REGIONAL ECONOMIC STATISTICS: INITIAL OBSERVATIONS ON THE FIRST REPORT FROM CHRISTOPHER ALLSOPP'S REVIEW
JOHN PULLINGER
ONS
12 JANUARY 2004
This document is also available in PDF format
1. This paper reviews Christopher Allsopp's first report from the perspective of ONS. It provides a brief summary and interpretation of the information requirements, including the detail required and quality issues. It goes on to consider the broader statistical context and links to other current statistical developments. It provides a general analysis comparing the desired position with the current situation before detailing the likely statistical developments needed to achieve the aims of the report. The paper also considers organisational requirements and timing issues along with initial views on how ONS might take forward the work needed to make the necessary improvements to regional economic statistics.
2. ONS welcomes and supports the findings of the review to date. Implementation will be challenging and will raise a number of questions about resourcing, timing and policy that will need to be addressed before it is possible to be clear just how much of the proposals can cost effectively be put in place. This will be the subject of further debate over the coming months. This document is in the form of initial observations. A formal response from ONS will await the publication of the final report of the review team.
THE INFORMATION REQUIREMENT
3. At its simplest the requirement is to be able to compare the economic performance of the countries and regions of the United Kingdom with each other and over time.
4. This requirement is broken down into seven areas of interest to users of statistics: productivity; flexibility; welfare and redistribution; prices and pay; public services; sustainability and European Union requirements. These requirements overlap, interact and complement each other in many ways but each represents the needs of particular constituencies that need to have their concerns addressed distinctly.
Productivity
5. The central requirement for understanding productivity is to provide for the constituent parts of the UK the key macro-economic aggregates available at national level, notably gross value added. Gross value added is especially salient given that it is the central measure of the public service agreement target on regional prosperity.
6. Coupled with this macro information, understanding of productivity also requires information on the drivers of productivity which have been identified in policy terms to be: investment; skills and human resources; innovation; competition; enterprise. Flexibility
7. The ability of regions to cope with shocks and to respond to broader economic and social changes may be manifest in labour markets, product markets and capital markets. The emphasis in the report is on labour markets. This encompasses, in particular, issues of mobility and the impact of housing, transport and relocation incentives.
Welfare and redistribution
8. The need for information is couched in terms of territorial justice. This covers the impact of benefits and tax credit systems, investment in and access to public services and the financial settlement relating to devolution.
Prices and pay
9. There is increasing interest in prices and pay at sub-UK levels to understand inflation pressures, differences in the cost of living, both levels and changes, and issues of pay flexibility. This also links to funding allocations, for example to local and health authorities.
Public services
10. Information about the levels of public expenditure by type (for example on health, education, transport) is increasingly of interest for analysis of funding decisions but also to assess broader policy questions such as the effectiveness and impact of policy in different geographical areas.
Sustainability
11. There has been considerable development of sustainable development and quality of life indicators in recent years. This is expected to grow. It links with development of indicators of neighbourhood renewal and rural issues.
European Union
12. There is a general need to be able to compare the development of countries and regions within the UK with those elsewhere in the EU and a specific need to provide information to assess eligibility for structural and other funds that are allocated at the EU level.
DETAIL REQUIRED
13. The summary of the information requirement provided above needs substantial elaboration to identify more precisely the concepts that need to be captured. As part of that process it will also be necessary to establish the detail to be collected in each area across a wide range of dimensions. The most important ones are: the aspects of gva; industry detail; regions and sub-regions; residence and workplace; periodicity; current prices and real terms; levels and changes; absolute amounts and rates.
Aspects of gva
14. National accounts are presented in three ways: based on income, expenditure and output. At the UK level all three are produced in a fully integrated way. At a sub-national level this may not be necessary and in any event may make little conceptual sense or be prohibitively expensive. The report proposes to continue with the current conceptual approach which focuses on measures of incomes, including the income measure of gva and household disposable income. This is to be complemented by information on domestic expenditure. It is not proposed to develop information on exports and imports. It is also proposed that real gva is produced using the output approach.
