Introduction

Increasing international globalisation and economic integration require more cross-country comparisons and analyses for a more comprehensive understanding of economic phenomena and processes. However, this task is complicated by a lack of harmonisation of statistical concepts and definitions and the risk is there to compare the non-comparable.  Much efforts have been made to better conceptualise and align national practices, in particular within the EU, but more needs to be done to   “translate” concepts and wordings used in a commonly understandable way and to “bridge” statistically between EU and Non-EU countries. This task is precisely that of OECD and it is not necessarily limited to OECD countries, but also includes “big players” worldwide.   

Business frames are of central importance in this context and the interface between the legal and administrative units on the one hand and the statistical unit on the other is of central relevance. The degree of harmonisation between the two determines the ability to fully exploit or not available information.

This paper summarizes and analyses national SME strategies in 26 OECD countries and 3 non-OECD economies. By its comprehensiveness, it intends to provide insight into the considerable diversity, but also similarities, across OECD countries. It is hoped to be of interest to the Experts participating in the 18th Roundtable on Business Survey Frames. The focus on SMEs is voluntary, since embedded in a larger OECD exercise which culminated in the Special Workshop on SME Statistics held at the 2nd OECD Conference of Ministers Responsible for Small and Medium-Sized Enterprises in Istanbul, 3-5 June 2004.

This analysis id based upon a special OECD questionnaire on the measurement issues of SME behaviour, with emphasis on how NSOs currently handle SME statistics and whether or not they see a need for change in the medium term.  The focus, hence, has been on the matching or mismatching between objectives in SME coverage and methods used rather than on purely technical aspects.  The information obtained was analysed and a first synthesis of findings was discussed at the OECD at a workshop held last September.  Since the workshop, a considerable effort has been made to complete and further analyse the information obtained and to take into account comments received from countries as well as from Delegates[1]

Generally speaking, the interest in SMEs, their life-span, performance, factors of success or failures and general characteristics is steadily growing. A better quantitative underpinning, higherr quality and better comparability of data are needed from a policy and an economist’s perspective. In a context of need  to stimulate economic growth and employment, policy decisions need facts and reliable figures.

I       Definition of SMEs

The characteristics of a SME reflect not only the economic, but also the cultural and social dimensions of a country. Not surprisingly, very different practices are used across countries and over time.  Some countries tend not to make a distinction between legal and statistical definitions.  This is the case for Canada, Greece, Portugal, Mexico and the Slovak Republic. The definition can be based on a threshold in revenue, like it is the case in Canada, it can be based on number of employees, as in the UK, or it can combine the number of employees and turnover for legal and statistical purposes like in Portugal. The Slovak Republic, Mexico and Greece use the number of employees as criterion.

In most EU countries, there is a distinction between the legal definition and the statistical definition.  The legal definition, based on EU recommendation number 2003/361/EC takes account of the number of employees, annual turnover, annual balance sheet and independence (see box 1 below).In some cases, the monetary thresholds have been adapted (Italy).  More generally, this definition has served as a basis for other European countries, such as accession countries (e.g. the Czech Republic) and Turkey.

Box 1:  The definition of SMEs according to the European Commission recommendation 2003/361/EC

- For legal and administrative purposes: Enterprises Employees Annual Turnover Annual Balance sheet Autonomous   Micro enterprise 1 to 9 < 2 million euro < 2 million euro 25% or more of the capital or voting rights of another enterprise Small enterprise 10 to 49 < 10 million euro < 10 million euro Medium enterprise 50 to 249 < 50 million euro < 43 million euro Large enterprise More than 250 > 50 million euro > 43 million euro   - For statistical purposes: The main criteria of SME statistics for statistical purposes are the number of persons employed.

Other EU countries use a simplified legal definition based on employment and turnover.  This is also the case with Hungary and Moldova.  Some EU countries do not have a commonly accepted legal and administrative definition; for instance the Netherlands and Spain.  Similarly, in New Zealand there is no common administrative definition; turnover is used by some, taxes on employee salaries and wages by other administrations.  In Brazil, different criteria and thresholds are used for different legal, fiscal and international trade purposes.

Denmark, France, Norway and Switzerland do not use a legal definition.  In Japan, the regular workforce, together with capital or investment, determine the size class, however, thresholds vary for the different activity classifications.  This is also the case in Korea, where SMEs are classified by the number of permanent workers, capital and sales.  In the US, the number of employees is used to identify small businesses in most sectors, except in the non-goods producing sectors where annual receipts are the criteria.

