Methodologies and Working papers. PEEIs in focus. A summary for the industrial production index edition

Methodologies and Working papers PEEIs in focus A summary for the industrial production index 2005 edition

Methodologies and Working papers PEEIs in focus A summary for the industrial production index 2005 edition

This report has been prepared based on information kindly provided by the STS coordinators of each country. The original draft of this report was written and produced by Simon Allen and Andrew Redpath of INFORMA s.à r.l. The opinions expressed in this report are those of the individual authors alone and do not necessarily reflect the position of the European Commission. For further information please contact: Thomas Jaegers Eurostat Statistical Office of the European Union e-mail: thomas.jaegers@ec.europa.eu Europe Direct is a service to help you find answers to your questions about the European Union. Freephone number (*): 00 800 6 7 8 9 10 11 (*) Certain mobile telephone operators do not allow access to 00 800 numbers or these calls may be billed. More information on the European Union is available on the Internet (http://europa.eu). Cataloguing data can be found at the end of this publication. Luxembourg: Publications Office of the European Union, 2012 ISBN 978-92-79-19515-0 ISSN 1977-0375 doi:10.2785/2911 Cat. No KS-RA-12-012-EN-N Theme: Industry, trade and services Collection: Methodologies & Working papers European Union, 2012 Reproduction is authorised provided the source is acknowledged.

Contents Contents 3 Introduction 5 What are PEEIs 5 PEEI in focus: reports and other methodological documentation 5 Purpose of this report 5 Part A: Data collection 6 1. Collection method and population coverage 6 1.1 Sources of the basic data 6 1.2 Nationally defined population 6 1.3 Coverage rate compared to full coverage 10 2. Selection of units 12 2.1 Basic method of selection 12 2.2 Stratification criteria used and sample rates 13 2.3 Sample size and coverage 14 2.4 Frequency of updating of the sample 16 3. Type of data collected 17 3.1 Basic method(s) used and their relative importance 17 3.2 Treatment of quality changes (if quantity data used) 24 Part B: Non response 25 4. Non-response rates 25 4.1 Snapshot of non-response 25 4.2 Development of response rates during one data collection exercise 26 4.3 Development of response rates over the course of a year 27 5. Treating non-response 28 5.1 Reducing non-response: present practice and plans 28 5.2. Treatment of non-response: imputation and reweighting 29 5.3. Replacement of non-respondents 30 5.4. Measurement of non-response errors/bias 31 Part C: Data processing 32 6. Data editing 32 7. Deflation 34 8. Index calculation/use of weights 36 8.1 Weights used and frequency of updating 36 8.2 Type of index 38 9. Benchmarking 39 10. Time series decomposition 40 10.1 Order of adjustment: choice of input series 40 10.2 Software used 40 10.3 Methods for WDA 41 10.4 Methods for SA 44 10.5 Aggregation approach 45 10.6 Revision of models 46 10.7 Base year of indices 47 Part D: Error measurement 48 11 Non-sampling errors 48 12 Sample errors 49 13 Comparison among sources 50 14 Bias 51 15 Dissemination of information about errors 52 Annex 53 1. Definitions 53 Industrial production index 53 Enterprise 53 Local unit 53 KAU 53 PEEIs in focus A summary for the industrial production index 3

Local KAU 54 2. Index compilation 54 Weights 54 Decomposition 55 Working-day adjustment 55 Seasonal adjustment 55 3. Classifications 55 NACE 55 CPA 56 PEEIs in focus A summary for the industrial production index 4

Introduction Important note: this report was compiled on the basis of information provided in 2004 and 2005; the methodology used by Member States may well have changed since then, in particular due to rebasing to 2005 and the introduction of NACE Rev. 2. Some countries provided updates in 2008. What are PEEIs Eurostat compiles European Union and euro-zone infra-annual economic statistics relevant for short-term economic analysis. Among these, a list of indicators, called Principal European Economic Indicators (PEEIs) has been identified by key users as of prime importance for the conduct of monetary and economic policy of the euro-zone. These indicators are mainly released through Eurostat's website under the heading Euroindicators. Eight PEEIs concern short-term statistics 1 (STS), of which six were included in the Council Regulation (EC) No 1165/98 of 19 May 1998 concerning short-term statistics (hereafter the STS Regulation) and the remaining two in the Regulation (EC) No 1158/2005 of the European Parliament and of the Council of 6 July 2005 amending Council Regulation (EC) No 1165/98 concerning short-term statistics. PEEI in focus: reports and other methodological documentation In the early 1990s Eurostat developed a database recording the methodology used for industrial and construction STS; in the second half of the 1990s this was extended to cover also the retail trade and services STS and was thereafter know as STS Sources. Since then the STS Sources database has been updated several times based on information provided by the EU Member States and the Candidate countries, as well as Norway and Switzerland. The latest update was done during 2004/2005. The latest update was done in 2007. In March 2004 it was decided to carry out a quality project for the PEEIs that would result in a yearly report. As noted above eight PEEIs concern STS, of which six could be considered as operational in early 2004. Four quality components were selected for this project and the quality indicators that should be included in the quality report were chosen. In June 2004 Eurostat presented further details of this quality project for the PEEIs to the participants of an STS working party. At the same meeting in June 2004 Eurostat proposed to consider a detailed, focused analysis for one PEEI every year, with the chosen PEEI changing each year. The first indicator selected to be a "PEEI in focus" was the industrial production index (hereafter referred to as the IPI), for which information was collected at the end of 2004 and into 2005. The retail volume of sales index was selected as the second indicator, with information being collected during 2006. Towards the end of 2004 Eurostat sent a pre-filled template for a PEEI in focus report on the IPI to the EEA member States as well as the (then) Candidate countries and Switzerland. The PEEI in focus report concerned information on five selected areas, including a specific focus on the seasonal and working day adjustment methods. The vast majority of Member States returned the reports having corrected and completed them. Among the EU-25 Member States replies were not received from the Czech Republic, Malta or Portugal: note that Malta does not produce an IPI on a regular basis. In addition, reports were received from Bulgaria, Romania, Turkey, Norway and Switzerland. Information included in the present report concerns only those countries that replied. Purpose of this report The present report aims to present in a simple manner the information collected in the PEEI in focus reports for the IPI. Its purpose essentially is to allow countries to compare their methods with those in other countries, based on information presented in a structured, and as far as possible, coherent manner. Wherever possible the report presents information for each of the countries, as well as presenting summary information for all of the countries as a whole. 1 i) industrial production index ii) industrial output price index for domestic market iii) industrial new orders index iv) industrial import price index v) production in construction vi) turnover index for retail trade and repair vii) turnover index for other services viii) corporate output price index for services. PEEIs in focus A summary for the industrial production index 5

