These are the slides presented at Surgeons Quarter , Edinburgh for the afternoon ESWG Seminar on Monday 17 June 2019 to provide delegates an understanding of economic statistics and regional analysis.
5. Compiling Economic Statistics
for Scotland – past, present and
future.
Compiling Economic Statistics
for Scotland – past, present and
future.
Sandy Stewart
Scottish Government
6. Compiling Economic Statistics for Scotland
• History
• What we currently produce
• Challenges
• Uses
• Future developments
• Understanding the Scottish Economy
• Long term times series
• Sub-Scotland analysis
• Other analysis for community planning
• Conclusions
15. Remote – a Scottish croft
Crofting
1755 – 20% of
population in
Crofting
counties
1961 – less
than 5%
33k family
members in
crofting
households in
2014 (0.6%)
18. History
Scottish Office
• 1988 – Index of Industrial Production and Construction (quarterly);
GDP Index (annual)
• 1990 – compilation of 1989 Input-Output tables (to complement
1973 and 1979)
1999 – Demands of Scottish Parliament
• 2002 - First quarterly real GDP series (1997-2001)
• Demand for more regular I-O tables, trade and GDP (I) and (E) as well
as (O)
19. History
2008 – Demands of Council of Economic Advisors
• October 2008 – SESCG proposal to embark on experimental SNAP
• Consistent quarterly GDP(O), (I) and (E) components
• Annual I-O tables 1998 to present
• Development of onshore Quarterly Supply Use framework (quarterly)
• First experiment estimates published October 2009
• August 2014 – QNAS – National Statistics status
• Now earlier releases of GDP and revised GDP/National Accounts each
quarter
21. What we currently produce for Scotland
• Input-Output tables – 1998 to 2015 on consistent basis
• Onshore only, consistent with UK National Accounts definitions, broadly
consistent with other measures
• Detailed onshore trade flows with ROW and RUK
• Quarterly, real-terms GVA and GDP
• Consistent with I-O weights, input into QNAS
• Quarterly National Accounts
• GDP by Expenditure, Income and Output components, quarterly imports and
exports to RUK and ROW, Revenues, Index of Manufactured Exports
• Other
• GERS, Retail Sales in Scotland, Oil and Gas Production Statistics, Labour
Productivity, Business Statistics, Labour Market statistics, Energy statistics …
22. New products
• Whole of Scotland Accounts – 2017/18
• North Sea satellite account
• Complements onshore I-O tables
• Quarterly Gross National Income (GNI) – 2017/18
• Direct Investment, Portfolio Investment, Other Investment, Employment flows
with ROW and RUK, North Sea
• Balance of Payments – longer term
• Needs Primary Income, Balance of Trade and some brainwork!
• Sub-Scotland economic statistics
• SESCG sub-group set up in 2018
24. Challenges
• No legal obligation to produce anything, unlike ONS
• But strong user demand
• Ministers, Parliament, Economic commentators, Public
• Especially – real terms GDP, public sector finances, trade ….
• And for new products – GNI, Balance of Trade, Balance of Payments ….
• And earlier estimates ….
• And longer time series/trends ….
30. Regional prices and deflators
• No regional prices/deflators available
• Use UK deflators, weighted by industry
• Adjust for domestic/export consumption for manufactured goods
• Input prices should also be considered …
• Regional double deflation
• Hedonic quality adjustments
32. Uses of Scottish Economic Statistics
• Public, Parliament and Media
• Modelling and Analysis
• CGE model – using I-O framework
• Multi regional CGE model
• Shock and impact assessments
• Forecasting the economy and tax revenues
37. Future Challenges
• Move from survey to administrative data
• HMRC VAT turnover data
• HMRC PAYE RTI data
• Keeping up with UK Blue Book Changes
• H-Approach Double Deflation
• Measuring structural changes in the economy
• Consumption of digital products (often free, contribution to economy hidden)
• Disruptions to traditional industries (travel, music, media, ..)
• Consumption and prices for ‘new’ service providers (AirBnB, Uber, ..)