The statistical definition is generally based on the number of employees or takes account of a mix of the number of persons employed and turnover.  However, a number of countries do not differentiate SME statistics collection from other statistics; this is the case, for instance, in Denmark, Germany, New Zealand and the US.  Norway uses the number of employees together with the independence criteria.  Finally, a simplified definition, based on the number of employees, is used in France and Finland in addition to the EU recommendation.

Size thresholds and the definition of an SME vary also according to the sector of economic activity.  Different combinations of criteria are sometimes used.  For instance, in Australia employment is used as a criterion for identifying size classes in all sectors but agriculture. Australia is considering introducing alternative or additional criteria for non-employing businesses. In the agricultural sector, size classes are defined using the estimated value of agricultural operations (EVAO) that is evaluated based on physical production criteria and sales value.

In Japan, capital or total amount of investment together with the workforce is used to define SMEs, but even though the same criteria are used, thresholds that apply to each element not only vary by sector but also by criteria. To illustrate this point: taking the criteria “workforce” and “capital/investment”, a SME in Japan has an upper bound of:

• 300 persons and 300 million yen in manufacturing, construction and transportation;
• 100 persons and 100 million yen in wholesale trade;
• 100 persons and 50 million yen in services industry;
• 50 persons and 50 million yen in retail trade.

The evidence gained from the strategy questionnaire clearly shows that the diversity and richness of SME characteristics, political strategies and economic conditions are unlikely to ever yield a commonly used and accepted definition of SMEs. This may also be unpractical from an analytical point of view.  However, the stocktaking so far allows already identifying 4 different dimensions for greater harmonisation and the possible elaboration of target definitions as shown in box 2 below:

Box 2:  Four dimensions for elaborating an SME target definition

II     The business frames used for SME statistics

The business frame is a key element for ensuring adequate survey population coverage, data reliability and continuity over time.  In some countries the business frame covers all sectors and size classes; it is not specific to SMEs.

For half of the countries that have replied to the OECD questionnaire, the business frame brings together several sources, sometimes according to sectors or size class  In Australia, Brazil (for agriculture, electricity and construction), Greece (for almost all sectors of activity), Spain and agriculture in the US, the business frame is based on administrative sources.

Statistical sources are used in Greece (agriculture) Korea, Mexico, Portugal, Switzerland and Turkey (agriculture) for the business frame.

The business frame brings together administrative sources and statistical sources in Germany (except for sectors 01-02-05 and 75), Hungary and Spain.  As specified, agriculture receives a different treatment in Austria, Germany, Greece, Italy, Mexico, Spain, Switzerland, Turkey, the UK and the US.  Only Germany, Spain and the US have specified that some services are not included in the main business frame.

In Australia, Austria, Greece, Italy, Mexico, Spain and Switzerland, all size classes are covered in the business frame for all sectors but agriculture, forestry and fisheries. It should be noted, however, that in Greece the register does not include the employment variable. Canada excludes non employers and businesses with less than 30 000 CAD in Goods and Services Tax sales, while New Zealand excludes businesses with Goods and Services Tax annual turnover less than 30 000 NZD, regardless of number of employees. In Spain, some non-employers in sectors 80, 85 and 90 are excluded as well as agriculture, forestry and fisheries. In the UK the 1 employment size band and some 2 employment size band that are not registered in administrative sources (with less than 56 000 GBP or paying wages below 4 000 GBP) are not covered. Brazil, the Czech Republic, Denmark, Finland, France, Hungary, Japan, Korea, Moldova, the Netherlands, Norway, Poland, Portugal, the Slovak Republic, Sweden, Turkey declare full size class and sectoral coverage.

The business frame[2] covers a variety of units in different countries. However, before drawing any conclusions, it should be said that different designations may well cover similar content. As an example: what in Canada is called ‘location’ may cover what in the EU is called ‘local unit’, or the Australian ‘Type of Activity Unit’ may well be a ‘kind of activity unit’.

In Australia, the business frame covers the administrative tax unit and the Type of Activity Unit (entity that can report production and employment for similar economic activities). In Brazil it covers the local unit and the enterprise.  Only some non-European countries report covering establishments, this is the case for Canada, Japan, Korea and the US. Further to establishments, enterprises are covered in Canada and the US.  The location is another unit covered in Canada and the tax unit is covered in the US.  In Japan the business frame covers establishments, as defined as a single physical location where economic activities, such as the production or supply of goods and services are conducted.