Part A: Data collection 1. Collection method and population coverage 1.1 Sources of the basic data The IPI is derived from statistical survey data in every country see Table 1.1. All of the countries compile a monthly index, although at least one (Germany) collects data on both a monthly and quarterly basis for the purpose of compiling the IPI. A number of countries also use administrative data. Denmark, Cyprus and the United Kingdom indicated that they use administrative data for particular activities, and it is quite likely that this is the case in other countries, but was not mentioned in their reports. Table 1.1: Main data sources Main sources Main sources BE (1) Survey NL Survey DK Survey and administrative data (for section E) AT (1) Survey DE Survey (2) PL Survey EE (1) Survey SI Survey EL Survey SK (1) Survey ES Survey FI Survey and administrative data (tax) FR Survey SE Survey IE Survey UK Survey and administrative data (energy data and for some food processing activities), and data from a trade bureau (for some metal related activities) IT Survey BG Survey CY Survey and administrative data (cigarettes and the majority of beverages) RO (1) Survey LV (1) Survey TR Survey LT Survey NO Survey LU (1) Survey CH Survey HU Survey and administrative data (VAT) (1) A size threshold is used, and above this threshold a census is carried out. (2) In addition a quarterly survey is carried out where a size threshold is used, and above this threshold a census is carried out. 1.2 Nationally defined population Table 1.2 shows the population that each country targets for its IPI: this population is hereafter referred to as the nationally defined population. The information on the population is given in terms of the activities covered, the size of units covered, and the type of statistical unit. For reference note that the STS Regulation requires an activity coverage of NACE Sections C to E, does not specifically mention size thresholds, and requires the KAU to be used as the type of statistical unit. Concerning the choice of the type of statistical unit, it should be noted that unlike most other STS indicators (the main exception being price and cost indices), many countries compile the IPI from product data (as will be seen later), and as such the choice of type of statistical unit plays a lesser role. Concerning the size coverage, it should be noted that in addition to the size thresholds mentioned here, to limit the nationally defined population, some countries also use further size cut-offs when selecting units for their surveys see point 2.1 below. Note that the systematic exclusion of part of the population (for example on grounds of their activity or size) leads to a bias in the resulting indicators when compared to the population required by the STS Regulation. For example, the evolution over time of the output of smaller units may be different from that of larger units. Some countries specifically adjust for this; for example, Hungary, the Netherlands and Finland use estimation methods or administrative data to ensure that their indicators reflect a complete coverage. PEEIs in focus A summary for the industrial production index 6

Table 1.2: Nationally defined populations Activity coverage Size coverage Unit >= 10 persons employed or turnover >= EUR 2.5 million for industrial enterprises BE Sections C to E (note that the industrial activity of nonindustrial enterprises is also included) >= 20 persons employed or turnover >= EUR 2.5 million for non-industrial enterprises Enterprise DK Sections C to E >= 20 persons employed KAU DE Sections C to F >= 20 employees Local unit EE Sections C to E >= 50 employees or turnover >= EEK 15 million Enterprise EL Sections C to E >= 10 employees KAU ES Sections C to E except Divisions 37 and 41 and Groups 23.3 and 35.5 >= 20 employees Enterprise FR Sections C to E >= 20 employees (sometimes 10) Not specified IE Sections C to E >=20 persons employed (sometimes lowered) Local unit IT Sections C to E except Divisions 10, 12 and 41 >= 20 employees (sometimes different threshold) Enterprise CY Sections C to E All size classes Enterprise LV Sections C to E >= 20 persons employed or industrial turnover >= LVL 300 000 Enterprise LT Sections C to E All size classes Enterprise LU Sections C to E except Divisions 10, 11, 12, 13, 23, 37 and 41 and Groups 33.1 and 40.3 and Classes 15.81, 18.22, 18.24 and 29.32 >=20 persons employed Not specified HU Sections C to E All size classes: >= 5 persons employed for the survey; administrative data used for smaller units and included in the index. Enterprise NL Sections C to E All size classes: >= 20 persons employed for the survey; smaller units are estimated and included in the index. Enterprise AT Sections C to F > 9 persons employed "Establishment" (KAU) PL Sections C to E >= 10 persons employed Enterprise SI Sections C to E except Group 40.3 and Divisions 11 and 41 All size classes Enterprise and local KAU SK Sections C to E (note that the industrial activity of nonindustrial enterprises is also included) >= 20 employees or turnover >= SKK 100 million Enterprise FI Sections C to E All size classes: >= 20 employees for the survey; administrative data used for smaller units and included in the index. Local KAU SE Sections C to E except Group 40.3 and Division 41 >= 10 employees (KAUs belonging to enterprises above this threshold) KAU UK Sections C to E All size classes KAU BG Sections C to E >= 10 persons employed Enterprise RO Sections C to E except Divisions 12 and 30 >= 20 employees Enterprise TR Sections C to E except Divisions 12, 37, 41 and Groups 40.2 and 40.3 All size classes except for the private sector in Section D which covers >= 10 employees Local unit (locally referred to as establishment) NO Sections C to E except Division 41 >= 10 persons employed Local KAU CH Sections C to E >= 5 persons employed in full time equivalent units "Establishment" Note: some countries may have mentioned the exclusion of some activities that in fact do not exist in that country, so their exclusion will not affect the comparability of results. Note: in some countries the size threshold may be based on a different measure of employment than the one stated, as the number of (paid) employees and the total number of persons employed are often confused. PEEIs in focus A summary for the industrial production index 7