• Output of MNEs – even harder for regions
39. Understanding the Scottish Economy
• QNAS goes a long way towards understanding current changes …
• But need more …
• … especially relating to long term analysis of employment, output and
productivity, and regional variation
• … but difficult because data sources and methods, industrial
classifications, and geographic boundaries change over time
40. Sub-Scotland Productivity Model – project
underway
1. Produce consistent nominal GVA for onshore Scotland over time
2. Using I-O tables, disaggregate GVA to industry groups
3. Add details on turnover and employment
4. Disaggregate industry groups to local authority areas
5. Deflate by sector and constrain to QNAS/GDP totals
6. Adjust for changes in employment patterns (male/female, full-
time/part-time, hours per job)
7. Compile modelled local productivity estimates.
42. Data sources and methods
• QNAS 2018 Q4 - for annual GVA 1998 – 2018 (21 years)
• Historic sources and analyses
• Presented in real terms (deflated by UK GVA deflator), per capita, and
log10
• FISIM and Imputed Rental adjustments
• Averaging of data sources
• Flexible – can adjust at any stage
43.
44. Sources of historic information
Title Author(s) Date published
1. The Scottish Economy – A statistical account of Scottish
life
AK Cairncross et al 1954
2. Survey of economic conditions in Scotland in 1954 Clydesdale & North
of Scotland bank
Ltd
1955
3. An Inquiry into the Scottish economy (1960-61) JN Toothill 1961
4. Scotland’s Economic Progress 1951-1960 – A study in
regional accounting
G McCrone 1965
5. Scotland – The Vital Market Credland, Luby,
Murray, La Frenais
1966
6. The Structure and Growth of the Scottish economy Johnston, Buxton
& Mair
1971
7. The Renaissance of the Scottish Economy? C Lythe & M
Majmudar
1982
8. Understanding the Scottish Economy Ingham & Love 1983
9. The Caledonian Blue Book, 1997 H Gibson, G
Riddington, D
Whigham & J
Whyte
1997
46. Data sources and methods
• IOCs – 98 industries - arguably too detailed, but deflators available for
IOC groups – consistency with QNAS and GDP analysis
• QNAS – provides IOC detail back to 1998
• Unpublished I-O analysis 1950 – 2014 (pre-dated ESA2007 changes)
• Further work needed to harmonise and make consistent with overall
GVA estimates
49. Sectors of interest
• Regional data system – manufacturing employment
• Sea Fisheries
• Coal Mining
50. Regional Data System – manufacturing
employment in Scotland 1950-1995
• Scottish Office responsible for upkeep of Scottish elements of DTI
database
• Pre-dates the ONS IDBR
• Coverage – all local manufacturing units with 10 and over employees
• Classified to SIC92
• Address, postcode and contact details
• 12,700 local units, 46 years’ data, 4MB
• At least 50 person years’ work!
54. Sea fishing
• Detailed employment and landings statistics since early 1900s
• GVA harder to estimate, but employment (possibly) good proxy for
geographical distribution.
55.
56.
57. Coal mining
• Detailed estimates of mine active in 1950, employment, peak year,
production in tonnes of coal
• Employment and output per worker good proxy for geographical
distribution
• Further refinements needed for recent years
61. Sources and methods
• Input-Output analysis at core – uses many sources IDBR, ABS,
Regional Accounts – good quality from 1998
• Pre 1998 harder – but have
• Data ACOP, ACOC, RDS
• Analysis , SEBs, academic work
• Work incomplete – work in progress
66. To conclude ….
• Scottish economic statistics have matured significantly since the
Scottish Parliament was established
• There are many challenges in producing regional economic indicators
– but not insurmountable
• To understand the changing economy (productivity) need to
understand:
• Long term changes / history
• Relative importance of different industries in different areas
• Need to understand the wider context – GDP, Employment,
Productivity don’t explain everything.
