Archive:Microdata linking - international sourcing

This Statistics Explained article is outdated and has been archived - for recent articles on structural business statistics see here.

Data from December 2013. Most recent data: Further Eurostat information, Main tables and Database. Planned update: June 2019.

Authors: Zuzanna Tilewska and Peter Bøegh Nielsen (Statistics Denmark), Pekka Alajääskö, Simon Bley and Anton Roodhuijzen (Eurostat, Structural business statistics and global value chains)

This article describes the microdata linking project launched in 2012, linking the results of the international sourcing survey to structural business statistics (SBS), international trade in goods statistics (ITGS) and foreign affiliates statistics (FATS) on enterprise level, using unique enterprise identification numbers. We start by presenting some of the main findings of this exercise, before outlining the organisation of the project and explaining some methodological issues of the microdata linking approach. We conclude by briefly addressing some other European microdata projects and the way forward for microdata linking.

European business-statistics compilers often face a dilemma: on the one hand, user and policy requirements demand additional information on the structure and development of European enterprises, while on the other hand, budget constraints and the reluctance to increase the burden on survey respondents and national statistical institutes put tight restraints on the enlargement of data requirements.

One solution for this dilemma is a new innovative approach for combining existing statistical information by linking the data on the microdata level without increasing the burden on enterprises. Microdata linking provides an opportunity to discover new information and to develop new statistics and indicators both by using existing data sets and by combining these with new data collections.

In the European statistical system (ESS) one initiative looking into this opportunity for business statistics was a project launched in 2010 linking the results of the survey on “international organisation and the sourcing of business functions” (hereafter referred to as the international sourcing survey) with structural business statistics (SBS) and international trade in goods statistics (ITGS). This was followed up with an exercise launched in late 2012 which broadened the scope to also include foreign affiliates statistics (FATS).

This article is part of an online publication on Microdata linking in business statistics.

Figure 1: Percentage of international sourcing for domestically and foreign-controlled enterprises

Figure 2: Percentage of international sourcing for domestically and foreign-controlled trading enterprises in manufacturing

Figure 3: Number of regions to which manufacturing enterprises are sourcing internationally

Figure 4: Number of regions to which manufacturing enterprises are exporting

Figure 5: Percentage of foreign-controlled enterprises exporting to the country of control, 2008

Figure 6: Percentage of foreign-controlled enterprises importing from the country of control, 2008

Figure 7: Export of goods intensity, by international sourcing status, 2011 (index 2008 = 100)

Figure 8: Employment (full-time equivalents), by sourcing status, 2011 (index 2008 = 100)

Figure 9: Employment in foreign-controlled enterprises (full-time equivalents), by sourcing status, 2011 (index 2008 = 100)

Figure 10: Employment in domestically controlled enterprises (full-time equivalents), by sourcing status, 2011 (index 2008 = 100)

Main statistical findings

Foreign-controlled enterprises more active in international sourcing and trade

Data from the international sourcing survey was linked with the inward foreign affiliate statistics (iFATS) to analyse the potential relationship between the enterprise being domestically controlled or foreign-controlled and the prevalence of international sourcing (see Figure 1).

Significantly more foreign-controlled enterprises than domestically controlled enterprises source internationally.

The international sourcing survey showed that international sourcing was most prevalent in the Nordic countries (Denmark, Finland, Sweden and Norway).

Linking confirmed that this was the case for both domestically controlled and foreign-controlled enterprises.

The data was further linked to international trade in goods statistics (see Figure 2).

Although Figure 2 shows the same pattern for the sub-population of traders as in Figure 1 for the complete population, the prevalence of sourcing for the traders is somewhat higher.

Exporting more geographically diversified than international sourcing

The international sourcing survey collected data on enterprises engaged in international sourcing broken down by region. Analysis of the microdata from the international sourcing survey shows that in all countries the majority of manufacturing enterprises sourced to only one region (varying from 51 % for Finland to 83 % for Sweden – see Figure 3).