Figure 1.2a lists the activities that countries indicated that they excluded from their nationally defined populations, indicating which are the most commonly excluded from the required NACE range of Sections C to E. It should be noted that some countries may have mentioned the exclusion of some activities that in fact do not exist in that country, despite the fact that exclusion will not affect the comparability of results for example, this may well be the case for Division 12 (mining or uranium and thorium ores) which does not exist in the vast majority of countries. Among the most commonly excluded activities are those in Section E (within Divisions 40 and 41). Another is the case of recycling (Division 37) which, while having particular characteristics in terms of processing, produces goods that can often not be easily distinguished from similar products produced from new raw materials: countries collecting product level data may have problems covering this activity, and the three countries (Spain, Luxembourg and Turkey) that noted that they do not cover Division 37 all use mainly or exclusively product level data (either in value or in quantity) to compile their IPI. Number of countries Figure 1.2a: Number of countries excluding specified activities (NACE) 8 7 6 5 4 3 2 1 0 41 12 40.3 37 10 11 13 15.81 18.22 18.24 23 (all of) 23.3 29.32 30 33.1 35.5 40.2 NACE codes The various size thresholds that are used by countries to limit the nationally defined population are summarised in Figure 1.2b. The most common size coverage is based on employment of 20 or more, followed by employment of 10 or more and full coverage. Figure 1.2b: Simplified presentation of employment thresholds to determine the coverage of the national sources: number of countries using specified threshold (1) 7 1 1 8 10 20 and more 10 and more No threshold 5 and more 50 and more (1) Simplifications: use of employees or persons employed not separately identified; occasional use of different thresholds ignored; different thresholds for nonindustrial enterprises ignored. Note that some countries additionally use financial thresholds, normally related to turnover. PEEIs in focus A summary for the industrial production index 8

Figure 1.2c provides a summary of the types of statistical units that were reported by the countries. As noted above, for countries collecting product level data the choice of type of statistical unit plays less of a role in determining the comparability of results. One important detail that is not available from the PEEI in focus reports, is the way that the statistical unit and the activity coverage are combined. For example, where the enterprise is used as the type of statistical unit, but product level data is collected, is the collection of the product level data restricted to enterprises whose principal activity is in Sections C to E, or does it also include the product level data of enterprises with a secondary activity in Sections C to E? Equally, when product level data is not collected, and the main method is a (deflated) value of output (such as turnover) and again the enterprise is chosen as the type of statistical unit, is only the output from industrial activities collected or the total output? Given the way that IPIs are compiled in nearly all countries, namely by compiling an index at a relatively low level of activity (3 or 4-digit level for example) and then aggregating using weights, the choice of type of statistical unit used to compile the weights is also important. Figure 1.2c: Number of countries using specified type of statistical unit 2 1 4 2 2 14 Enterprise KAU Local KAU Enterprise and local KAU Local unit Not specified or "establishment" (1) (1) Note that the establishment is not an EU type of statistical unit, and it may correspond to a variety of types of statistical units, commonly a local KAU, but also possibly a local unit or a KAU. PEEIs in focus A summary for the industrial production index 9

1.3 Coverage rate compared to full coverage Table 1.3 shows the size of the nationally defined population (as determined by the activity and size coverage presented in Table 1.2) in terms of the number of units, and its distribution between the five main industrial groupings (MIGs). Furthermore, it shows the proportion of total value added of Sections C to E that is covered by the nationally defined population for each of the MIGs and the industrial total. Note that this coverage rate should not be confused with the proportion of value added accounted for by units that actually received a questionnaire (in other words the sample) which is treated in point 2. In a few cases the coverage rates are not consistent with information given elsewhere in the reports. Equally, the number of units in the nationally defined population as a whole is not always consistent with the sum of the nationally defined population for each MIG. Table 1.3: Size and coverage rate of the nationally defined populations Coverage rates (nationally defined population relative to Number of units in the nationally defined population full STS requirements) in terms of value-added (%) Industry total Intermediate goods Capital goods Consumer durables Consumer non-durables BE 41 424 12 415 7 053 4 101 17 558 128 : : : : : : DK 2846 890 1025 272 646 13 83 87 80 86 87 58 DE (1) 45 796 19 624 13 675 1 881 10 463 153 94 : : : : : EE (2) 937 287 145 74 273 158 87 85 84 83 85 98 EL 5 595 2 640 817 495 2 500 89 : : : : : : ES 30 924 12 684 6 588 2 236 8 555 861 95 95 96 91 92 99 FR 24 024 10 060 5 655 1 018 7 212 259 91 92 93 88 83 98 IE 1 961 690 370 168 609 124 97 97 97 97 97 99 IT : : : : : : 76 : : : : : CY 6 335 2 095 1 144 898 2 103 95 100 100 100 100 100 100 LV 1 211 476 119 64 454 98 87 84 83 84 87 95 LT (3) 8 708 3 440 881 682 3 425 280 : : : : : : LU 226 98 61 1 60 6 89 95 89 100 91 50 HU 15 683 5 481 3 445 972 5 449 336 97 97 96 97 97 99 NL 6 833 2 308 2 068 405 1 931 121 100 100 100 100 100 100 AT (4) 8 257 2 745 1 889 922 2 544 157 94 94 96 85 92 96 PL (5) 31 058 10 178 4 722 1 862 13 232 1 064 95 93 92 93 94 100 SI 18 702 8 984 2 652 1 668 5 011 388 100 100 100 100 100 100 SK 2 069 760 432 94 579 115 85 : : : : : FI (6) 5698 2295 1434 1210 271 488 65 69 55 58 52 95 SE (7) 7422 3197 2193 411 1595 26 93 91 95 93 92 94 UK 156 542 59 787 30 403 11 927 53 683 742 100 100 100 100 100 100 BG 8 084 2 133 1 118 454 4 222 147 97 95 94 92 96 100 RO 7 883 1 951 920 758 3 900 354 96 92 96 93 90 96 TR : : : : : : : : : : : : NO : 1 780 1 082 203 1 191 : 89 86 92 86 92 : CH (7) 18 043 8 050 3 848 1 491 3 985 669 39 41 44 28 35 49 (1) Coverage rates refer to 2000. (2) Coverage rates refer to 2003. (3) Coverage rates were not provided and are assumed to be 100% as the survey is complete. (4) Nationally defined population includes units with more than 9 persons employed. (5) Coverage rates based on sold industrial production from 2003, not value added. (6) The coverage rate is believed to refer to the survey only: it is assumed that the inclusion of the administrative data increases the coverage rate to 100%. (7) Coverage rates based on employment, not value added. Note: the number of units for total industry should equal the sum of the number of units for the five MIGs, but this is not the case in all countries. Energy Industry total Intermediate goods Capital goods Consumer durables Consumer non-durables Energy PEEIs in focus A summary for the industrial production index 10