In the EU and other European countries the situation is less harmonised than could be expected.  In fact, the EC Regulation 696/93 on the statistical units prescribes the list and definitions of units for all EU members (see next paragraph). Several non-members use the same framework. Enterprises are covered in most European countries, except Finland, France, Hungary, Norway and Portugal. In these countries it is the legal unit that is covered.  The legal unit may be different from the enterprise by the fact that it may take several legal units to form an enterprise.  Further to enterprises, European business frames would typically cover a local entity (either local unit or local kind of activity unit).  Extra dimensions could cover the legal unit or the enterprise group as is the case in Denmark, Finland, France, the Netherlands, Sweden and the UK.  In Hungary the business frame covers legal units:  companies, partnerships and sole proprietors.  Sweden has the largest variety of coverage with no less than 7 different units for statistical and administrative uses.

The Council Regulation [(EEC), No. 696/93 of 15 March 1993] on statistical units for the observation and analysis of the production system in the Community lays down a list of eight (types of) statistical units defined on the basis of three criteria (legal, geographical and activity criteria):

• the Enterprise;
• the Institutional Unit;
• the Enterprise Group;
• the Kind-of-activity Unit (KAU);
• the Unit of Homogeneous Production (UHP);
• the Local Unit;
• the Local Kind-of-Activity Unit (local KAU);
• the Local Unit of Homogeneous Production (local UHP).

In the UN classification ISIC, statistical units[3] are the entities for which information is sought and for which statistics are ultimately compiled. These units can, in turn, be divided into observation units and analytical units. The statistical units in the International Standard Industrial Classification (ISIC) Rev. 3 comprise:

• the enterprise;
• enterprise group;
• kind-of-activity unit (KAU);
• local unit;
• establishment;
• homogeneous unit of production.

For all countries that responded to the questionnaire, the SME business frame, if it exists, is embedded in the business frame used for all enterprises. One would therefore expect exclusions in the business frame to apply to the SME business frame as well. This is most often the case, however, in few cases countries reported no exclusions to the SME business frame, while some exclusions were reported to the business frame.  The SME business frame in Finland excludes family farms (sector 01-02 and 05).  The reliability of the classification is questioned in Greece.

The business frame is managed by the NSO for most countries. In Denmark the Central Business Register is managed by an agency under the Ministry of Economic and Business Affairs, the NSO supplements it with additional information. Greece specified that the business register is compiled by the fiscal authorities, managed by the NSO and dynamically updated using survey results. In the US, statistical institutions are decentralised. The National Agricultural Statistics Service maintains a sectoral business frame for agriculture covering farms, and the Bureau of Labour Statistics maintains another frame covering business establishments.

In most countries the business frame is not used by other institutions for conducting surveys.  Where other institutions can use the business frame, commonly those institutions are government or “authorised” bodies.  Among the different institutions, the Czech Republic reports the use of the business frame by private entities for marketing purposes. Greece reported making segments of the business register available to non-governmental bodies for use for surveys but also data analysis, in the limits of confidentiality rules.  In Korea, public and private institutes can use the business frame to carry on fact finding surveys on SMEs and their human resources, calculating SME business indices.  In the case of New Zealand, other government agencies can fund surveys conducted by Statistics New Zealand.  Interestingly, the results of such funded surveys fall into the public domain after release by Statistics New Zealand.  In Sweden other public institutions responsible for official statistics and other private and public institutions conducting business surveys can access the business frame.

The update interval for the SME business frame is generally less than a year.  This is the case in Australia, Austria, the Czech Republic, Canada, Denmark, Finland, France, Moldova, the Netherlands, Poland and the Slovak Republic and Sweden.  Brazil, Greece, Germany, Italy, Hungary, Korea and Spain have their SME business frames updated yearly, as it is for the manufacturing sector in Japan.  In Belgium, Japan for other sectors than manufacturing, Mexico, Norway and Switzerland the business frame is updated less often than once a year.  Few countries report different update intervals based on business size, complexity, or sector of activity.  This is the case in New Zealand and Turkey.

Several countries report a real time update of some dimensions of their business frames.  This is the case in Denmark, Portugal, the UK and the US. In Denmark, the Statistical Business Register is updated daily reflecting the compulsory reporting by owners of the legal units.  Yet, delays in reporting cause lags in business frame updates.  Some sources for Portugal are available real time (on a daily basis), while some other sources are updated once a year.  In the UK births and deaths are updated daily, and employment and turnover yearly. In the US, updates to the frame follow administrative update cycles (weekly, monthly and quarterly).  Other sources affecting larger enterprises and multi-establishments are updated annually.  Results of the quinquennial economic census can only feed the business frame every five years.

As concerns business demography, about half the countries consider that their business frame is fully able to capture births and deaths[4]. This is the case in Australia, Belgium, Canada, Denmark, Hungary, Italy, New Zealand, Norway, Poland, the Slovak Republic, Spain, Turkey, the UK and the US.  Nevertheless,  most countries in this group signal difficulties in tracing structural changes, and delays in capturing deaths, often linked to communication delays with administrative sources.