Industry detail
15. European requirements are currently onerous in terms of providing industry detail and the quality of the figures is likely to be substantially lower at the detailed level than at higher aggregations. The report proposes that less detail is really needed. Indeed at current levels of quality the detail available may add little to our understanding of regional economies. This may be true, but depending on the method of collection it may be that the detail can become a by product of the collection of higher aggregations rather than a cause of added cost to the Exchequer or burden on business. How this issue is resolved will depend on the balance between "top down" and "bottom up" collection. This issue will be considered later in this paper.
Regions and sub-regions
16. The focus of the report is on the countries and regions of the UK. However, it is recognised that differences between variables within regions are often greater than those between regions. In many areas information at lower levels of geography will materially help understanding of change at the regional level. Where possible it will be useful to design systems that enable information for lower levels of geography to be produced.
Residence and workplace
17. In the south east of England in particular there are substantial differences between information that relates to where people live and that which relates to where wealth is created (that is, where they work). It is always important to consider whether residence or workplace based measures are appropriate. For some questions both will be useful.
Periodicity
18. At the UK level there is a well articulated set of measures with monthly, quarterly and annual periodicity. The report proposes that measures at the sub-UK level are needed generally on an annual basis. There may be exceptions. For example in Scotland a range of quarterly measures already exists and across the UK there are a number of indicators, for example on the labour market, available more frequently.
Current prices and real terms
19. The general requirement is for "real" changes, that is with changes due to price movements excluded. At present most sub-UK information is at current prices only. Meeting this requirement will present a significant challenge.
Levels and changes
20. The PSA target is expressed in terms of growth rates. It is clear, however, that the broader need to understand economic progress requires information about levels and changes. This is true of all variables, including prices.
Absolute amounts and rates
21. Comparison between areas can often most appropriately be done by producing a rate e.g. gva per head. The appropriate denominator will need to be available in each case, and the precision of the denominator may be as important as the precision of the numerator in compiling the necessary measures. The need for rates means that delivery of improved regional information will be dependent on the availability of reliable information on population (resident and workforce) and other denominators.
QUALITY ISSUES
22. In order to assess the statistical requirement it is necessary to understand the levels of quality required. To attain the levels of quality currently available at the national levels for sub-UK statistics is likely to be prohibitively expensive. The requirement is probably for a lower level of quality in any event, but it is important to be quite clear about this in managing expectations. Four aspects of quality are especially important: relevance, precision; timeliness; revisions.
Relevance
23. Increasingly it is difficult to assign a location to an economic transaction. The combination of globalisation of business and the role, speed and complexity of information transfer in the economy mean that geography is often meaningless to the transactors. In these circumstances geographical allocation of transactions can be arbitrary and do nothing to aid our understanding of what is going on. Indeed it can give false signals. At smaller levels of geography this problem increases. There are likely to be measures that are useful at the UK level but which have little or no value at regional level. It will not be sufficient therefore simply to choose measures that work at the national level and replicate them at the regional level. In each case it will be necessary to understand clearly the concept to be captured and why it will help us understand the regional economy.
Precision
24. The report makes clear the goal to treat "all regions equally". Given that the primary aim is to allow robust comparisons between each of the 12 top level countries and regions this is a reasonable goal. It carries the consequence that it will be proportionately more expensive to produce results of the required quality in smaller regions. Thus, other things being equal, the sample size required for a large region like London, will be the same as that required for a small country, like Northern Ireland. This will have funding consequences.
Timeliness
25. It is clear that availability of the key regional aggregates is currently too slow to meet user needs. It is proposed that early estimates are produced from indicators ahead of the publication of more comprehensive annual surveys. This mirrors the approach taken for the national accounts but has consequences for the need to enhance short term sources of statistics as well as annual structural surveys.
Revisions
26. Revisions are always inconvenient. The more partial information is when first results are published the greater is the likelihood of future revisions. In developing a system for regional economic statistics it will be necessary to be quite clear about the nature and scale of likely revisions so that explicit decisions can be made about the trade off between timeliness and certainty.