Figure 1.3a: Share of MIGs in the total number of units in the nationally defined population (sum of countries with complete available data) 33% 2% 38% Intermediate goods Capital goods Figure 1.3a summarises the distribution of the population by MIGs, using an aggregate for all countries with a full set of information available as presented in Table 1.3. The coverage rate of the national defined population relative to a full coverage for NACE Sections C to E is presented in Figure 1.3b. Consumer durables Consumer nondurables 7% Energy 20% Figure 1.3b: Coverage rates of the nationally defined population relative to the STS requirements. 100 90 80 70 60 50 40 30 20 10 0 CY LT (1) NL SI UK HU IE BG RO ES PL (2) AT DE SE (3) FR NO LU LV EE SK DK IT FI (4) CH (3) BE (5) EL (5) TR (5) (1) Assumed to be 100%. (2) Coverage rates based on sold industrial production from 2003, not value added. (3) Based on employment, not value added. (4) The coverage rate is believed to refer to the survey only: it is assumed that the inclusion of the administrative data increases the coverage rate to 100%. (5) Not available. PEEIs in focus A summary for the industrial production index 11

2. Selection of units 2.1 Basic method of selection This point looks at how units are selected from the nationally defined population in order for them to be the subject of the statistical survey. Units can be selected in a number of ways, essentially by taking all units, or a sample. If all units are taken it may be a case that really all are taken, or all above a certain size. If a sample is taken it may be a random sample, or a purposive/judicious sample. Table 2.1 shows the methods used in each country, with the countries regrouped according to the method used. BE EE IE LV LU SK RO DK CY LT HU NL PL SI SE UK BG NO CH Method of selection Exhaustive survey within the nationally defined population Sample Table 2.1: Use of sampling and cut-off criteria If sample, threshold (inclusive) for exhaustive strata If cut-off, criteria used 200 persons employed 20 persons employed or CYP 1 million 50 persons employed 50 persons employed 50 persons employed 50 persons employed 20 employees Not specified 150 "employment" 100 persons employed 100 persons employed Not specified 75% of gross production value for 4-digit NACE level for each federal DE state 70% of total production of survey commodities and sample of EL establishments at 4-digit level ES Cut-off NACE class: 95% of production value; product level: 95% of production value. FR 80% of production IT 70% production value by product AT 90% of total production in each NACE Division is covered. FI Gross value between 40% and 100% TR (1) 80% of production value in Section D (1) Cut-off for Section D; exhaustive survey for Section C. Note: in general it can be assumed that all samples are random samples - this was explicitly mentioned by several countries using sampling, although in Poland a purposive sampling is used.. Note: in some countries the size threshold may be based on a different measure of employment than the one stated, as the number of (paid) employees and the total number of persons employed are often confused. The simplest method (though often with a high cost and burden) is to take all units, in other words a census. Because of the high cost this is rarely feasible, except when the population has in fact been reduced, normally by excluding smaller enterprises, thus generally excluding a large number of smaller units: typically smaller units account for a very high proportion of the number of units, a smaller proportion of employment and an even smaller proportion of output. Approximately one quarter of the countries use this method. Note that all of the countries that survey all units within the nationally defined population use a size threshold to determine the nationally defined population, so none of them carries out a complete census to compile the IPI. A similar method is to include only the larger enterprises, not by systematically excluding the smaller ones, but by selecting the largest enterprise, and then selecting the next largest and so on until those selected represent a certain proportion of the activity of the whole population, for example 70% of production. This can be done by looking at the results of a more complete survey, for example an annual product survey (such as PRODCOM) or an annual survey providing structural business statistics. Again, approximately one quarter of the countries use this method. PEEIs in focus A summary for the industrial production index 12

