Demography Of Small and Medium Enterprises

5th Working Group Meeting

Report of Meeting

“Information” session

There has been approval of funding for DOSME for 10 countries. However, pending a tender process for the project management the money is not available. The 10 countries are the pre-accession ones. Albania and Macedonia are not included, but may take part in a parallel “Balkans project”. In practice this may mean that they will continue in the project but be funded from another source. Balkans project finance is, however, likely to be even slower than DoSME2, and there may need to be an advance from the participating countries to undertake B5/6 on the promise of later repayment. Arto Luhtio will keep the project in touch with developments.

TES courses are also subject to tender arrangements in the future, and there may be regional or national courses in future. CECs should soon receive an invite to a course on statistical units in Luxembourg in Oct 2000. Eurostat will fund attendance by CECs. A specific TES course for Bulgaria is also planned for May. The business registers course that is held annually in Oslo is not certain to continue this year. There is in any case no funding for CECs for this course.

Additional information on various topics is available from the Eurostat web site http://europa.eu.int/comm/eurostat. Information on the next roundtable on business survey frames will be available, probably from autumn 2000, from www.stats.govt.nz/nzroundtable. The meeting takes place in Wellington at end of October/ early November. The next meetings will be in the USA in 2001 and in Portugal in 2002.

EU work on business demography will be organised through three separate sessions with 5 member states in each. These will be brought together at the June business registers meeting. There should be a clear route for the DoSME project to contribute its experience – Arto Luhtio would assist, but John Perry should be involved too.

Document 63 contained information relating to the first Structural Business Statistics PHARE project. The first project meeting will be in Warsaw in July (and the second in Slovenia in the autumn?). There will be a joint DOSME2/ SBS project seminar in Autumn 2001. The Netherlands National Economics Institute (NEI) is the current contractor for the SBS project

There had been a Eurostat meeting on enterprise groups in Luxemburg in April. Eventually it is hoped to have an extra chapter in the registers manual for enterprise groups, and to amend the business registers regulation, but this may take some time to achieve. Directors of statistics in member states have endorsed his approach.

The forthcoming business registers committee meeting on 26/27 June in Luxembourg is scheduled to cover enterprise groups (from the April meeting), demography (feedback from the 3 sessions), coverage of NACE sections N and O, and implementation of the business registers regulation (the member states questionnaire). The UK completed questionnaire was shown to CECs for information.

CECs were requested to provide information to Infostat on any courses or seminars that take place in relevant areas. Eurostat will examine publicising on their web site (BRNET within CIRCA).

Progress on B4 (doc 57)

The report covering the main phases of B4 was presented. The main issue raised was the contents of the CDs to be provided to close the project:

	for Eurostat and experts all central databases, documentation and the web pages
	for Infostat all data including ancillary data files, all counts, documentation, web pages and packages; and
	for the countries their own central database, documentation and the website.

Countries will need to archive their own input data (register extracts and so on); if CDs from Infostat are in open form, these could be added if a suitable CD writer is available. There was a question of how to deal with the B4 publication, since the data must be written to CD before they are published.

Individual Country Reports (key points only – Paul Smith)

Albania

	same number of businesses in each trimester was expected, but not achieved. This would be checked.
	42% of new creations in Tirana; earlier high concentrations in the capital were related to the earlier political collapse, but this does not seem to be true for B4 – there were no obvious external influences.
	no specific economic changes to report; survey ran smoothly.

Bulgaria

	There is a new organisation in 1999 for BULSTAT; the sample had been taken under the old organisation.
	Additional questions had been added to the B4 questionnaire.
	Employment information is only available on BULSTAT with a 1-year delay, so can’t be used for the sample base. Hence the problems with outliers.
	A new statistical law from 17/6/1999 means there is now a single code for each business, so other sources will be available for sampling.
	Two big (>500 employment) units were used as donors in imputation. This highlights a general need to restrict the size of potential donors.
	there is still a high percentage of units with start dates pre-1994 among new registrations.

Czech Republic

	Problems with foreign natural persons creating businesses – higher than before, and hard to contact
	More information in B4 was obtained from indirect controls
	Imputation was done the same way as B3 (NB slightly different approach to other countries, but fine, see earlier notes)
	Economy not buoyant – GDP falling, inflation and unemployment up (and see report)

Estonia

	Postal response rate was low. A change of the Estonian postcode system may have contributed.
	Improved data quality by using an administrative source to correct the main data file; useful in checking sector of activity
	A new law on social taxes has been introduced, which will increase the rate of natural person business creations
	May be trouble for later surveys, as there is no employment information in 1999 – it will be available with a 1-year delay

Hungary

	Survey running smooth, with lower imputation rate than previously.
	Economy doing well – GDP shows 4.5% growth in 1999.