67. To conclude ….
• Scottish economic statistics have matured significantly since the
Scottish Parliament was established
• There are many challenges in producing regional economic indicators
– but not insurmountable
• To understand the changing economy (productivity) need to
understand:
• Long term changes / history
• Relative importance of different industries in different areas
• Need to understand the wider context – GDP, Employment,
Productivity don’t explain everything.
71. ONS Economic Forum, Edinburgh
Gary Koop, Stuart McIntyre, James Mitchell and
Aubrey Poon
June 2019
72. 1. To produce and disseminate timely model-based quarterly regional estimates of
nominal GVA to the same timetable as the UK’s first estimates of quarterly GVA for the
UK as a whole.
2. To produce historical quarterly estimates of regional GVA. Draft scoping report on
feasibility of incorporating greater levels of regional and sectoral levels of
disaggregation using ‘Big Data’ econometric methods
3. To produce real or volume GVA estimates, using the ONS’s real regional GVA data.
4. To explore the possible use of underlying micro-level and administrative data to
produce model-free (or less model dependent) quarterly regional output data
Aim of our project (phase 1)
73. 1. To produce and disseminate timely model-based quarterly regional estimates of
nominal GVA to the same timetable as the UK’s first estimates of quarterly GVA for the
UK as a whole.
2. To produce historical quarterly estimates of regional GVA. Draft scoping report on
feasibility of incorporating greater levels of regional and sectoral levels of
disaggregation using ‘Big Data’ econometric methods
3. To produce real or volume GVA estimates, using the ONS’s real regional GVA data.
4. To explore the possible use of underlying micro-level and administrative data to
produce model-free (or less model dependent) quarterly regional output data
Aim of our project (phase 1)
74. Academic papers:
• KMM - UK Regional Nowcasting using a Mixed Frequency Vector Autoregressive Model with Entropic Tilting
(joint work with Gary Koop and James Mitchell) forthcoming in Journal of the Royal Statistical Society: Series A
• KMMP - Regional Output Growth in the United Kingdom: More Timely and Higher Frequency Estimates, 1970-
2017 (joint work with Gary Koop, James Mitchell and Aubrey Poon), ESCoE working paper 2018-14
ESCoE Blogs (ESCoE.ac.uk/regionalnowcasting):
• November 2018: “More timely and more frequent regional growth estimates”
• February 2019 “Regional growth in 2018 – what happened? ”
• May 2019 “Differences in economic growth across the UK remained substantial in the year to Q1 2019”
Outputs…
77. ONS’s historical regional GVA data from 1966 – 1996, combined with ONS’s equivalent
data from 1997 onwards, UK GVA , wider UK macroeconomic indicators, and most
recently regional CBI Survey data and claimant count data. For more details see the
working paper(s).
BVAR – econometric model estimating interdependencies between all these variables.
Widely used in the macroeconomics field, and extensively by central banks and others.
Lots of testing of different specifications and ways of estimating these models – see the
working papers for the full details.
Key issue is how to incorporate more timely information into our model for regional GVA
growth to produce more accurate (and timely) estimates?
Data/Model
78. Different sets of results to show you:
• Very brief overview of KMM (2019) results…
• Results from KMMP (2019)
• In sample fit / historical data
• Dating the cycle(s) / recession profiles
• Economic (inter)connectedness
• Latest estimates, and comparison to the UK
• Comparison to Scottish data
Outline of progress
86. Regional connectedness…
Examine regional connectedness using Diebold and Yilmaz (2014) variance
decomposition approach. This provides two measures:
The "connectedness from" measure.
• This is a measure of how information in other regions impacts the forecast error
variance of region i.
The "connectedness to" measure.
• This is a measure of how information in region j influences the forecast error
variances of other regions.
91. Substantial work undertaken to produce model based estimates of regional
economic growth (real and nominal) to (approximately) the same time scale as
UK GDP
Enables better understanding of the evolution of regional economic growth in
real terms, now and historically
Additionally enables exploration of regional economic cycles and economic
interconnectedness over time.