Whether units are excluded by only taking units above a certain size, or by taking the largest units until a certain proportion of output is covered, the systematic exclusion of part of the population leads to the risk of a bias in the resulting indicators, as the evolution over time of the output of smaller units may be different from that of larger units. This is similar to the exclusion on size 2.2 Stratification criteria used and sample rates In the case of the IPI it is normal to use activities (NACE headings) or products (CPA or Prodcom headings) as one of the criteria for drawing a sample, and size is grounds of a part of the population from the nationally defined population. The alternative to these methods is to select units using a sample, and this is the most common method for the IPI, used by 12 of the 26 countries. A sample may be a simple sample drawn from the whole population, or a stratified sample. More information on stratified sampling and sample rates is given in the next sub-point. commonly also used see Table 2.2a. Region is a further criteria that may be used. Table 2.2a: Criteria used for sampling (among countries carrying out sampling) Criteria 1 Criteria 2 Criteria 3 DK Activity "Employment" CY Activity Employees Value of production LT Activity (NACE 4-digit) Persons employed HU Activity Persons employed Location of headquarters (capital/countryside) NL (1) Activity "Employment" PL Activity Persons employed Region SI (1) Activity Employees Turnover SE Activity Employees UK Activity "Employment" BG Activity (NACE 3-digit) Persons employed NO (1) Activity Employees CH (1) Activity "Employment" (1) Activity is not mentioned as a stratification criteria, however it can be assumed that this is used. Note: in some countries the size threshold may be based on a different measure of employment than the one stated, as the number of (paid) employees and the total number of persons employed are often confused. Table 2.2b: Countries carrying out sampling: sample rates (%) Employment thresholds (1) 1-4 5-9 10-19 20-49 50-99 100-199 200+ DK 0 0 0 20 45 75 100 CY Not specified 100 100 100 100 LT 17 17 56 56 100 100 100 HU (2) 0 10 14 22 100 100 100 NL (3) 0 0 0 87 100 100 100 PL 0 0 10 10 100 100 100 SI SE Not specified Not specified UK 1 1 5 5 19 19 100 BG 0 0 24 24 24 100 100 NO 0 0 17 33 50 100 100 CH Not specified (1) Use of employees or persons employed thresholds not separately identified. (2) Units with less than 5 persons employed are covered using administrative data. (3) Units with less than 20 persons employed are estimated. Nine of the 12 countries using sampling provided information on the sample rates by employment size class. Of these nine, five had sampling rates of 100% (full coverage, full enumeration, census) from an employment level of 50, and in one of these cases from a lower level. PEEIs in focus A summary for the industrial production index 13

2.3 Sample size and coverage Table 2.3 shows the number of selected (surveyed) units, and their distribution between the five MIGs. Furthermore, it shows the proportion of value added of the nationally defined population that is covered by the selected units. It should be noted that the number of units in the sample does not in all cases equal the sum of the sample sizes for the five MIGs. When considering the proportion of value added of the nationally defined population that is covered by the selected units it is useful to also bear in mind the importance of a size threshold in determining the nationally defined population, as was presented in Table 1.3. Table 2.3: Sample size and coverage rate Coverage rates (share of value-added among selected Number of units in sample units, relative to the nationally defined population) (%) Total Intermediate goods Capital goods Consumer durables Consumer non-durables BE 7 269 2 946 1 769 499 2 217 0 94 96 95 91 91 0 DK 1 307 369 486 117 322 13 81 76 78 74 85 100 DE (1) 14 710 7 563 3 819 663 2 612 53 76 80 75 80 71 81 EE 937 287 145 74 273 158 87 85 84 83 85 98 EL 2288 1049 268 198 878 14 95 : : : : : ES 13 205 5 710 2 156 882 3 978 479 79 79 80 78 78 80 FR 6 773 2 263 1 287 225 2 986 12 82 83 90 61 70 86 IE 1 961 690 370 168 609 124 100 100 100 100 100 100 IT 3140 : : : : : 89 : : : : : CY 832 346 76 58 339 13 100 100 100 100 100 100 LV 1 211 476 119 64 454 98 100 100 100 100 100 100 LT (2) 3 332 1 250 404 228 1 312 138 95 92 96 93 93 99 LU 171 90 33 1 41 6 94 98 83 100 86 98 HU (2) 4 414 1 490 957 265 1 365 337 94 91 94 99 89 100 NL (2) 6 139 2 200 1 952 328 1 638 21 : 90 92 96 90 : AT 6 313 2 194 1 315 707 1 981 117 89 91 89 80 86 89 PL (3) 12 808 4 357 2 344 722 4 562 823 83 80 83 81 75 100 SI 1731 775 311 118 459 68 72 72 75 82 61 79 SK 2069 760 432 94 579 115 : : : : : : FI 1269 443 186 227 48 365 61 65 52 53 42 92 SE 2057 913 539 171 416 18 : : : : : : UK 9 139 3 323 2 568 2 021 608 619 32 17 5 10 8 97 BG 3 076 882 467 149 1 445 133 92 90 90 78 86 99 RO 7 883 1 951 920 758 3 900 354 96 92 96 93 90 96 TR 4809 : : : : : 80 : : : : : NO 1410 625 341 87 357 : 71 73 67 69 70 : CH (4) 7000 2000 1100 380 1000 80 15 10 12 7 9 6 (1) Coverage rates based on gross production value. (2) Coverage rates based on turnover. (3) Coverage rates based on value of sold industrial production, 2003. (4) The total is greater than the MIGs because the total also includes units that provide information through trade associations, rather than directly. The number of units provided is in fact the number of activities in all of the "establishments" sampled. The coverage rates are given in employment terms, and do not seem comparable with other countries. Note: the number of units for total industry should equal the sum of the number of units for the five MIGs, but this is not the case in all countries. Energy Total Intermediate goods Capital goods Consumer durables Consumer non-durables Energy PEEIs in focus A summary for the industrial production index 14

Figure 2.3a: Share of MIGs in the total number of units in the sample (sum of countries with complete available data) 29% 4% 38% Intermediate goods Capital goods Figure 2.3a summarises the distribution of the sample between the MIGs as an aggregate for all countries with a full set of information available in Table 2.3. It can be compared with Figure 1.3 which showed a similar distribution between the MIGs for the nationally defined populations. Figure 2.3b shows the distribution of the sample between the MIGs for all countries. Consumer durables Consumer non-durables 8% Energy 21% 100% Figure 2.3b: Share of each MIG in the sample (in terms of the number of units) (1) 80% 60% 40% 20% 0% BE DK DE EE EL ES FR IE CY LT LU HU NL AT PL SI SK FI SE UK BG RO CH Intermediate goods Capital goods Consumer durables Consumer non-durables Energy (1) Italy, Latvia, Turkey and Norway: incomplete or not available. PEEIs in focus A summary for the industrial production index 15

Figure 2.3c ranks the countries according to the absolute size of their sample. Relative to the size of their industrial economies most of the countries that are EU Member States since 2004 and the Candidate countries have quite large samples in this respect it should be noted that many of these are quite small countries. Nevertheless, the larger countries, Romania and Poland also have quite large samples relative to the size of their industrial economies. Among the countries that were EU Member States prior to 2004, Belgium and Austria both have quite large samples relative to the size of their industrial economies. The largest Member States have generally the largest samples in absolute terms, with the notable exception of Italy, but relative to the size of their industrial economies their samples are quite small, except in Spain. 16 14 12 14.7 13.2 12.8 Figure 2.3c: Sample size: number of units (thousands) 10 8 9.1 7.9 7.3 7.0 6.8 6.3 6.1 6 4 2 0 4.8 4.4 3.3 3.1 3.1 2.3 2.1 2.1 2.0 1.7 1.4 1.3 1.3 0.9 0.8 0.2 DE ES PL UK RO BE CH FR AT NL TR HU LT IT BG EL SK SE IE SI NO DK FI EE CY LU LV (1) (1) Not available. 2.4 Frequency of updating of the sample Nearly all of the countries using a sample for selecting units update their sample every year, Cyprus and Switzerland being the only exceptions. Among the countries using a cut-off (selecting only the largest units until a particular fraction of an output indicator has been reached), four update their selection every year, two less frequently, and one did not specify how frequently the selection is updated. Table 2.4: Frequency of updating the sample Frequency DK Annual CY 5 years LT Annual HU Annual NL Annual PL Annual SI Annual SE Annual UK Annual BG Annual NO Annual CH Multi-annual: last in 2001-2003 DE Annual EL 5 years ES (1) Base year change FR Annual IT Base year change AT Annual FI Annual TR Not specified (1) More frequent updates are made to account for births and deaths. Countries using a sample Countries using cut-off PEEIs in focus A summary for the industrial production index 16

3. Type of data collected 3.1 Basic method(s) used and their relative importance One of the key issues relating to the compilation of the IPI is the choice of "basic method". This stems from the difficulty of collecting value added data on a monthly basis, combined with the requirement for a volume index. As a consequence a variety of methods are used. Three main methods can be identified, and these are: method 1 - measuring as a value the output of the observation unit as a whole, and then using a deflator at some stage of the compilation to produce a volume index; the output may be measured in terms of turnover, production value or value added; method 2 - measuring as a value the output (sold or total production, or production intended for sale) for products, and then using a deflator at some stage of the compilation to produce a volume index; method 3 - measuring the physical quantity (tonnes, litres, pieces etc.) of output at the product level. In addition to these three basic methods, two supplementary basic methods are used: method 4- measuring the hours worked; method 5- measuring the physical quantity (or deflated value) of inputs (raw materials or energy). All of these methods present advantages and disadvantages, and most countries use a combination of basic methods, with one of the first three basic methods dominating. The methods used for various activities are presented in Table 3.1, which for each method used indicates approximately the proportion of value added for which this method is used. PEEIs in focus A summary for the industrial production index 17

Deflated output of the observation unit (turnover, production value or value added) Deflated output at the product level (deflated production values) Table 3.1: Basic methods used (part 1) Quantity of output, at the product level (physical quantities) Hours worked Other methods NACE concerned % value added NACE concerned % value added NACE concerned % value added NACE concerned % value added Method NACE concerned % value added BE (2) (6) DK DE (6) EE EL (3) ES A. 14 (not 1421), 171, 176, 191, 2111, 241, 2441, 2710, 2735, 274 (production) B. The rest (sold production) 80% Sections C and D excluding Division 11 and Group 35.1 (turnover) 82% 11 and Section E 17% 35.1 1% 1513, 1581, 2112, 2461, 271, 2732, 2960 (turnover) 5% The rest 78% 101, 102, 111, 1583, 231, 401, 402 8% 3511, 3512, 352, 353, 45 9% Section D (turnover adjusted for changes in stocks) 83% Sections C and E 17% parts of 17, 18, 19, 20, 22, 24, 25, 29, 31, 34, 35 and 36; 30, 32, 33 21% The rest 78% 2852, 351, 353 1% 1519, 1551, 2330, 351, Input (quantity 37, 45 20% or value) 41 1% FR 183, 242, 244, 268, 286, 291, 292, 294, 295, 297, DL (not 33.5), 343, 361 (turnover) 28% The rest 66% 283, 285, 296, 351, 352 6% IE Not specified (turnover, net selling value) 54% Not specified 46% IT (4) DG, DJ, DK, DL, DN 11.2% Sections C and E, Subsections DA to DF, DH and DI 83% DK, DL, DM 6% CY (8) LV All NACE (gross output) 100% LT All NACE (turnover) 100% PEEIs in focus A summary for the industrial production index 18

Table 3.1: Basic methods used (part 2) Deflated output of the observation unit (turnover, production value or value added) Deflated output at the product level (deflated production values) Quantity of output, at the product level (physical quantities) Hours worked Other methods NACE concerned % value added NACE concerned % value added NACE concerned % value added NACE concerned % value added Method NACE concerned % value added LU (5) 174, 204, 2212, 2222, 2524, 2663, 2682, 2751, 2811, 2812, 2852, 2862, 2875, 2922, 2951, 2953, 2956, 2971, 314, 342, 364 15% The rest 2212, 2222, 271, 2922, 311 (hours of manual 75% workers) 10% HU All NACE (turnover corrected for stock changes of own production) 100% NL A. 15(52.4%), 17, 19, 20, 21, 22, 23(10.6%), 24, 25, 26, 27(42.2%), 28(58.3%), 29(29.7%), DL, 34(40.4%), 35(23.4%), 36(43.0%), 37 (turnover, with/ without correction for change in inventories) B. 27(57.8%) (value added) A. Input A. 55.6% B. 1.6% 15(30.8%), 16, 23(89.4%), 34(59.6%) 28(41.7%), 29(70.3%), 27% 35(67.6%) (volume of consumption) 8% B. Unspecified A. 15(9%) B. 15(7.7%), 18, 36(57%) A. 1% B. 5.9% 10(90%), 11(92%), 13(63%), 14(77%), 15(73%), 16(90%), 17(71%), 18(57%), 19(57%), 20(52%), 21(90%), 25(80%), 26(80%), 26(74%), 27(84%), 28(59%), 29(54%), 30(65%), 31(53%), 32(51%), AT 22(57%), 23(63%), 24(51%), 33(50%), 36(71%) 34(52%), 35(52%), 37(81%), 40(77%), 38% 41(92%) 62% PL All NACE (sold production) 100% SI All NACE 100% 5 products (in Divisions Input (quantity SK 138 products 20% 395 products 80% 29 and 35) 0% or value) 4 products (in Groups 22.1 and 22.2) 0% PEEIs in focus A summary for the industrial production index 19

Table 3.1: Basic methods used (part 3) Deflated output of the observation unit (turnover, production value or value added) Deflated output at the product level (deflated production values) Quantity of output, at the product level (physical quantities) Hours worked Other methods NACE concerned % value added NACE concerned % value added NACE concerned % value added NACE concerned % value added Method NACE concerned % value added FI 28,29,33,35,36 (parts of) (turnover) 1% 17, 20, 22, 24-37 (parts of) 12% All NACE Divisions (parts of) 22, 24, 25, 28, 29, 31, 32, 82% 34, 35, 36 (parts of) Input (quantity 4% or value) 10, 132, 1581, 1582, 233, 131, 151, 1551, 1583, 16, 201, 2111, 296, 311, 351 to 353, SE The rest (turnover) 75% 2112, 232, 2611 to 2613, 271 to 275, 40 20% 355, 37 5% Rest of Section D (turnover adjusted for change in stocks of finished goods and work in progress) 73% Sections C and E, Divisions 15, 23, 27 and Group 34.1 27% UK BG All NACE (production value) 100% RO All NACE 100% 22, 35 (parts of) 1% TR Groups 29.7, 32.1 and 34.1 and Class 29.31 20% The rest 80% NO 10-21, 23 to 27, 36 to 40; 29.7, 31.1 22, 28 to 35, except 85% groups 29.7, 31.3 15% CH (7) Sections C and D 100% (1) The distinction between quantity and volume indicators was probably misunderstood by some countries. (2) Hours worked are productivity corrected. (3) A mixture of value and quantity data by product - no further details provided. (4) The method concerning deflated value of output at the product level may in fact be the quantity of output at the product level; note that Subsection DK is allocated to two different methods. (5) For some activities more than one method is used. (6) Value added shares were provided including Section F, but could not be adjusted to exclude this. (7) In different parts of the report reference is made t deflated turnover and product quantities. (8) Not specified. PEEIs in focus A summary for the industrial production index 20

In Figure 3.1a the countries have been regrouped according to the dominant (sometimes exclusive) basic method used. For countries that mainly use method 1 (value of output of the observation unit as a whole, rather than product data) no use at all is made of method 2 (value of output by product). Equally countries that mainly use method 2 (value of output by product) only tend to use method 1 (value of output of the observation unit as a whole) to a small extent. Countries that use mainly method 3 (physical quantities of output by product), do not use method 1 (value of output of the observation unit as a whole) except France, but several also use method 2 (value of output by product). Method 4 (hours worked) is used by many countries, regardless of their main basic method. All of the countries that use one method exclusively are countries that have been EU Member States since 2004 or are non Member States: only EE, SK and TR of the countries that have been EU Member States since 2004 and Candidate countries are using a mixed method. 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% BG UK SE PL Figure 3.1a: Basic methods used, share of value added (1) NL HU LT LV EE DK IE BE DE Deflated output value (of the unit) Deflated value of product output Physical quantity of product output Hours worked IT Other (input or unspecified) (1) The distinction between quantity and volume indicators was probably misunderstood by some countries; CY not shown as method not specified. (2) In different parts of the report reference is made to deflated turnover and product quantities. SI RO CH (2) NO FI SK TR ES LU FR AT Table 3.1b: Countries mainly using deflated output of the observation unit: activities where they use physical quantity of output at the product level (1) Section Division DK EE NL SE UK 10 x x 11 x x x C 12 x x 13 x x x 14 x x 15 x x x 16 x x 20 x D 21 x 23 x x x 26 x 27 x x 34 x x E 40 x x x 41 x x (1) The distinction between quantity and volume indicators was probably misunderstood by some countries; Ireland, not available. Table 3.1b shows, for countries using mainly method 1 (value of output of the observation unit as a whole), which are the activities where they use method 3 (physical quantities of output by product). There are a few activities where only one country uses this method, and in all of these activities the country concerned is Sweden. Parts of Sections C and E, and Divisions 15 and 23 are activities where countries mainly using method 1 use method 3 instead. PEEIs in focus A summary for the industrial production index 21

Table 3.1c shows, for countries using mainly method 3 (physical quantities of output by product), which are the activities where they use method 2 (value of output by product). The range of industrial Divisions not included in the table, in other words where no country using mainly method 3 chooses to use method 2, is quite short, and includes the following Divisions: 10, 12, 16, 37 and 41. There are a number of industrial Divisions where several countries that use mainly method 3 choose to use method 2, and these are the following Divisions: 20, 22, 29, 32, 34 and 36. Table 3.1c: Countries basically using physical quantity of output at the product level: activities where they use deflated output value at the product level (1) Section Division ES LU AT (2) SK FI TR C 11 11 14 14 15 15 17 17 174 173, 1754 17 18 18 1824, 183 19 19 19 20 20 204 20 20 21 21 22 22 2212, 2222 22 2215, 2222 to 2225 22 23 23 232 24 24 24 2442, 2452 25 25 2524 2522, 2523 D 26 2663, 2682 2612, 2615, 2661, 2665, 2666, 2682 27 2751 27 28 2811, 2812, 2852, 2862, 2875 2811, 2822, 283, 2851, 2863, 2875 29 29 2922, 2951, 2953, 2956, 2971 291, 292, 2932, 294, 295, 296 2931, 297 30 30 30 31 31 314 311, 312, 3162 32 32 32 321 33 33 33 33 34 34 342 34 341 35 35 3511, 352, 353, 354 36 36 364 36 3614, 3622, 364, 365, 3663 E 40 402 (1) The distinction between quantity and volume indicators was probably misunderstood by some countries. (2) For each NACE Division there are deflated production values as well as physical quantities. This table shows only those which are concentrated in one of the categories. PEEIs in focus A summary for the industrial production index 22

