2 Months
• Earlier estimates would require
– alternative more timely sources; or
– negotiated early access.
Services
•
five components of source data:
– the Retail Sales Index (RSI)
– the Monthly Inquiry into the Distribution and
Services Sectors (MIDSS)
– "Manual Series"
– Other public sector series
– Deflators
RSI
•
•
•
•
Consistently delivered around t+25 days
No significant processing burden
Deflators delivered with Series
Scottish Retail series would be an “easy”
component of earlier estimates
• 5% weight in current GVA estimates
MIDSS
• Delivered monthly with a 2 month lag
• New months data incorporate revisions
• Could apply standard forecasting methods
– Such as ARIMA
• Earlier estimates could incorporate 2/3
months + forecast data.
• 29% weight in current estimates
Public Sector Measures
• Typically stable annual series
• Already use forecasting in GVA indices
• Earlier estimates would be on same basis
as current ones
• 22% weight in current GVA Indices
Deflators
• Most are relatively timely and not
particularly volatile
• Transport deflator exhibits more volatility
– Probably due to correlation with fuel prices
– Possibly implying that triangulation possible
“Manual Series”
• These are volume series supplied directly from
source
– Such as passenger miles
– Parcel volumes
– Banking series
• Possibly the biggest challenge to a Scottish
Flash Estimate
– Not always particularly timely
– Often volatile
Comparison of ARIMA forecasts for one of the Financial Services Data Sources
160
140
120
100
Actual Data
ARIMA
Forecast at
2007 Q4
ARIMA
Forecast at
2008 Q1
80
60
40
20
19
95
19 Q1
95
19 Q3
96
19 Q1
96
19 Q3
97
19 Q1
97
19 Q3
98
19 Q1
98
19 Q3
99
19 Q1
99
20 Q3
00
20 Q1
00
20 Q3
01
20 Q1
01
20 Q3
02
20 Q1
02
20 Q3
03
20 Q1
03
20 Q3
04
20 Q1
04
20 Q3
05
20 Q1
05
20 Q3
06
20 Q1
06
20 Q3
07
20 Q1
07
20 Q3
08
20 Q1
08
20 Q3
09
20 Q1
09
Q
3
0
ARIMA
Forecast at
2008 Q2
ARIMA
Forecast at
2008 Q3
ARIMA
Forecast at
2008 Q4
Production
•
four components of source data
–
–
–
–
The Monthly Production Inquiry (MPI)
The MPI top-up
"Manual Series"
Deflators
MPI and top up
• Similar picture to MIDSS
• Monthly supply, with 2 month lag
• Top up currently used to ensure panel
continuity
– will change when we move to ratio estimation
to be more standard sample boost
• MPI has about 14% weight in current GVA
index
Manual Series
• volume series direct from source
– GW hours of electricity
– Volume of slaughtered poultry
• Possibly the biggest challenge to a
Scottish Flash Estimate
– Not always particularly timely
– Often volatile
• But perhaps not as hard as it looks?
Electricity Series - effect of forecasting data for one Electricity company
120
100
90
80
70
Actual Electricity Series
forecast at 2007Q4
forecast at 2008Q1
forecast at 2008Q2
forecast at 2008Q3
60
20
01
Q
1
20
01
Q
3
20
02
Q
1
20
02
Q
3
20
03
Q
1
20
03
Q
3
20
04
Q
1
20
04
Q
3
20
05
Q
1
20
05
Q
3
20
06
Q
1
20
06
Q
3
20
07
Q
1
20
07
Q
3
20
08
Q
1
20
08
Q
3
Volume of electricity generated
110
Production Deflators
• Supplied by ONS at t + 10 weeks
• Much more volatile than services deflators
• Forecasting methods look ropey
• Negotiated earlier access to ONS
deflators
– Perhaps partial information to inform
forecasts
Potential
Approaches
Reweighting UK Flash Estimate
– Quick and straightforward
– Almost immediate Scottish Headline Figure
– Initial look suggests not that informative
– Might improve with unrounded figures
– Not a true Scottish estimate
– Lacks detail
Analogous Methodology
to UK Flash
• E.g. to produce estimate at t + 50-70 days
• Partial data + ARIMA – Holts Winters etc.
• Some sectors too volatile for robust series
using this approach
– But could work to identify better sources
– Or supplier collaborated forecasts
Partial Early release
• For example of
– growth estimates for some sectors
• e.g.
– “index of services excluding finance”
– Perhaps with a finance index carrying heavier
caveats
Bring forward existing series
• Evaluating the possibility in any case
• 2 weeks is “challenging buy feasible”
• Will still lag UK estimates
– Still some obstacles if this approach would
maintain existing methodology
Discussion
• Focus of efforts has to be guided by user
requirements
• Resource issues not necessarily trivial
• Briefing and National Statistics
implications will be an issue