Births may sometimes be over-estimated as a change in company name, ownership or business type may be interpreted as birth.  In the UK, the smallest businesses are excluded from the frame, therefore many births and deaths may not be captured.  However for those which are captured the quality of data is estimated to be very good.

Some countries report that the business frame is better able to capture births than deaths.  This is the case in the Czech Republic, France, Greece, Mexico, the Netherlands, Sweden, and Switzerland.  Where the quality of information concerning deaths is weaker, alternative sources, if available, are used.  In Austria, Finland, Germany, Japan, Korea, Poland, Portugal, both births and deaths are partially covered.  In Korea deaths are deduced comparing participation to surveys over two consecutive years.  Absence of survey reply in the second year is seen as an indication of end of activity.  Similarly, in Greece the new register is compared to the previous year’s using the tax number of the enterprise as key variable.  Births are therefore easily identified, while a dead enterprise may still have a tax number and not be identified as dead.  In Brazil, Hungary, Portugal absence of operation signs for SMEs is assumed as death, thus delaying capture.

Austria finds insufficient information in administrative sources to improve traceability of changes.  In Finland, the coverage of very small business is found as needing improvement.  In Germany there is a long time lag before births and deaths appear in the register.  Hungary considers that the unique ID number for each business in the Register, tax office and statistical office simplifies capturing births and deaths. Only administrative data are available for SMEs with 1-4 employees, the statistical office is therefore fully dependent on administrative sources for updates. In Japan, updates are based on the Establishment and Enterprise Census that is conducted every five years.  Five years is also the frequency of the Economic Census carried out in Mexico, resulting in a poor capture of deaths.  In the Slovak Republic the cause of the birth or death is difficult to identify.  Merger, takeover, split off, breakdown, change of legal form or real birth and death are recorded in the same way.  This is also the case in Portugal where the business register does not keep record of changes of ownership, mergers or break-ups.  In the Netherlands, units do not report cessation of activities.  In Portugal, access for the NSO to fiscal information on sole proprietors was ceased in 1999.  The business register can therefore not be updated for this category of businesses, nor can births be captured.  There is still place for improvement in the economic classification of businesses that are obtained from administrative records.

In more than half the responding countries the quality of the business frame is considered as fully appropriate.  Where it is not, countries have identified different priority areas for concentrating future efforts.  Improvement in coverage, classification, traceability and timeliness are identified tracks for future work.  Some countries target improvement of coverage of small businesses, the inclusion of employment information or better contact information in the business frame.  Others consider improving business classification by industry and tracking changes in economic activity.

Ability to trace change through a unique identifier code for the local unit/establishment would allow to better capture unit activity and de-registration.  Similarly the introduction of links between legal units in enterprise groups and groups of companies would enhance identification of changes in size, ownership and location for units involved in events such as mergers and splits.  As concerns timeliness, more frequent updating of the business register would allow earlier recording of deaths.

Concerning reported new developments, Japan sees the use of administrative sources as a means to improve the accuracy of the business frame. The Statistical Frame of Establishments and Enterprises should provide a population list as from 2004, feedback from SFEE should also improve accuracy of the Establishment and Enterprise Census. Similarly, Korea is planning to develop a business frame.

Box  3:  Business Frames for Structural Business Statistics

III    Data collection and compilation strategies

In the majority of countries which responded, the NSO is fully in charge of the collection of official statistics. This is the case in the Czech Republic, Finland, Greece, Hungary, Italy, Mexico, Moldova, the Netherlands, Norway, Poland, Portugal, the Slovak Republic, Switzerland and Turkey. The NSO is partially in charge of official statistics in Australia, Austria, Brazil, Canada, France, Japan, Korea, Spain, Sweden and the UK. Germany and the US are the only countries in the sample under review where the NSO is not in charge of official statistics on SMEs.

Other institutions collecting data on SMEs generally do so without a mandate from the NSO.  Ministries, interagency bodies or research institutions in charge of SME are often other occasional or systematic collecting institutions.  The NSO is often associated in the collection either with a consultative role in the design or edition of questionnaires, or provision of the frame.  The NSO can also coordinate sampling or receive and treat data collected by other bodies to merge with other sources and perform new analyses.  In other cases these institutions may sponsors annual compilation of data by the NSO.