STATISTICAL CONTEXT
27. The Allsopp report calls for a framework that helps make sense of the many and varied demands for regional data. This is will be most important in its own right but will also enable the new information created as part of this review to benefit from and contribute to wider developments in other areas. Parallel developments that will support the creation of better regional data are: statistical developments being led within the devolved administrations and the Greater London Authority; development of regional statistics within the European Union and internationally; re-engineering of the national accounts and development of an integrated system of population statistics.
28. Of direct relevance is the framework that has been developed for neighbourhood statistics. Neighbourhood statistics has been developed primarily to support neighbourhood renewal policy but it does cover the economic domain as well as other policy areas. The Allsopp proposals start from the economic domain but also require information beneficial to social policy concerns. It may be that the best way to set out the statistical requirements is to use the neighbourhood statistics framework which elaborates nine policy domains: access to services; community well being; crime and community safety; economy; education skills and training; health; housing; physical environment; work.
29. The current set of regional economic statistics are primarily the product of a top down approach. This means that the UK level statistics drive the requirement and sub-UK statistics are available as a by product. In the past this has served us well since it has enabled useful information to be compiled cheaply and easily. The increased requirement for reliable regional information means that this approach is no longer adequate and it will be necessary to plan to meet regional requirements more directly. The neighbourhood statistics system, however, is based on the bottom up principle and seeks to provide information at the lowest level of geographical detail and then aggregate. The experience gained already with neighbourhood statistics will be of significant value in delivering improved regional economic statistics.
STATISTICAL REQUIREMENTS
30. The statistical requirements from this agenda are substantial. They cover analytical work as well as enhancements to statistical sources.
Analytical issues
31. The prime need is for information that is consistent over time and across the countries and regions of the UK. Before considering new collections of information we should consider the extent to which existing information could be accessed and analysed in a way that meets the requirements. In many circumstances it will in any event be necessary for there to be analytical input to ensure consistency. This will be especially true where different sources are used to compile statistics in different areas and where administrative statistics, which are subject to changes in definition from year to year, are used.
32. Specific further areas of analytical work will be: the creating of analytical datasets capable of use at the regional level, for example enhancement of the business data linkage dataset; modelled estimates to create early estimates or to compile statistics on topics where only proxy information is available at the regional level. Good thinking at an early stage will also enable us to consider "hybrid" data collections which combine business data, household data and administrative data in the same instrument. This could be especially useful in understanding questions of productivity and flexibility. There is some useful experience of this technique in the Department of Trade and Industry. It has also been used in Canada and some other countries. The proposed UK wealth and assets survey system is likely to follow this kind of approach.
Administrative data sources
33. There is substantial potential in existing administrative data sources, especially those held by Inland Revenue (PAYE and others), Customs and Excise (VAT and others), the Department for Work and Pensions (benefits data and others) and the Treasury (public expenditure information system). Information that can be linked at the unit record level for statistical purposes has great potential for minimising the business compliance costs on businesses and improving quality, but will require substantial initial investment in creating new systems to join datasets together reliably. There will also be legal and administrative concerns to be addressed. The experience of other countries, for example, Canada shows that substantial gains can be made from this approach.
34. Current evidence suggests that many administrative data sources are especially poor at capturing location effectively. The availability of a single reliable national address register will be of special value in improving the quality of these sources. Other quality concerns about administrative data relating to scope, coverage and validity for use in statistical analysis will require a substantial investment in quality management and in deriving methodological approaches that remove discontinuities in data sources. This will be particularly necessary where administrative rules vary over time or are different or are applied differently in different parts of the UK.
35. There may well be value to be gained from exploring the potential of private sector administrative datasets, for example, retail sector and finance sector databases, to provide valuable economic information at the regional level. As well as these sources there will be data held by local authorities and others that could be of value. It will be useful to review and extend work that has already been done for neighbourhood statistics to test the potential of these sources.
Business register
36. A critical source that will need to be upgraded is the interdepartmental business register. Current proposals to establish a single UK comprehensive business directory will need to be carried forward if the new information requirements are to be met.