By linking data from the international sourcing survey to international trade in goods statistics, it was possible to analyse the regional patterns for all enterprises with at least 100 employees involved in international trade, focusing on enterprises in manufacturing since these dominate the international trade in goods (see Figure 4). The analysis showed that:

Patterns in export markets are more diversified than in international sourcing.
Typically manufacturing enterprises trade with 3-5 regions.
The pattern was even more pronounced in the EU-15 countries where enterprises often traded with more than five regions.
In Member States that joined the EU in 2004 and 2007 the number of enterprises trading in one or two regions was quite substantial.

Foreign-controlled enterprises often trade with enterprises in the controlling country

By linking data from the foreign affiliates statistics (FATS) to international trade in goods statistics it was possible to gain more information on the trade patterns broken down by ownership. The findings of this analysis are consistent with the findings in the article "International sourcing of business functions", showing that 70 % to 80 % of sourcing enterprises are carrying out so-called international insourcing, i.e. moving business functions to a foreign affiliate or daughter company within the same multinational enterprise group.

Foreign-controlled enterprises tend to export goods to the ‘country of control’. In 2008, between 49 % (Lithuania) and 80 % (Slovakia) of all foreign-controlled manufacturing enterprises exported goods to the country of control (see Figure 5).
Imports of goods from the country of control are even more common, with between 72 % (France) and 92 % (Lithuania) of foreign-controlled enterprises importing goods from the country of control (see Figure 6).

Export intensity growth rates higher for enterprises not sourcing internationally

Linking the enterprises from the international sourcing survey to the trade statistics provided the value of exports, and linking them to the structural business statistics provided information on the total turnover. This made it possible to calculate the export intensity (exports of goods as a proportion of total turnover) for groups of enterprises defined by their sourcing status (see Figure 7). The results of this exercise gave a somewhat mixed picture:

In all countries except Ireland, France and Sweden, enterprises not sourcing internationally saw their export intensity grow more or fall less between 2008 and 2011 than enterprises sourcing their core function internationally.
In six countries, enterprises sourcing core functions internationally saw their export intensity grow more or fall less than enterprises sourcing support functions internationally.
The opposite was true in Finland and Norway.
In Slovakia and Bulgaria no difference was found between sourcing core functions or support functions.

Employment grew less or fell more in enterprises sourcing internationally

When analysing the effects on employment in this context, one should bear in mind that international sourcing is here defined as enterprises moving business functions abroad – functions that were previously carried out in the home country are now carried out in another country. In other words, the international sourcing of a business function will by definition lead to a decrease in employment. In the linking exercise described in this article, it was not possible to distinguish between this employment decrease due to international sourcing and any other changes in employment due to other economic circumstances.

In order to analyse the effect of international sourcing on the country's employment situation, data from the international sourcing survey was linked to data from structural business statistics. To some extent the effects seemed to depend on what business functions were sourced internationally, core functions or support functions (see Figure 8):

Between 2008 and 2011, enterprises not engaging in international sourcing generally showed a larger increase or a smaller decrease in the employment (measured in full-time equivalents) they were providing in their home country, compared to enterprises that sourced core functions internationally. This was the case in all countries except Finland and Lithuania.
In most countries, enterprises sourcing support functions internationally also showed a larger increase or a smaller decrease in the employment they were providing in the country in question than those sourcing core functions internationally.
Comparing enterprises not engaging in international sourcing and those sourcing only support functions internationally gives a more mixed picture. In five countries employment showed a more positive development among enterprises not sourcing internationally while in the remaining five the opposite was true.

This data was also linked to foreign affiliate statistics to analyse possible differences between foreign-controlled and domestically controlled enterprises. The results of this can be seen in Figures 9 and 10.

Data sources and availability

The ESS microdata linking project on international sourcing

The microdata linking approach New statistics on enterprises have traditionally been produced by carrying out surveys. Microdata linking presents an innovative approach to obtaining new information on the economic performance of enterprises by linking different existing statistical sources at individual enterprise level (microdata level). This approach does not require new surveys to be carried out and thus does not increase the burden placed on enterprises.

When attempting to link two or more sets of data, two things are vital for a successful exercise:

There has to be a large enough intersection of responding units.
There has to be a unique identifier or at least a very reliable matching approach.