Latvia

	Difficult to contact foreign legal persons, often not resident in Latvia – in some of these cases there is only information from the taxpayers register about activity
	From 1 January 2001 there will be a new law on commerce, which will change the commercial and registration processes, and may induce many new registrations.

Lithuania

	Pre-1990 there were only special lists and no registration; post-1990 there was a registration system, but it was complicated and many dormant businesses did not register
	only after 1996 have banks been in the register. Some banks with >50 employees have been included in B4 (implies discontinuity)
	no “limited liability company” category in the Lithuanian register – just with or without foreign capital
	the register is specially prepared from several sources for DoSME, and there is sometimes a mismatch of information
	there are many problems contacting sole proprietors, and in many cases they refuse to participate in the survey
	there are doubts about the cessation dates – many businesses “temporarily cease activity” [sic] at the time of the survey through a distrust of the tax and government systems.
	there is a lag between registration date and the start of activity, which may be to do with understanding the questionnaire (and also the first bullet point)
	in 1999 (the year of the B4 survey) there were three prime ministers and their cabinets in Lithuania, leading to a mistrust of governments. This is also reflected in low GDP and exports, and a big budget deficit.

Macedonia

	Only 204/2000 postal responses received, so most contact with interviewers
	Alternative addresses for some businesses were found from tax offices and similar
	Many businesses were surveyed from indirect controls
	GDP in 1999 was influenced by the Kosovo crisis, but still the economy grew in several indicators (official GDP estimates – 1998: US$3540m, 1999: US$3635m)

Poland

	Sample nearly the same but number of dead units decreased, and smaller non-contact rate. Non-response unchanged, however.
	New regional organisation from January 1999 with 16 regions (instead of 49), though the original offices still exist and form another tier
	Foreign owners investing in Poland agreed to keep employees for several years, but in many cases this period is just ending, increasing unemployment and business creations
	social security reform has obliged enterpreneurs to pay more; many have responded by asking employees to open their own businesses which receive raw materials and create outputs, leading to an artificial proliferation of businesses. A new code for these businesses has been added to the register
	there are many “newly-created” enterprises in health protection, which are actually budgetary units for state health protection which are now partly independent of (and partly dependent on) the state.

Romania

	There was a reorganisation of national statistics in 1998, and about half the DoSME team changed
	The tax authorities previously identified temporary cessation as permanent cessation, but this was changed during 1999, and quality is expected to improve
	3.2% GDP growth (1999 on 1998)

Slovakia

	Economy worse in 1999 than 1998 (except balance of trade).
	About 20% fewer creations in B4
	4 outliers with large employment – one split from a former company, one newly formed from a state enterprise, one new limited company and one stock company created from a co-operative (so 3 changes of legal form, not true creations)
	619 delayed registrations through change of activity of businesses were added to B4

Slovenia

	New business register was formed from two other registers in 1997, but some units were entered on the register with a delay
	Started to enter doctors and dentists for the first time (hence more “creations” in 1998)
	Telephone follow-up of partial non-response; partially successful
	1 extreme outlier was the result of a response by the wrong enterprise (!) and should be treated as inactive

Other general discussion points

	Latvia want DoSME 2 to include money for translation of working documents
	Exchange rate fluctuations can cause problems for participating countries (especially with a weak Euro)
	A session in DoSME 2 on survey quality would be useful

Comments from CECs (John Perry):

Albania

Used DOS version of imputation program

Checked data using SPSS

Recorded IACTIVE to 0,1,2 (not 1,2,3 as in the written report)

Bulgaria

Re-registration is a problem

Additional questions were included in the questionnaire (impact?)

Employment was added to the register after the register extract was taken (could this be used for the CDB?)

Imputation was a problem but resolved

Czech

Foreign natural person a problem resulting in more indirect estimation

Economic situation is more optimistic

Estonia

Low response to mail questionnaire

Extra cross-checking of data introduced

Expected more legal persons (as was case)

Number of employees is now included on the register only after a year’s delay, thus affecting the register extract

Hungary

Very high contact rate this year

Economic improvement noted

Latvia

Difficulties in making contact (foreign owners as in Czech)

Need extra money for survey (getting expensive to run)

Economy now stabilising after “Russian” crisis

Some changes in law will affect future surveys (not B4)

Lithuania

Change in de-registration process could affect results

Banking now added to the register

Doubts about the accuracy if inactivity dates (businesses mysteriously die at the time of the survey)

Economic situation poor

R. Macedonia

Low mail response

Poor addresses also affected actual contact

Lots of dead units

Kosovo could have affected classification of businesses

Upward trend noted in economy

Poland

New arrangement of regional offices (now only 160

Many non-responses (could have affected imputation)

No confidentiality or outlier problems

Large number of newly created enterprises because of growing unemployment and to changes in contracts of workers (going from employee to self-employed)

Less investment and increased demand difficulties noted

Too many enterprises in health sector through restructuring of state enterprises

Romania

Also followed up the dead units (30% actually found active)

Used DOS version of imputation

Checks done of source data resulted in finding data entry errors on 150 units

Slovakia

Percentages quoted do not quite add up to those in document 58 (weighting?)

Four large units are not real creations

Some reactivations (as in Romania)

Economic situation not optimistic

Slovenia

Doctors etc now added to register (explains some lags and large RY 08 value)

Some combination also of two business registers

Quality

Document 62 needs an extra paragraph to define “ghosts” and the base-line issues. The C survey is to be excluded from the graphs. B4 is to be added to the report and graphs.

B4 Publication

Changes were agreed as follows:

Drop 1 January 1995 columns
Consider using only sole proprietors and partnerships for tables 18 to 21 (B4 only – there is still and issue regarding earlier years)
Consider using only legal persons for table 22 (B4 only – there is still and issue regarding earlier years)
Combine primary and basic vocational education for Albania
Add definitions for all tables
Add table of RY distribution to methodology
Comment on Republic of Macedonia (Eurostat to confirm this name) and include set of tables as in document 58, subject to separate discussion with Republic of Macedonia
Comment on error in table A of B3P: “B2” instead of B3 in heading and republic of Macedonia entries wrong (active/ inactive counts switched and counts instead of percentages)
Include B4 questionnaire and comment on errors in B3 (1999 instead of 1998 dates in 1) and 16)

Answers to specific queries on data in the tables were:

Table 2, Estonia. Due to the new social law (see commentary)
Table 7, Latvia. High part-time figure shows the influence of a high rate of unemployment.
Table 7, Lithuania. Social insurance for employers has increased. Many employees are now asked to work part-time so that employers don’t have to pay social security.
Table 9, Romania. This is a real effect from a new fiscal policy, which encourages smaller enterprises. This ties in well with the increase in physical persons in Table 2.
Table 9, Slovenia. Enterprise creation is becoming more stable, more can afford their own premises…
Table 11, Albania. See earlier commentary.
Table 12 Hungary. (to be completed)
Table 12, Slovenia. A side effect of adding doctors and dentists to the business register in 1999. Not a real change.
Table 14 (general). What should the definition for this table be?
Table 14, Slovenia. One outlier affected the 1998 data, which may need revision.
Table 14, Romania. (to be completed)
Table 20 (general) A warning about different definitions/interpretations could be added.
Table 21. Separate A and B surveys into different tables (superseded by decision to drop survey A information).

Polish Publication

The working group agreed to the publication by Poland of a comparison of the five first wave of accession countries:

Consider putting only graphs (no tables) in the publication
Look at the unweighted estimated for 50+ employment units to determine sensitivity of weighting (small samples/ large variance)
Examine further the non-actives (which included both closed and not trading) to ensure that the explanation is correct.
Include definitions (as for B4) of content of graphs/ tables – should be consistent with proposed revised treatment of B4.

We need to consider adding 1998 (B4) data at some stage. Poland will publish following clearance from Eurostat and project co-ordinator. Eurostat will check on procedures. Eurostat will publish only electronically on the web site. Publication is to be ready electronically by 3^rd week of May.

Some interest expected in Eurostat and in the EU DGs concerned with enterprise and enlargement

Longitudinal estimates

The impact of “dormant” estimates on the population estimates was demonstrated to be major and could not be explained. We cannot progress the work without an explanation.

Issues raised country by country were:

Albania

survey includes follow-up of dead units so few quality problems

Bulgaria

quality of cessation dates is a major problem (the years could be ok where present but they were not used in the estimation)

Change in register causes extreme difficulties

Some extra dates are held on the national register but have not been added to the central database (Infostat need to consider how to deal with this)

Czech

Treatment of dormant units has produced unacceptable estimates

Estonia

Some corrections have been made to date (Infostat to confirm also in CDB)

Hungary

Some dates entered wrong way round (year – month)

Some dates only as years (December entered for estimation purposes)

Latvia

Problems caused by the A/B1 survey procedures (units missed from A were added to the B1 survey)

Lithuania

Inconsistencies between B2/C1 were removed (also on the CDB)

B2 low active rate could cause problems

Poland

[see Word document]

Romania

Natural persons unlikely to be dormant

Slovakia

Survey dates were often used as “real” dates in the actual survey

Slovenia

An extra (IIb) improved version was produced with new cessation dates from administrative sources (not added to the CDB)

Stricter control is needed on CDB for dates and the B5/B6/C2 questionnaire design needs improvement to deal with these issues.