Summary…
92. Continuing production and dissemination of estimates
Regional short—term indicators
Developing ONS capacity with these methods and techniques
Next steps…
95. Economic Statistics Centre of Excellence
Improving the quality of regional economic indicators
June 2019
96. Improving interregional trade flows data
Data map of existing trade related data
Paper on feasibility study of using alternative data sources for estimation of inter-
regional trade flows within the UK
Paper on estimation of Scotland-RUK and Northern Ireland –RUK inter-regional trade
flows
Paper on inter-regional trade flows for all four UK regions
Phase 1
97. Improving interregional trade flows data
Data map of existing trade related data
Submitted in December 2017
Paper on feasibility study of using alternative data sources for estimation of inter-
regional trade flows within the UK
Submitted in April 2018
Paper on estimation of Scotland-RUK and Northern Ireland –RUK inter-regional trade
flows
Estimates of trade between Scotland, NI and RUK submitted in September 2018
Paper on inter-regional trade flows for all four UK regions
Estimates of trade between 4 nations of the UK submitted in March 2019
Phase 1
98. Existing supply and use tables taken as “true”
Provide estimates between NI & GB, and Scotland and rUK
Flows were then split up using proxy information, either from the point of view of the
exporter or importer
Road and Maritime Freight data key sources
Wales was more tricky – 2007 tables were simply re-based
• Transparency about this simplicity of approach was really important
Overall approach
99. During the construction of the estimates, several issues were raised which have
informed our recommendations and proposals for Phase 2.
This includes:
• Differing data collections and coverage inconsistencies
• Different approaches to producing trade estimates in the regional tables
• A need to ensure consistency with UK approaches following ESA 2010
Extremely high policy interest in the results - now not just from a devolution perspective
but also due to Brexit.
Also in these findings – Welsh Government trying to learn lessons
Findings – data issues
100. Findings – summary (2015)
Scotland NI Wales England
Scotland 1,112 1,606 52,071
NI 1,065 147 10,053
Wales 1504 1901 42,477
England 64,432 14,553 38,534
FROM
TO
Scotland NI Wales England
Scotland 2,014 1,107 51,668
NI 1,881 96 9,288
Wales 1,168 1,806 42,909
England 63,953 13,746 39,086
TO
FROM
No
re-exporting
Re-exporting
102. Develop a strategic approach to the collection of trade information and estimation of
trade within the UK, with different approaches for goods and services
This has three strands
• Support & facilitate discussions between ONS and the devolved administrations on
trade collections in goods, developing a strategic approach to these surveys across
the UK;
• A qualitative review of service sector companies; and
• A quantitative appraisal of methods used for estimating inter-regional service trade,
testing the sensitivity of current approaches and providing recommendations.
Inter-regional trade & Supply and Use
103. Develop a framework for the consistent production of Supply Use Tables for Scotland,
England, Northern Ireland & Wales
Currently recruiting! Closes today!
Inter-regional trade & Supply and Use
104. Building on the University of Southampton work
Initial Estimates produced November 2017
Follow up in February 2019, investigating small area estimation techniques for stabilising
price weights
There were a number of recommendations which we will be taking forward in Phase 2
Regional Consumer Prices
105. Reviewing Additional Data Sources used in the construction of national weights, and for
regional expenditure estimates and other regional work.
Undertake an investigation into estimation and smoothing techniques to be used in
conjunction with multiple data sources
provide quality measures which a regional CPI or Consumer Prices Index, including
owner occupiers’ housing costs (CPIH) should aim to achieve; this will help identify what
is an acceptable level of temporal variability over time.
assess the temporal stability of the regional price quotes and the capacity for price
estimation to be improved using model-based methods
Regional Consumer Prices
106. Produce an updated set of regional CPIH estimates
Provide recommendations and proposals for future work and possibilities for regular
publication
If possible, provide an assessment of the national administrative sources which may be
used in the future to inform regional prices
Regional Consumer Prices
107. Economic Statistics Centre of Excellence
Improving the quality of regional economic indicators
June 2019