The international sourcing microdata linking project

Background

The microdata linking project on international sourcing was set up by Eurostat at the end of 2012 and carried out in 2013 by the national statistical offices in nine EU Member States (Bulgaria, Denmark acting as coordinator for the project, Finland, France, Ireland, Lithuania, the Netherlands, Slovakia and Sweden) and Norway.

The two main objectives of the project were

to test the centrally developed methodology for linking datasets and producing standardised outputs across a number of participating countries, and
to produce new information on enterprises that responded to the 2009–11 international sourcing survey, carried out in 15 countries including all those taking part in the microdata linking project.

Data sources

The core dataset used in the project was the data from the 2009–11 international sourcing survey (IS survey). This data was linked at individual enterprise level to structural business statistics (SBS), international trade in goods statistics (ITGS), inward foreign affiliate statistics (iFATS) and to additional information from national business registers (NBR).

The original sample for the IS survey was selected as a subset of the SBS population, which means that in theory there would be a matching SBS data set for every IS data set; thus fulfilling the condition of a large enough intersection of responding units.

A unique identifier for IS and SBS was found in the national business registers (NBR) used as a frame for the microdata linking. Business registers play a pivotal role in linking datasets. In most countries, the business register is the central register containing enterprise identification numbers and information on multinational groups. It is where samples for surveys are compiled and the main breakdowns (such as economic activity) are kept up to date.

ITGS is not directly based on NBRs and therefore, as a matching approach, VAT numbers were used to link ITGS data sets to the NBR and thereby to IS and SBS.

Control group

To be able to benchmark the stratum of enterprises engaging in international sourcing, it was important to designate a control group. A simple approach would have been to use all the enterprises in the linked dataset not engaging in international sourcing, however, this may have led to a skewed result, as the structure of the control group in regard to other variables like enterprise size or turnover may have been very different from the structure of the group of sourcing enterprises. The approach taken in this project was therefore to construct a control group of non-sourcing enterprises from the survey population with the same characteristics as the sourcing enterprises in terms of economic activity.

ESS project carried out in three phases

In 2013 the project was carried out in three separate phases by the participating national statistical offices.

The first phase involved the construction of the linked microdata files. The project coordinators produced standardised guidelines explaining in detail how the linked datasets in each country were to be structured and provided code to ensure that identical tables were made in all countries. Each country recorded information from all the data sources mentioned above in one database. The microdata files were stored locally at the national statistical offices throughout the project and were not shared with third parties.

In the second phase of the project, the linked files were tested for consistency. Although each dataset being used in the project had already been carefully edited, it was necessary to carry out further checks to ensure, for example, that enterprises were represented by the same statistical units across different datasets and over time, as the reporting units used for specific enterprises could have differed across the data sources mentioned above. In fact we found these differences and corrected for this. The tests used during this phase of the project were devised by the project coordinators and implemented locally by the national statistical offices.

In the third phase of the project standardised statistical output was created in each country, consisting of a descriptive analysis, longitudinal analysis and regression analysis.

Context

The main findings of the linking exercise launched in 2012 are presented at the beginning of this article. However, some more general conclusions can also be drawn from the experience gained through these microdata linking exercises.

Microdata linking has significant potential as a way of gathering new statistical evidence without increasing the burden placed on respondents.
A coordinated approach to microdata linking is a cost-effective way for national statistical offices to undertake microdata linking and ensures harmonised, comparable results across countries.
Microdata linking gives new insights into the behaviour of enterprises and the impact of this behaviour on various aspects of European economies.
By linking various data sources, it is possible to conduct regression analysis, and by extension, other types of statistical analysis such as multivariate analysis and cluster analysis.

Next step in microdata linking

Microdata linking is considered to be an important strategic approach which will be very useful in the production of business statistics in the future. Eurostat has therefore launched a new project on microdata linking and has set up specific linking exercises to be carried out in 2014 and 2015. Nine EU Member States (Austria, Denmark, Finland, Germany, Ireland, Latvia, the Netherlands, Portugal and Sweden) and Norway are participating in the project. The general objective is to develop and implement microdata linking methods in order to increase the value of data already available, in particular structural business statistics.

This new project will benefit greatly from the experience gained through the microdata linking projects mentioned above.

Further Eurostat information

Publications