The following should be noted:

	The use of the word “survey” in the tables is misleading: it is actually the “questionnaire” used in the survey. Thus “B2” refers to B2 for new registrations (the B2 survey) and also to first contacts from the A survey through the C survey.
	The study looks at the issues from an economic theory point of view.
	The estimates for 1996 dormants and active are not correct. Thus the dip noted in numbers of businesses in 1996 is not real. This is because “dormant” businesses in the A survey were resurveyed only in the C survey in 1997. As a result if they were actually dead they were given a death date in 1997. 1995 would have been more appropriate. This would also affect the estimates of dormants in 1995 (currently too high).

Hardware/ software issues

The data entry package can be rewritten very quickly for the new working environment. No action will be taken until DOSME2 contracts are in place

Bulgaria does not have access to SAS or SPSS. This needs to be rectified.

Longitudinal database

Presented on a slide show.

Issues:

	Date calculation does not work if months are missing. Need some rules for this.
	File type of four characters (eg A1C1) is sufficient to explain the different surveys and forms but it may be necessary to extend this for C2. No action will be taken until contracts are in place for DOSME2.
	Consistency of data remains an issue, especially of dates. We need to consider further what needs to be done.

Table 1 of the B4 publication

Republic of Macedonia (name to be confirmed for publication by Arto Luthio) data are to be included in the table with a 1 January 1999 creations figure in the first column and an appropriate footnote.

Table 1 is to be moved to the main text of the report, as it is on a different basis from the other tables. This will help the reader.

GDP figures will be presented in ECU (as they relate to 1998) per head and PPP figures will also be included. The correct figure for Slovenia is 8756 ECU per head in 1998 Republic of Macedonia will provide figures in $ for conversion.

Options for calculation of table 1 estimates were considered on the basis of an example of data for Slovenia, which illustrated the problems of:

	Re-registrations
	Changes in rules for registration (impact of adding doctors)
	Lags in registrations

	Survey totals	Latest yearly totals
Direct estimates
Adjust for re-registrations
Adjust for future surveys
Adjust for dormant units
Adjust for re-registrations and future surveys
Adjust for future surveys and dormant units
Adjust for re-registrations, future surveys and dormant units

DOSME 2

Only options 4 and 5 were considered in any way acceptable for the DOSME 2 project. The gap in funding causes major problems of continuity for the DOSME surveys. The lack of a B5 survey in 2000 can be met only by making contact with B5 units in 2001, 12 months later than for previous surveys. This will require special care with the questionnaire and the collection of information for dead units. The DOSME 1 project had already shown that it was difficult to collect the vital death dates only 9 months after the end of the year of registration. A further 12 months delay can only reduce the quality.

Three countries expressed an interest to carry our B5 surveys in 2000 without project funding. Poland has already placed the B5 survey in its survey programme for 2000, although finding funding within the office would be difficult. Also, Poland felt that some central co-ordination would be essential if such B5 survey data were to be useful. Republic of Macedonia also wished to conduct a B5 survey under their own funding. Their situation was compounded by their exclusion from DOSME 2 and the uncertainty of “Balkans Project” funding. Because they were late entrants to the project, they needed central support for survey processing more than other countries. Bulgaria intends also to proceed with a B5 survey, although funding has not been confirmed. Albania was also very keen to conduct a B5 survey with their own funding. Other countries felt that they could not fund B5 type surveys within their own budgets, although Slovakia, Czech republic and Hungary had all put such a survey in their year 2000 programmes.

Eurostat was urged to find a solution. Possibilities included obtaining funding from the under-spend on other current PHARE project. Arto Luhtio would investigate urgently. CECs might also investigate the possibility locally of funding from the SBS projects. Arto Luthio would also investigate the possibility of a bridging contract for Infostat to provide continuity until the DOSME 2 project starts. There was also a need to maintain communication between Eurostat and CECs during this period.