Business surveys
37. A wide range of business surveys will need to be enhanced, with substantial boosts to sample sizes. Redesign of business surveys will also involve the collection of more information about individual establishments for multi-site businesses. This will need careful thought to avoid the creation of discontinuities in existing series and to make sure that only variables that can sensibly be disaggregated at establishment level are collected at that level. It will all need good planning to get maximum benefit from the current ONS statistical modernisation programme. The main business surveys that will need to be enhanced are: the Annual Business Inquiry; the Annual Register Inquiry; the short period inquiries of output; the retail sales inquiry; consumer prices collections; producer prices inquiries; Vacancy Survey of Enterprises. Other new or expanded survey instruments may well be necessary to capture aspects of productivity and flexibility. In addition it will be necessary to find ways to improve the regional coverage of current attitudinal surveys of business, which are primarily conducted by private sector bodies.
Household surveys
38. The development (in planning) of a population register and a continuous population survey will assist the development of regional economic statistics. It will be necessary in addition to consider specific enhancements to the survey modules relating to the Expenditure and Food Survey and the Labour Force Survey. Enhancement of information about household spending patterns to provide the necessary detail at regional level will be particularly costly but vital, both in its own right and to provide weights for price indices. Development of the International Passenger Survey and the potential of other social surveys will also need to be evaluated. The proposed wealth and assets survey system could also provide significant relevant information.
ORGANISATIONAL AND OTHER REQUIREMENTS
39. The report describes a concept of "devolution with constrained discretion". This means two things. First it means that the organisational arrangements must ensure comparability between areas. The discretion available to devolved administrations and other sub-UK bodies needs necessarily to be constrained if comparability is to be achieved. Second it means that devolved administrations and other regional and local bodies need to be partners in a regional economic statistics system not just users of it. They may be well placed both to supply information and to quality assure it on the basis of local presence and knowledge.
40. The goal is to create UK wide statistics that allow comparisons between the constituent parts to be made with confidence. To achieve this goal will mean that the sources and methodologies will often vary between areas. Management of this in a way that respects the goal but also the interests of each of the stakeholders will be complex. Strong leadership from ONS will be vital. A series of partnership groups might also take on a remit to further the delivery of the recommendations of this review as part of a wider role to improve relationships between the UK level statistics system and, respectively, the devolved administrations, the English regions, local authorities and other groups with an interest in regional statistics (including the private and voluntary sector and academia). In developing the relationships with the English regions it would be possible to take forward the proposal in the review report for ONS to establish a regional presence.
41. We need to fit such development and implementation work into the established National Statistics planning arrangements so that the developments are all of a piece with other developments in the areas of statistics concerned. And this will need to cover the further work that the review team will be doing on the degree to which the changing economic structure of the UK is being properly reflected in official economic statistics. But to the extent that we may be establishing a changed statistical system with a different centre of gravity, we may need to look radically at many of our existing arrangements rather than assume that the current planning structures will suit the future.
42. As well as these organisational requirements there are two further areas that need urgent attention and will affect the successful implementation of the findings of the review team: access to administrative data and compliance costs. It will be necessary to tackle current barriers to the availability for statistical purposes of administrative data, especially tax data. It will also be necessary to be clear about the impact of increasing burdens on businesses. The review report notes that the burden of statistical form filling on businesses is very small compared with other burdens placed on business by government but this issue has in the past been contentious. It also needs to be considered alongside the question of access to administrative data. If we can demonstrate that by reusing for statistical purposes data already held by the government in order to avoid collecting it a second time on a statistical form then any increases in form filling burden may be seen to be more acceptable.
TIMING
43. Full implementation of the report will take time, probably through to 2008. In the short term progress can be made by following through on current improvement programmes to regional economic statistics and neighbourhood statistics. Early access to tax data will also allow progress to be made on some fronts. These improvements will still be in the context of a system designed to produce national results and will therefore inevitably be limited. A business case is being worked up for the 2004 spending review to achieve the longer term gains. The costs will depend on the extent to which changes can be made in a way that benefits from the current ONS modernisation programme and from wider developments, including a national address register and a national population register.
There will also be a significant challenge for us in accessing the necessary skills. Substantive gains to quality can be expected progressively to come on stream from late 2006 through 2008.
ONS January 2004