Almost all countries have different treatment for core and specific statistics.  But this is not the case for the Czech Republic, Mexico, New Zealand (where statistics on SMEs are not compiled separately), Norway, Portugal and the US.  The Czech Republic collects all variables annually through a comprehensive questionnaire.  In Mexico, variables are collected through a quinquennial census.  Norway uses the Structural Business Statistics framework through statistical and administrative data. In Portugal, two Eurostat-led data collection exercises yield most of business statistics variables.  Structural statistics have been improved as a result of Commission Regulation nº58/97 as it precisely defines the contents of the data set to be produced.  However, they do not give any specification of variables on SME.  Only core variables are presented in the Statistics of US Business, these come from the Census Bureau’s Business Register that integrates administrative, survey and economic census data.

Many countries report the aim of minimising sample surveys.  Some countries have developed tools to monitor the response burden.  The strategy developed is therefore to collect core variables through the integration of census/surveys based data and administrative data.  When they are not collected together with core variables, specific variables are added to other collection exercises or collected through specifically designed ad hoc surveys.  This generally applies to all businesses.  Variables for smaller businesses (micro-enterprises) are sometimes estimated, or collected through sample surveys.  Some countries compile longitudinal data to estimate variables.  Some countries consider cost sharing for specific variables collected on demand and on an ad-hoc basis.

In Japan, SME related statistics that are surveyed by other governmental bodies are reprocessed at the NSO for core statistics. Specific variable are collected using single year sample surveys on topics covered by the White paper on SMEs.

Obstacles encountered in SME data collection are as diverse as low response rates; limitations of response burden; timeliness; poor quality of sampling frame; the legislative framework as it limits the variables that can be collected; small sample size; acceptance from respondents; actual availability of variables and level of detail.  The low response rate and the size of SME population are the most commonly quoted obstacles.  The quality of the business frame, poor use of data collection tools, volume of data collected, and compiled, poor coverage of the smallest businesses are less frequently reported.  Insufficient resources for surveys have also been mentioned.

Australia, Belgium, the Czech Republic, Finland, the Netherlands, New Zealand, Sweden and the UK have improved their use of administrative data, in some cases using a common data collection exercise, using unique statistical and administrative definitions In Australia the household based survey has been extended and the input data warehouse (IDW) is used to store administrative data received from the Australia Taxation Office. The IDW facilitates increased data confrontation and sharing. In Belgium the improved use of administrative sources resulted in a 50 to 70 percent reduction of the amount of questions asked.  In Finland the sample sizes have been decreased.  Finland, Japan, Moldova, the Slovak Republic have improved their communication tools with questionnaire respondents through clearer, simpler and more user friendly electronic questionnaires, better information on data use.  In some countries, following a failure in response to statistical surveys the respondent is interviewed.  Hungary sees the provision of their SME statistics to international and regional organisations as an improvement.  Portugal improved the methodology of the Structural business survey and the use of data estimation.  In Spain, sample coordination has reduced the response burden.  In Sweden, questionnaires on economic data have been adapted to the accounting systems of the enterprises.  In the UK methodological changes have allowed burden reduction.  The same objective has led extension of electronic reporting for the Economic Census to all businesses in the US.  Also, in a pilot project for smaller single-establishment enterprises, the Census of wholesale trade and the annual trade survey have been consolidated to avoid duplication of data collection.

Future development plans in Australia include the further development of the input data warehouse to include unit level data. It is also envisaged to develop a business longitudinal database.  Better use of administrative data is also relevant to Austria, Belgium, Brazil, Finland, Hungary, Italy, Moldova, New Zealand, Portugal, the Slovak Republic, Spain Sweden, Switzerland, Turkey and the UK.  Korea plans to make an obligation to register SME balance sheets.  The Netherlands plan to improve the business register and develop integration.  In Portugal, Spain, Sweden, Turkey and the US improved electronic data collection tools are considered.  Portugal plans to use estimation procedures for small enterprises.  Sweden will introduce cut-off samples with models.  Switzerland considers multiple survey collections and Internet-based techniques for data collection.  Data linking is considered in the UK.

Box 4:  Collection and compilation strategies

IV   SME data linkage with administrative sources

Very different access patterns for NSOs to individual records could be identified, ranging from full to no access.

In Australia, Belgium, Canada, Finland, Germany, Hungary, Italy, New Zealand, Norway and Sweden, the NSO has full access to individual administrative SME data in practice.  In Austria, France, Greece, the Netherlands and Poland, although the NSO is entitled to access individual SME data, this happens only partially in practice.

In Brazil, the Czech Republic, Moldova, Portugal, the Slovak Republic, Spain, the UK and the US, NSOs have only partial access to individual identifiable administrative data.  Japan, Korea, Mexico, Switzerland, Turkey do not grant the NSO access to individual identifiable administrative data on SMEs.