DOSME 1 Review

As this was the last meeting of the project, it was necessary to review the work carried out. The project would finish on budget and on time and had met the objectives. The B4 publication was nearing completion and would be with Eurostat earlier than an previous publication. Participants were invited to comment:

	The project had strengthened relations between the SBS and registers within the statistical offices
	The input of western experts had been very helpful
	Infostat had been crucial in ensuring the success of the data processing
	The seminar had dealt well with quality issues and the wide range of participants had been a particularly welcome feature
	The project had helped in starting up another project in the Czech republic (harmonisation of business registers under national PHARE funding)
	Much use had been made of the data in the Czech Republic – it was the only source of this type of information
	Imputation training had been excellent
	The project provides a good basis for future work on measuring register quality.
	The project had improved co-operation with those responsible for SMEs in Slovakia, Bulgaria and Republic of Macedonia.
	The survey was useful in allowing other questions to be added, thus minimising the cost of that work. A good example was in Slovakia.
	Imputation methods had been transferred to other surveys in Slovakia and Hungary.
	The survey had helped improve questionnaire design for other surveys in Slovakia.
	The transfer of knowledge on conduct of surveys had improved work on other surveys in Romania.
	The tools developed for the project were very good (Romania).
	Romania had found Salomon very useful in their general survey work.
	The project had a huge impact in Albania, improving communication in government and with other organisations.
	The funding level was correct for survey work in Albania.
	Estonia had used the project in re-organising its business register.
	Estonia had developed business demography as a result of the project.
	The methods of classifying businesses introduced by the project had improved the quality of data more generally.
	Documentation of project was considered excellent
	Organisation of meetings, and the opportunity for countries to organise meetings, was a very good feature.
	Common standards for software had been implemented successfully.
	The survey had achieved what had not been done by most EU member states.
	Co-operation between countries had been excellent: they had worked as a team.
	Continuity with PECO Panel and in the publications had been achieved, an essential feature of longitudinal work.
	This had been a unique co-operative survey project over a long period that had encouraged joint working

Some comments raised issues for the future:

	More documentation of the methodology was needed in the publications.
	The frequency of the C surveys (even worse with the delay in DOSME 2) creates problems for measuring survival rates.
	The objectives were now different from those at the start of PECO Panel and we need to ensure that continuity is achieved.
	Only Romania had adopted Salomon. It was not flexible to national survey needs and there were limitations in what it could do. Future versions need to recognise this and actively take on board comments.
	A closer link with other existing survey methods was desirable.
	The next stages should concentrate more on quality measures.
	Statistical units, profiling and dealing with mergers/ take-overs were not covered by DOSME but were important in the SBS work. Training on units should be part of the programme of work (the planned seminar?).
	There should be greater integration with other statistical work (which can be achieved by close working with the SBS project).
	More work on standardising definitions, e.g. of dormant units, was needed.
	It was difficult to manage the resource conflicts with other work.

Eurostat stated that CECs would become more involved with EU work. Communication between CECs and with Eurostat was vital. CECs should inform Eurostat of events that were related to DOSME or business registers in general.

DOSME 2 First Meeting

The participants agreed to the Bulgarian offer to host the first meeting. Those tendering for the contract would be advised by Eurostat.

John Perry

Project co-ordinator

LIST OF DOCUMENTS

Doc47final.doc:	Longitudinal estimates – draft final results (included in document 53a)
Doc53a.doc:	Longitudinal estimation – final summary report
Doc53b.doc:	Longitudinal estimation – final country reports
Doc54.doc:	Longitudinal estimation - manual
Doc54annex.doc:	Annexes to document 54
Doc5354.ppt:	Presentation of documents 53 and 54
Doc55.doc:	List of participants for 5th meeting
Doc56:	Agenda for 5th meeting
Doc57.doc:	Infostat report on B4
Doc57.ppt:	Presentation of document 57
Doc58.doc:	Draft B4 tables
Doc58mk.doc:	Draft B4 tables for Republic of Macedonia
Doc58gdp-euro.htm:	GDP figures in Euros
Doc58gdp-pps.htm:	GDP figures (pps)
Doc58b4ry.doc:	Table showing country by RY (trimester)
Doc58Sl.xls:	Table showing table 1 data for Slovenia
Doc59.ppt:	Polish publication
Doc60.doc:	DOSME part 2 plans
Doc61.doc:	Longitudinal database issues
Doc61.ppt:	Presentation of document 61
Doc62.doc:	Quality of B1 to B3 survey results
Doc62.xls:	Graphs for document 62
Doc63.doc:	Structural Business Statistics - Implementation, Development and Analysis (Papers from February 2000 meeting)
Doc64.doc:	Hardware/ software in year 2000
Doc65.doc	CEC reports on B4 survey and economic situation
Doc66.doc	Report of 5^th (final) working group meeting
Ukquest.xls:	UK response to year 2000 Eurostat registers questionnaire