Table 3.1d shows the activities where countries use method 4 (hours worked), regardless of the main basic method that they use. The only country that has been an EU Member State since 2004 using the hours worked method is Slovakia. The activities where several countries use method 4 include the following Divisions: 28, 29, 31 and 35 (in particular Group 35.1). Table 3.1d: Activities where hours worked are used: organised by NACE Division Section Division BE DK DE ES FR LU NL SK FI SE NO C 10 10 13 132 15 1519, 1551 1581, 1582 22 2212, 2222 22 22 23 233 233 24 24 25 25 27 271 28 2852 283, 285 28 28 28 D 29 296 2922 29 2911, 2922, 2924, 296 29 296 29 (not 29.7) 30 30 31 311 31 311 31 (not 31.3) 32 32 32 33 33 34 34 34 351, 352, 353 351, 353 351, 352 35 3511 35 35 351 351 36 36 37 37 37 351, 352, 353, 355 35 PEEIs in focus A summary for the industrial production index 23

3.2 Treatment of quality changes (if quantity data used) When basic method 3 (physical quantities of output by product) is used or method 5 (raw material or energy inputs) using physical quantities rather than values, the issue of changes in the quality of the products is important. If there is a change in the quality of a product produced, this should be reflected by a change in the IPI, as this is a volume index. Table 3.2 shows all of the countries that use basic method 3, and indicates the extent to which this issue is addressed; as can be seen few countries make a systematic effort to adjust for quality changes. Table 3.2: Treatment of quality change by countries using physical quantities by product as one of their methods (1) Countries using this method as their most common method. Countries not using this method as their most common method. SI RO CH (2) % of value added using this method Treatment of quality changes 100 None 100 None 100 None No explicit treatment of quality changes for minor activities and old products. When quality changes occur for NO 85 major activities this is treated as a new product and introduced in the index-calculation in January. Output quality changes are not recognised as such. Quality changes are recognised to some extent as an annual volume index is calculated and in doing so quality changes are approximated by setting acceptance limits to the unit value ratio. The acceptance limits are based on the producer price index, in which quality changes have been taken into account. If the unit value changes by more than the producer price index of the industry in question FI 82 multiplied by +/? 40 per cent, it is concluded that we have a new product. SK 80 None TR 80 Not specified ES 78 None LU 75 Not specified FR 66 None AT 62 A new basket of products is introduced every 5 years. IE 46 None. CY :None NL 27 None; this method is used for relatively homogeneous products. Not specified, except for motor vehicles: the quantity is quality adjusted by using data from the Society of Motor UK 27 Manufacturers and Traders on the engine size and fuel types of vehicles produced in the month. SE 20 None DK 17 Not specified EE 17 In the case of significant quality changes new products are taken into account and the time-series recalculated. DE 8 None (1) The distinction between quantity and volume indicators was probably misunderstood by some countries. (2) In different parts of the report reference is made to deflated turnover and product quantities. PEEIs in focus A summary for the industrial production index 24

Part B: Non response 4. Non-response rates The issue of non-response can be evaluated from a number of perspectives, three of which are presented here. The first is a simple snapshot, of the non-response rate, the second shows how this evolves over the course 4.1 Snapshot of non-response Table 4.1a: Non response rate (unweighted) Non-response (%) Timeliness, if specified BE 27.9 After 1 month DK 0.8 After just over 1 month For the preliminary results - timescale not DE 10.5 specified EE 0 EL 5 ES 2.5 FR 6 IE 50 After 65 days IT 8-10 CY 8 LV 0 LT 4.9 After 1 month LU Less than 5 HU 5.6 NL 29 AT 17 After 60 days; weighted by value added PL 3.7 SI 2 SK 5.9 FI 0.6 After just less than 1 month SE 12.6 UK 22 BG 1.96 RO 8.3 TR 3 NO 5 CH 20 of one monthly round of data collection, and the third looks at the differences in response rates for the 12 calendar months. Table 4.1a provides a snapshot of the rate of nonresponse, country by country. As will be seen in the next point, the non-response rate develops over the course of a period of data collection, and so to be able to compare non-response rates it is important to take a snapshot at the same point in time, either after a fixed period of time such as 35 days from the end of the reference period, or at a benchmark moment such as the first date for publishing the index. Unfortunately, the benchmark point in time was not specified: a number of countries spontaneously indicated the reference time at which the non-response rate was measured and this is also indicated in Table 4.1a. Similar information for each of the MIGs, and by size class is provided in Table 4.1b for the countries that provided this information. LT (persons employed) HU (persons employed) NL (employees) Table 4.1b: Non response rate (unweighted) (%) Intermediate Total industry goods Capital goods Consumer durables Consumer nondurables Size class Energy 1-9 12 : : : : : 10-49 4 : : : : : 50 and more 1 : : : : : 5-49 11 11 11 12 14 0 50 and more 2 2 2 3 2 0 Total (5 and more) 6 6 6 7 6 0 Less than 50 21 25 28 18 16 28 50 and more 20 22 22 14 23 3 SE (employees) 10-19 16 18 11 28 16 0 20-49 12 13 11 14 8 0 50-99 15 16 19 7 11 0 100-199 14 13 14 30 12 0 200-499 6 6 5 21 6 0 500 and more 2 1 1 0 4 0 PEEIs in focus A summary for the industrial production index 25