Session No. 3
Paper No. 5
Country: Italy
Results of integrating several statistical and administrative sources for the analysis of international groups
Enrica Morganti
Italian National Statistical Institute
* The author would like to thank Giuseppe Garofalo for his supervision during the realisation of this work and his assistance in the revision of the final text. The author is the only one responsible for remaining mistakes and the opinions expressed do not involve the Italian National Statistical Institute.
Enquiries and correspondence should be addressed to: or .
Key words: affiliates, business registers, enterprise groups, globalisation, group head, group policy, majority ownership, parent unit, subsidiary unit, ultimate beneficial owner.
1. Introduction
During the last two decades the increase in the liberalisation of international transactions, the development of information technology and the decreasing in the costs of international transports, have pushed enterprises towards the adoption of global strategies, involving not only investments in different countries, but also non-equity relationships such as alliances, co-operation agreements and partnerships across borders. These facts have substantially changed the structure of the world system and international economic relations are increasingly involving intra-firm trade within Transnational Enterprise Groups (TEGs).
This globalisation process has determined an increase in the demand for information on transnational groups from policy makers, economists and the general public, with reference to the number of units, their economic activity and dimension, production, intermediate consumption, R&D, assets, major financial flows, as well as demographic issues such as birth, death, mergers and acquisitions.
The main issues are to evaluate the weight of foreign controlled enterprise groups versus domestically controlled enterprise groups on the domestic economies on one hand, and to monitor the degree of concentration on the domestic markets on the other.
Dealing with these aspects requires the definition of theoretical and operational definitions of enterprise groups, and the development of statistical tools to manage information minimising the burden on enterprises.
The Italian National Statistical Institute has started to collect data on enterprise groups since 1996, using different available sources[1]. This paper presents the main results obtained from the integration of several statistical and administrative sources for the setting up of a Business Register on enterprise groups.
The paper is organised as follows: in paragraph 2 the impact of globalisation on statistics is revisited, paragraph 3 discusses problems arising from different definitions and concepts that can be found when dealing with enterprise groups literature. In paragraph 4 the content and descriptive aspects of the available data sources are considered, while paragraph 5 contains the main results of the integration. Finally paragraph 6 draws some conclusions.
2. The Impact of Globalisation on Statistics
Globalisation can be defined as the existence of “relationships between economically productive units residing in different countries, related by legal, financial, contractual or other informal links (alliances, co-operation, etc.)”[2] and their socio-economic consequences; it implies a completely new way for enterprises of defining strategies, which has changed the international economic system of trade. The central economic unit to be investigated becomes the enterprise group rather than the enterprise unit, and the relevant attribute to study globalisation is the nationality of the head of group rather than the residence of the legal unit. To this attempt National Accounts data have appeared to be inadequate, since they refer to the concept of residence and the economic variables considered are truncated at the national border level.
The General Agreement on Trade in Services (GATS), signed at the World Trade Organisation in 1994, was the first international agreement explicitly requiring statistical figures on globalisation for the service sector[3]. It implies that, for their negotiations, the participating countries are able to supply data on international trade in services. Although some of these information are already covered by Balance of Payments statistics, this is not the case for one of the most important mode of supply international trade in services, which is through establishing a commercial presence abroad[4].
In 1996 the European Community co-ordinated a pilot study aimed at providing Foreign Affiliate Trade Statistics (FATS) for member countries. The European Community itself had already pointed out the importance of collecting information on enterprise groups in 1993, explicitly considering the definition of the unit enterprise group within the Community Regulation on Statistical Units[5]. At the end of 1995 Eurostat reinforced this effort with the establishment of a Globalisation Reflection Group in charge of preparing a Report assessing the direct or indirect impact of Globalisation on the different statistical domains and providing standard rules and definitions for member countries.
The final draft report of the Reflection Group individuated several aspects through which Globalisation can influence statistics, and consequently political economy decisions in the domestic countries.
The faster-pace growth of foreign trade and the increase of less tangible goods and more complex trade agreements, as well as of stages of successive transactions, have made more difficult to measure the volume of international transactions and to correctly establish the countries of destination and origin. Moreover the outsourcing strategies of firms, their decisions on location of production and distribution units, have influenced geographical breakdowns of international transactions, increased intra-firm trade flows and then lowered the quality of existing statistics on foreign trade. Foreign Direct Investment (FDI) data are at the moment the only available source on linkages between resident and non-resident enterprises, but their scope is neither adapt to the study of the activities of Transnational Enterprise Groups (TEGs) nor to the comparison with domestic business statistics.