Where access to administrative and other sources is partial, differences in the definition of variables have been commonly seen as a major impediment to the use of administrative sources, different observation units, classification and the absence of a unique identification number have also been mentioned.  Sometimes technical problems hinder access to administrative sources.  Countries generally have identified the main sources they would like to access.

The absence of a unique identifier, differences in definitions, and timeliness of data are reported as limitations to the use of administrative data.  Austria is an example of use of administrative and fiscal sources.  This has been possible thanks to the adequation between the units of the register of enterprises and the units of the social security administration and fiscal units.

Administrative sources have direct and prominent roles in countries and areas of activity where data substitution and sample supplementation takes place.  This is the case in Australia, Denmark, France, Sweden, the UK and the US for core variables for all activities, Finland for core variables and some specific variables for all activities, Hungary for SMEs with 1 to 4 employees, Moldova, the Netherlands (excluding agriculture), and Norway for all variables for all sectors.

Administrative sources have direct but limited roles for the imputation of non-response and sample supplementation in Australia.  Austria and Italy use administrative sources for core variables for all sectors excluding agriculture, Canada for core variables for all sectors, Finland for some specific variables for all activities, Sweden for specific variables covering all activities.  Administrative data are used for estimating core variables for all activities in Portugal.

Administrative sources are used for quality control in Austria, in the Czech Republic for core variables for all sectors, in Finland for some specific variables, in Greece for core variables in some sectors and in Hungary for SMEs with 5 to 249 employees.  Australia uses them for data confrontation with the view to improve data quality.  Administrative sources intervene for the estimation of figures for enterprises for Belgium.  In the Slovak Republic, administrative data are only used for updating the activity status of units.

Box  5:  Access to and linkage with administrative sources for SMEs

V    Dissemination strategies for SME statistics

SMEs are often not direct users of official business statistics, while they usually make use of macroeconomic indicators to define their short-term plans.  SMEs may prefer to use reports or studies carried out by consultants, rather than accessing directly products provided by NSOs.  Sometimes they may also be unaware of the existence of such statistics or do not know how to access those. Therefore, they often feel burdened by statistical surveys, without seeing benefits from their contribution.

A wide dissemination of SME data is, therefore, important to meet user needs.  Rapidity and cost effectiveness are criteria to be taken into account.  Countries generally disseminate data annually.  Pricing policies largely differ and both free-of-charge and charged practices were observed.

Dissemination products are designed totally by the NSO for all countries in the sample except France, Korea, partly the Netherlands, the UK and the US.  In Austria, SME statistics are contained in other publications.  The same time lag for publications occurs for core and specific variables.  It takes a great majority of countries between 12 to 18 months to publish SME data; few countries publish data for reference years older than 18 months or together with publications with a shorter periodicity.

Few countries report no specific interest in SME statistics. User interest and needs are quite diverse and NSOs have adopted different response strategies.  Australia has conducted a household- based survey to gather demographic and structural data on areas such as home-based businesses.  Austria, the Slovak Republic, Sweden, Switzerland, Turkey and the US compile specific products on demand.  Canada and New Zealand have a specific programme for gathering data on SME financial aspects.  In France the statistical department in charge of dissemination is integrated in the Ministry that conducts SME policies.  Portugal also considers the Structural Business Survey results as key vehicle for a comprehensive dissemination of SME data.  Switzerland intends to improve structural business statistics to more fully capture breakdowns by size-classes.  The UK has launched the Urban Renewal Programme to meet demand on SME statistics.

Interest in or demand for SME statistics in conjunction with other statistics than business statistics has driven Australia and Japan to envisage specific longitudinal databases.  A prototype longitudinal database should be available in Australia by fall 2004, followed by a production database in 2006. Italy aims to produce a structural data warehouse.  The Netherlands plan to define a special view on their output database.  New Zealand has followed a different approach in the way that several government bodies have sponsored a single economy-wide survey on several aspects of business activity.  Cost recovery seems to be an accepted practice for such cases.

A particular important demand for SME statistics is related to the employment and finance dimensions.  Here, the aspect of data linking and, possibly, embedding demographic variables into the comprehensive inquiry frame of structural business statistics could be of strategic importance.

Box 6:  SME dissemination and data demand

Concluding remarks

This comprehensive stocktaking exercise has delivered highly interesting results for follow-up and further thought and consultation for devising a strategy for improving the information base for SMEs. The elaboration of an accepted target definition of SMEs, respecting differences in national practices and perception, but allowing re-aggregations across common size-classes and better comparability across surveys and sectors as well as a recommended choice and definition of variables would provide a good starting point.  It is this aspect where OECD is called upon to “bridge” between EU and Non-EU economies of OECD, but also those participating in the so-called “Bologna Process”.

The questionnaire replies revealed that the NSOs generally pursued a user-driven approach by consulting public and private stakeholders for their survey design. Issues of concern primarily concentrated on excessive response burden, possible duplication of data collections, a generally low response rate and sometimes questionable data quality.

Insufficient feedback to SMEs was recognized as well as sometimes inadequate data availability and breakdowns. As key obstacles figured the low response rate which is, of course, linked to the sheer size of the survey population. Proposed strategies to improve this situation include an increased use of administrative data, a better integration of registers and detailed inventories and documentation.

Although the quality of business frames was generally considered as appropriate, concerns were expressed as to the lack of comprehensive coverage, confidentiality issues and allocation difficulties of activities and industries.  Difficulties in correctly measuring the entry and exit were often considered as impediment to tracing changes.

In many countries, the importance of NSOs in coordinating surveys has been recognized.  They could probably play an even more important role in driving a process to allow combination of sources and elimination of duplication in data collection. The notions of “information system” and “input data warehouse” deserve further thought and follow up.

NSOs often reported that, despite access to administrative data, its use and usefulness were limited because of incompatible basic definitions, different classifications and thresholds and lack of a common identifier.

SME data collections do not seem to present distinctively different characteristics from other data collections.  They are generally regarded as an additional dimension to structural business statistics.  A number of respondents stated the need to include the dynamic dimension (demography) into the more static (structural) dimension.  A clear requirement is the possibility to link up statistical systems, in particular with respect to employment and finance.

ANNEX: SELECTION OF QUESTIONNAIRE RESULTS

Table 1:  Specific obstacles to SME data collection

COUNTRY	Low response rate	Size of SME population	Poor use of data collection tools	Volume of data collected	Volume of data compiled	Other
Australia		X	X	X	X	Small sample size, Provider load consideration, data availability, lack of sample design in many “business surveys’ to collect non-employing businesses, lack of dedicated small business survey
Austria	X		X	X	X	Low unit-non response and item non response
Brazil
Canada						Not SME specific. Limited sample sizes resulting in lower data quality. Currently lacking business size indicators for data on the supply of financing.
Czech Republic	X					Not SME specific
Denmark						Not SME specific: as we primarily use registers the major obstacle is to identify and remove dead enterprises before sending the questionnaire.
Finland	X	X				Therefore large sample size
France						Not SME specific
Germany
Greece	X	X				Many data of various kinds are asked from SMEs. In some cases like in the Manufacturing sector SMEs are surveyed exhaustively because their number is limited (over 10 employees). This means that in any particular data request we have to survey the same enterprises and this adds a lot to their burden. Furthermore, SMEs are not very well organised and some of the data requested cannot be easily available. This makes them unwilling to collaborate and surveys become very expensive as the collection of data via interviews becomes unavoidable.
Hungary	X	X			X	The major difficulty is the collection/compilation of data of SMEs with 1 to 4 employees. These data have to be derived from tax data. The development of methodology and variables are under way.
Italy	X	X		X	X	The particular Italian productive structure which is based on the micro enterprise (95% of the total enterprises has less than 10 persons employed)
Japan	X	X				Low response rate, difficulty in acquiring sufficient budgets for surveys
Korea						Not SME specific. Regarding BSI surveys for economic status, it has been found that some cases show distorted results because respondents are psychologically affected by reports on government policy changes and by media reports on economic status. As a result, surveys conducted during economic downturn tend to produce results that point to worse economic situation
Mexico		X		X	X	The main SME data collection obstacles are the costs
Moldova		X	X			Timeliness of getting results
The Netherlands	X	X				Poor quality of sampling frame, low response rate.
New Zealand						Not SME specific. One of Statistics New Zealand’s strategic goals is to maintain the cooperation of data providers.  This includes managing compliance loads such that our direct surveying activity is minimised.
Norway						No specific
Poland	X	X		X	X	The most important is limitations in information for micro – enterprises. According to polish law micro –enterprise can lead one of  four type of book keeping. In the simplest version we can only collect information on turnover, expenses and employment.
Portugal			X			The quality of the business frame in what concerns the identification and characterisation of the enterprises (example: legal form, addresses). Portuguese administrative regulations tends to allow less demanding accounting rules to very small enterprises (companies with less than 50 persons employed and sole proprietors) and reduced fiscal forms. Therefore the collection of economic data on these enterprises has becoming less accurate
Slovak Republic	X	X				Low response rate – approximately 70%
Spain						User demand on special disaggregations at very small domains. Response burden
Sweden	X	X				Non response, timeliness, errors in economic activity classification of enterprises with less than 10 employees
Switzerland	X	X		X		Although these obstacles are not SME specific they are particularly acute for them - Acceptance by enterprises - Availability in enterprises of collected variables -Statistical burden- Level of detail (88% of all enterprises are micro-enterprises with less than 10 Full time equivalents) - Confidentiality
Turkey						Not SME specific Compilation of SMEs data by size class, four digit activity level and provinces level. Absence of budget for special survey on SMEs. Low response rate Volume of data collected
UK		X				Issue of burden. We have a Survey Control Unit to monitor the compliance costs and burdens on businesses. Of motherly rules mean that small businesses are surveyed less frequently than large (generally only every 3 years). Burden would be so bad on small businesses it must be managed. Sources – do not hold local unit details and this affect sub national estimates for multi site SMEs, we get round it using statistical methods. Administrative Sources – their approach is inconsistent with a statistical register (e.g. Multiple VATS
United States
Count	12	14	4	6	6

