Welsh Assembly Government

1
Welsh Assembly Government
An index of output in the retail and wholesale
sectors in Wales
Andy Milne

• Llywodraeth Cynulliad Cymru

2
Index of Distribution (IoD)

• Assembly
• Andy Milne - Statistician
• Sue Hibbitt - Higher Statistical Officer
• ONS
• Ian Richardson
• Kim Swain

3
Why a Welsh Index of Distribution (IoD)?

• Lack of regional economic data
• Not timely and it is in current prices
• We already estimate a Welsh Index of Production
  and Construction (IoP/C) and there was call for
  service sector data
• Pressure on ministers
• Scotland already produce quarterly estimates of
  GDP
• Distribution is a fair-sized sector (12 of Welsh
  economy)

4
How we set up the index

• A scoping study was conducted by ONS in 2002 and
  the final report delivered in August 2002
• The IoP/C for Wales (mainly) uses a combination
  of directly collected volume data and
  post-stratified monthly production inquiry data.
• The logical option was to copy this approach and
  go for the least invasive approach to estimating
  the index.
• It was anticipated that the post-stratification
  of GB monthly turnover and retail sales data
  using short-term employment data (STES) data
  would be fine. Use synthetic techniques to
  overcome the need for expensive and burdensome
  data collection
• But this is where things started to go array

5
Distribution in Wales from Regional Accounts
ABI/2 estimates

• Distribution is split into three, two digit
  classifications car sales repairs, wholesale
  and retail
• IoD 12 of Gross Value Added (GVA) for Wales
• 2 cars
• 4 wholesale
• 6 retail
• Using post-stratification to estimate cars
  wholesale seemed fine
• The results for retail however seemed very
  similar to the UK estimates

6
Retail

• An analysis of the employment data in Wales
  showed a very interesting patterns retail
  turnover is dominated by a small number of very
  large companies. For example think about the
  large supermarket chains
• See handout
• The data demonstrates how most of the employment
  in Wales is in companies that are not Welsh
  registered. This off diagonal analysis
  demonstrates that the retail sector is dominated
  by non-Welsh retailers
• What is the problem? It means that Wales is
  getting a very high proportion of the turnover
  from these large companies and the
  post-stratification technique is effectively
  apportioning similar quarterly movements in Wales
  as for the UK. In crude terms if UK goes up by an
  index point, the Welsh figures show a
• similar growth

7
Why is this technique not valid?

• In effect an increase in employment in Wales
  compared to the companies overall employment for
  that quarter shows a increased growth in retail
  in Wales
• However the data showed an exact movement for
  Wales the UK. This misses a major driver for
  this sector. productivity
• The town centre survey relating to 1997
  demonstrated that there is a notable difference
  between productivity per head of staff of urban
  and rural shops and in particular across regions.
  If an example of this is given London is more
  profitable per employee than in Wales. However
  our synthetic analysis was ignoring differentials
  of profitability and assigning similar levels to
  Wales as well as other areas.

8
What we did next

• Where do we go from here?
• Considered other approaches. The nature of retail
  always suggested that local level data may be
  available. For instance each supermarket must
  surely hold point of sale information relating to
  turnover for each store
• Local collection had not proved very successful
  in the past both the town centre study and the
  Scottish Executive have struggled to collect such
  data, we were warned off it by the general
  consensus in ONS, but I felt sure it may work
• Enter Mike Tom superstar. Those companies not
  willing were visited we re-assured and
  persuaded them to contribute

9
Next stages

• 30 biggest retailers in Wales represents 50 of
  the sector
• Turnover for the rest are estimated using post
  stratification techniques
• The data is deflated using RSI UK deflators
  (similar to the Welsh IoP/C). If any regional
  deflators should come along we would consider
  these.
• Problem with back data many companies could not
  supply long series of back data
• Problem with seasonal adjustment. We have not got
  a long enough time series of clean data to run
  X12. Clearly each part of distribution is likely
  to be effected by seasonally

10
Next Stages

• We hope to publish in Summer 2004
• It will remain an experimental series until we
  feel it is justified to be included under the
  heading of a National Statistic
• We hope to carry on producing the data on a
  quarterly basis
• We are already looking at other indices for
  Wales hotels, restaurants and catering is next
  on our shopping list and work has already
  started on this followed by transport..we hope
  to get moving with this soon

11
Conclusion

• If at first you dont succeed
• Try to think outside the box occasionally (my
  problem is staying inside it)
• Its the people that matter. Mike Tom , Rhian
  Jenkins, Ian Richardson and Kim Swain supported
  by Ole Black pulled the stops out to ensure we
  overcame the obstacles in the way
• It takes time. A baby takes 9 months, and
  elephant takes 24 months and IoD takes longer!

12
The employment post-stratification technique

• Within ONS the technique is known as MET 5
  (method 5)
• this is the formula used Grossed
  TurnoverIDBR/grossed employment
• Grossed turnover is the total grossed turnover of
  all sampled LUs in the classification in Wales
  over the quarter
• Grossed employment is the total grossed
  employment of all sampled LUs in Wales in the
  classification
• IDBR is the employment held on IDBR for all LUs
  in the classification in Wales