Eurostat-ECE Seminar on Business Registers June 21-22, 2005 The MNE project Eurostat-ECE Seminar on Business Registers June 21-22, 2005

Background The Multinational Enterprise Project (MNE) arose from the session on globalization at the June 10-12, 2003 Conference of European Statisticians in Geneva Objective: to identify areas where the measurement of the activities of multinational enterprises could be improved Assess feasibility of having MNEs report in an integrated fashion to several national statistical offices (NSOs), taking into account the confidentiality legislation governing the respective NSOs

Project structure and governance Five participating countries Canada, France, Italy, The Netherlands, United Kingdom Bureau of the Conference of European Statisticians (CES) acted as the steering committee Roundtable on Business Survey Frames has served as a sounding board

Study hypotheses Areas where improvements might be identified the description of the structure of the MNE in the registers and other files of the NSOs the measurement of important economic and financial variables relevant to national accounting the coordination of measurement activities between NSOs

Study design Common questionnaire Integrated collection Four variables: Sales, operating profits/loss, capital expenditures, number of employees List of legal entities covered by data supplied Integrated collection Home country NSO collects from MNE on behalf of all NSOs and shares data with them Starting point is data from public annual report for global entity, split by MNE, by country

Analysis Compare data received from MNEs as part of study (directly or by data sharing from other NSO) to data obtained in regular statistical program Provide measures of “nearness” of both sources

Reactions of MNEs Each participating country contacted a minimum of 3 MNEs MNE had to have at least one production facility in each of the five countries 11 MNEs agreed to participate 8 had operations in all 5 countries 3 had operations in 4 of the 5 countries All agreed to data sharing waiver Coordinated approach operationally feasible

Analysis results Structure Variables Counts of legal entities by MNE, by NSO, common Variables Measure of nearness Difference= ((MNE response – Internal results)/MNE response)*100 If difference < -10 % then the code was set to “-2” If -10%<= difference < -5 % then the code was set to “-1” If -5 % <= difference <= 5 % then the code was set to “0” If 5 % < difference <= 10 % then the code was set to “1” If 10 % < difference then the code was set to “2”

Legal structure of MNEs

Legal structure of MNEs In all, there are 50 data points for which comparisons were possible 22 exactly the same count of legal entities 28 differences 20 small differences (1 or 2 legal entities) 8 large differences For 6 of these 8, problem lies with the structure of MNEs in their home countries Legal structures in home countries are larger Errors are therefore potentially larger Getting the home country structure right is therefore key to the overall measurement of the MNE

Sales Of the 45 measurements that are available, 20 are within 10% of each other and 25 are more that 10% apart. Reasons differences in the structure of the MNE reported in the previous section MNEs were not able, or willing, to provide intra-enterprise sales between geographic segments

Operating profits or losses Only 19 measurements for which it is possible to make comparisons, out of a possible 52 Only variable that was considered confidential by the MNEs Of the 19 measurements, 5 are within 10% of each other and 14 are more that 10% apart

Capitalized expenditures 32 observations for which comparisons are possible 13 are within 10% of each other and 19 that are more than 10% apart Often, no data against which to compare in regular statistical program because units are too small, or surveyed at establishment level

Total number of employees 46 observations for which comparisons can be made 23 within 10% of each other and 23 that are more than 10% apart Most accurately measured variable

Resident vs foreign affiliates Overall, the data obtained for resident MNEs directly by NSOs tend to be more comparable to internal sources than the data obtained for foreign affiliates of MNEs through data sharing with other NSOs

Conclusions Data collected in a coordinated fashion from the MNEs for this study provided results that were different from the data collected by survey or administrative instruments in the regular programs of the National Statistics Offices 28% within 10% 36% more than 10% divergent 36% cannot be compared for a variety of reasons

Conclusions Employment variable most accurately measured Followed closely by the sales variable Capitalized expenditures were less well measured than either Operating earnings variable was the most poorly reported and the least accurately measured

Conclusions A centrally coordinated collection approach, such as used for this study, might elicit a defensive reaction by MNEs that is not encountered when each national component is approached individually The centrally coordinated approach to collection may in fact result in less data being collected from the MNEs overall than independent approaches to each national component

Recommendations Terminate study in its present form at this point Work be continued under the auspices of the Eurostat project on the creation of a European register of enterprise groups, taking into account specific learnings and experiences from the project