Another important field where globalisation have a relevant impact is competitiveness analysis, because of the practice of transfer pricing. It has been shown that a high percentage of international transactions within enterprises of the same group takes place at non market prices. The possibility that transactions are not valued arms’ length is very high especially for complex or specific intermediate products or for services, where units values are difficult to evaluate on a standard basis. As a result transfer pricing can affect trade balances, intermediate consumption and value added figures. If a large number of international transactions take place at prices which are different from market prices, competitiveness analysis based on international trade flows only could be irrelevant[6].
In order to define political economy strategies it is important to know how the property of production means is allocated over the economy. Globalisation, may imply that some economic variables become out of control for domestic policy makers because of foreign controlled enterprises. This could affect the effectiveness of labour market policies and employment, as well as industrial policies for investments, technological progress, sectorial and regional development.
Finally globalisation means liberalisation of financial and trading activity and of FDI, reinforcing the role of TEGs in the world economy. This liberalisation process may have negative impact on the concentration degree of domestic markets if it is not accompanied by an increase in their contestability, which could ensure equal opportunities of access to all firms. In operative terms this can be achieved if sunk costs are very much reduced or null. In the other case competitiveness law and regulations are required to guarantee real competitiveness conditions[7].
In order to study all these statistical and economical implications of globalisation we firstly need a theoretical as well as an operational definition of the unit enterprise group.
3. Definitions of Enterprise Groups
A definition of enterprise groups relies on the concept of control. Literature on enterprise groups indicates several concepts of control according to the fields of investigation, which yield to different representation of the unit enterprise group[8].
A wide comprehensive concept of control can be found in the existence of a group policy strategy[9]. Although legal units which are part of a group maintain their juridical autonomy on how to reach their medium-term targets, a group policy implies the existence of a long-term global economic strategy developed by the group head and followed by all the unit enterprises of the group. This can be achieved notably by appointing appropriate managers and directors.
According to Eurostat[10], the group head is defined as the “parent legal unit which is not controlled either directly or indirectly by any other legal unit”. In Italy the model of Public Companies is still underdeveloped, while the structure of financial relationships among economic units is characterised by the relevant presence of enterprise groups which refers to families, so the term legal units for the group head often includes physical persons , orto institutions.
This concept of control allows us to enlarge the one given by the Italian Civil Code (art.2359), which recognises the existence of direct control (comma 1) if one of the following situations applies:
1)if one legal unit owns more than 50% of the shareholders’ or members’ voting rights in the ordinary assembly of another legal unit,
2)if one legal unit owns a percentage of enough voting rights to exercise a dominant influence on the decision taken by the ordinary assembly of another legal unit,
3)if one legal unit is under the dominant influence of another legal unit pursuant to a contract entered into with that legal unit.
and the existence of indirect control (comma 2), when for cases 1) or 2) to apply voting rights of majority owned affiliated enterprises are computed.
The civil code does not include in the concept of control the minority owned affiliates, which are treated as a separate category in the second paragraph of art.2359.
On the other hand a concept of control based on the group policy criterion could re-introduce this category of enterprises in the definition of enterprise group, if they are part of unitary economic strategy pursued by the group head, by virtue of a direct or indirect relevant influence.
Although there seems to be general agreement on the theoretical concept of control within the literature on enterprise groups, the formulation of an operational one appears more controversial. For example from an operational point of view statisticians are often forced to use the majority ownership concept as a proxy for control, and the ultimate beneficial owner as a proxy for the group head[11]. Majority ownership and control have to be –in principle- two distinct concepts, since the former does not necessarily implies the latter, if the majority owner enterprise does not effectively exercises a group policy, and on the other hand control can be based on other situations which do not necessarily require a majority ownership.
Another example of the difficulty of application of the theoretical definition is in the identification of the subject which exercises the group policy. Also in this case the operational definition suggests the attribution of group head to those legal units which draw up consolidated accounts. Even if consolidated accounts are the correct instrument to represents the group as a whole economic agent, they seem too restrictive when used to identify the group head. First of all because only juridical persons can draw up consolidated accounts, and group heads represented by physical persons are excluded, and secondly because there may be different legal units, drawing up consolidated accounts, which are still part of the same group in virtue of a common group policy. In this case the true head of group should be found going up through the chain of control, until the subject who exercises the group policy is reached. Since this guideline seems to be the mainly accredited one at Eurostat it is useful to consider it deeply.
The Seventh Council Directive on Consolidated Accounts (83/349/EEC ) requires all enterprises which are parent undertakings[12] to draw up consolidated accounts in order to give “a true and fair view of the assets and liabilities, the financial position and the profit and loss of all the undertakings consolidated taken as a whole”. The Directive considers situations of control either “based on majority of voting rights […]” or “on agreements, where these are permitted; whereas Member States in which the possibility occurs must be permitted to cover cases in which in certain circumstances control has been effectively exercised on the basis of a minority holding”, which can include the case of minority ownership affiliated. The Italian law (D.Leg. 127, 9/4/91) applied the principles of the European Directive[13], and expanded the concept of control given in the civil code, as in the Directive’s text cited above.