 

Table 2:  Most important cases of differences

between administrative and statistical sources

 

COUNTRY	Definition of variables	Statistical unit	Classification	Absence of a unique identification number	Timeliness	Comment
Australia						The Australian business number is used as key identifier for legal entities of business units
Austria	X	X	X
Brazil			X
Canada				X for non incorporated businesses
Czech Republic	X					Better access to current sources and access to the employment office data
Denmark		X		X		It is possible to organise the enterprise/the business in more than one legal unit (i.e. more than one identification number). This can give us units with a lot of employees but no turnover and vice versa. This problem occurs almost only in case of bigger enterprises and almost newer for small ones. The small units are generally easier to handle - for instance when matching different data sources.
Finland	X	X
France						Generally the correspondence with administrative concepts is quite satisfactory. Information on financial links collected by the tax administration is currently not available, due to technical and not legal problems.
Germany			X		X
Greece	X		X
Hungary	X
Italy	X	X	x
Japan
Korea						Source data on import/export license provided by Korea Customs Service or corporate credit information of financial institutions.
Mexico						Better access to data available at the Secretaría de Hacienda y Crédito Público (SHCP) and Instituto Mexicano del Seguro Social (IMSS)
Moldova	X			X		Better access to tax and social security registers
The Netherlands	X	X
New Zealand
Norway
Poland						Need to access social security data
Portugal						There is a legal barrier to the access of individual fiscal information. This counters updates of the business register
Slovak Republic	X	X	X	X		Need to access the tax register at the individual data level.
Spain			X	X	X
Sweden		X				Definitions and coverage
Switzerland						Better access to social security registers and data. Customs data, tax data: VAT in particular
Turkey						Ministry of Finance, Ministry of Industry, Chamber of Industry, Chamber of Commerce, Turkish union of Chambers.
UK						Self employed system, corporation tax system, flat rate agriculture. Self employed system (Tax returns data)
United States						The Census Bureau does not collect data that specifically target SMEs
Count	8	7	7	5	2

 

Table 3:  Business Frame Coverage

COUNTRY	Admin. unit	Type of activity unit	Enterprise	Kind of activity unit	Local unit	Establishment	Location	Legal unit	Enterprise group unit
Australia	X, ABN unit	X
Austria			X	X	X
Belgium			X
Brazil
Canada			X			X	X
Czech Republic
Denmark			X	X, as defined in council regulation 696/93	of the enterprise, as defined in council regulation 696/93
Finland				X	X			X	X
France
Germany			X		X as defined in council regulation 2186/93
Greece			X, as defined in council regulation 696/93
Hungary								X: companies, partnerships and sole proprietors
Italy
Japan						X. as defined as a single physical location where economic activities such as the production or supply of goods an d services are conducted
Korea
Mexico
Moldova			X		X			X
The Netherlands			X						X EU definition
New Zealand
Norway				Local kind of activity unit used as establishment				X used as enterprise
Poland								X
Portugal								X regulation 696/93
Slovak Republic			X in line with council regulation 696/93		X in line with council regulation 696/93			X
Spain			X Unit with capacity to implement one or more economic activities
Sweden
Switzerland			X in line with EU regulation		X in line with EU regulation
Turkey
UK
United States			An enterprise is an economic unit comprising one or more establishments under common ownership or control in the business register’s implementation. This unit represents the top-level US parent company and all subsidiary US companies of which the parent owns or controls a majority (more than 50 percent) interns plus all EITNs and establishments affiliated with the parent and its US subsidiaries			An establishment is an economic unit, generally at a single physical location, where business is conducted or where services or industrial operations are performed> Examples incl. a mine, factory, warehouse, sales office, grocery store, bank, hotel movie theatre, doctor’s office, museum and central administrative office.