Still the Council Regulation on Statistical Units, pointed out that accounting groups are not suitable for statistical analysis since “they do not constitute mutually exclusive additive groups of enterprises”, just because of the presence of minority owned affiliated enterprises which are consolidated with the equity method. Then in order to use consolidated accounts for enterprise groups analysis the following amendments are suggested:
1)Consider accounting groups at the highest consolidation level, i.e. at the group head level,
2)Include in enterprise group units whose accounts are entirely integrated in those of the consolidating company,
3)Add majority-controlled units whose accounts are not included in the overall consolidating by virtue of the application of one of the criteria allowed by the Seventh directive, i.e. different in the type of activity or small relative size,
4)discount temporary links of less then one year.
The first amendment restricts the domain of possible group heads, while the second one leads again to the exclusion of the minority owned affiliates. This second amendment in particular might be too strong, also because several definitions contained in other EEC Regulations and Directives, as well as in Italian laws, indicate the presumption of control also when a participating interest of at least 20% is found.
Different solutions to guarantee that statistical unit groups were mutually, exclusive, additive units could than be studied and identified in order to deal with these cases, instead of adopting such radical exclusions[14].
Finally the correct definition of control and the identification of the group head have important implications on the attribution of the nationality to the group itself, whose criteria are still one of the main issue under discussion at Eurostat[15] and OCSE[16].
4. Available Data Sources
Eurostat Globalisation Reflection Group final report suggests, without any hierarchical priority, several sources of information which can be used to collect data on TEGs[17]:
1)Making use of existing units (groups defined by the domestic legislation or by enterprises themselves, and associated information (information from consolidated accounts, data from market listings).
2)Collecting data constructed by respondent, with given rules for the delineation of groups – e.g., through surveys (direct enterprise surveys e.g., on foreign affiliates, administrative records).
Use of administrative data on individual international transactions (e.g., trade, investment) to reconstruct annual flow data through the use of identifiers (data on intra-firm trade from customs forms, construction of enterprise-level foreign investment data).
3)Statistical “construction” of units and data based on existing information from linked registers (combining enterprise data with information on financial relationships in order to construct data on enterprise groups, construction of truncated enterprise group data through linking of registers from different countries).
With reference to the Italian experience the available data sources are the following:
1)The Italian National Statistical Institute (ISTAT) Survey on Employment for years 1995-96,
2)The Italian National Statistical Institute (ISTAT) Survey on Scientific Research for year 1995,
3)The Italian Exchange Rate Office (Ufficio Italiano Cambi – UIC) Survey for year 1996,
4)The Archive of the Italian National Committee of Inspection on Quoted Societies (Commissione Nazionale per le Società e la Borsa – CONSOB) for year 1996,
5)Bureau van Dijk data bank,
6)The Italian National Statistical Institute Business Register on Enterprise Units (Archivio Statistico delle Imprese Attive – ASIA) .
In order to correctly treat these different sources our first object of study has been a deep comprehension of the specificity of each survey and of the concepts and definitions thereby used. Problems of uncertain or ambiguous definitions of the used terminology existed not only for ‘external’ sources, but also for some ISTAT Surveys, where questions on enterprise groups were not preceded by a definition of control.
In the followings we briefly discuss the specific contents of each source:
1) Survey on Employment (1995) – 7,616 observations.
This survey is conducted on a sample of enterprises with less than 50 employees and on the whole universe for those of greater dimension. For the 1995 year the section referring to enterprise groups was preceded by the definition of control to be applied by the respondents. The survey gives information on 42,000 enterprises on:
1)Whether it belongs to a group,
2)Nationality of direct controlling legal unit,
3)Whether Italian the fiscal code of direct controlling legal unit,
4)Whether not Italian the nationality.
2) Survey on Employment (1996) – 3,575 observations
The survey is the same as the one described above for year 1995, but information on group enterprises are limited to point 1).
3) Survey on Scientific Research (1995) – 723 observations.
The survey is limited to enterprises with R&D departments. Questions regarding information on enterprise groups were not preceded by any definition of control.
This survey provides information on:
1)Whether it belongs to a group,
2)The geographical area of the majority owner.
4) Italian Exchange Rate Office Survey (1996) – 2,824 observations.
The survey is based on the Statistical Currency Communication (SCC), a compulsory declaration required to all enterprises which have transaction over 20 million of Italian lira with foreign partners. The SCC contains information on: