Data Sources and Methodology of the 60-Industry Database of the Groningen Growth and Development Centre. Version: October 2005

Transcription

1 Data Sources and ethodology of the 60-Industry Database of the Gronngen Growth and Development Centre erson: October 2005 Introducton The 60-Industry Database of the Gronngen Growth and Development Centre provdes a comprehensve nternatonally comparable dataset on ndustral performance at a detaled ndustry level for 26 OECD countres and Tawan (see table 1). It enables the user to desgn ndcators such as shares of output and employment by ndustry and to analyse trends n value added and labour productvty. arables covered nclude current value added, value added deflators, persons engaged and hours worked for 56 ndustres for the perod The 60-Industry Database updates and extends prevous work at the Gronngen Growth and Development Centre (GGDC) and the Natonal Insttute on Economc and Socal Research (NIESR) recently descrbed n O ahony and van Ark (2003) (downloadable from For most varables and countres, the OECD STructural ANalyss (STAN) database s taken as the pont of departure, whch n turn s largely based on recent natonal accounts of ndvdual OECD member states. 2 In some cases, own countres natonal accounts data s drectly used for reasons of greater detal or more up-to-date fgures. The STAN data and those from the natonal accounts are complemented, updated and backdated as well as further dsaggregated by the use of nformaton from ndustry surveys and hstorcal sources for ndvdual countres to provde a comprehensve data set for the perod In general the method employed was to use natonal accounts aggregates as control totals and the other data to dvde these totals nto sub-ndustres. 3 In a lmted number of cases, STAN and natonal accounts also nclude workng hours per person employed (or per ob), but for many countres alternatve sources, such as labour force surveys or other labour market statstcs, had to be employed to obtan these estmates. 1 Seres on the number of employees and labour compensaton are not avalable for all countres. 2 See 3 Appromxately 50 per cent of ndvdual cells n the 60-Industry Database s drectly derved from OECD STAN, whereas the other 50 per cent represents extensons. 1

2 Country Coverage Table 1 provdes a lst of countres covered n the database. The database covers 26 countres, of whch 26 are OECD members as well as Tawan. All 15 OLD EU countres 4 are covered, and aggregate estmates for the EU as a whole are also provded. Table 1 Lst of countres n databases Belgum BE Unted Kngdom UK Denmark DK Unted States US Germany* DE Canada CA Greece GR Australa AU Span ES Norway NO France FR Swtzerland CH Ireland IR Japan JP Italy IT South Korea SK Luxembourg LU Tawan TA Netherlands NL Czech Republc CZ Austra AT Hungary HU Portugal PT Poland PO Fnland FI Slovak Republc SL Sweden SW *separate seres are provded for West Germany for and for unfed Germany from Industry coverage Data are provded for 56 ndustres n total. 5 The ndustres are classfed accordng to the Internatonal Standard Industral Classfcaton (ISIC) revson 3. Ths classfcaton s very close to the European NACE rev 1 classfcaton system. Table 2 provdes a lstng of the ndustres. The ndustry dvson s more detaled than n STAN whch allows a focus on ndustres whch are charactersed by hgh ICT-nvestment shares and/or ICT-goods producton. For example, addtonal entres were ncluded for fve of the sx ICT producng ndustres (semconductors, ISIC 321; communcaton equpment, 322, fber optcs, 313; rado and T equpment, 323; and nstruments, 331). Separate entres are also ncluded for professonal and other busness servces (ISIC and 749) and agrculture and forestry (01 and 02). 4 OLD EU-15 refers to membershp of the European Unon untl 30 Aprl Despte the coverage of 56 ndustres, we have called the database 60-Industry database as we am to expand the ndustry detal a lttle further n the near future. For example, n due course, nng and quarryng wll be dvded nto nng and quarryng of energy producng materals and nng and quarryng except energy producng materals, Food products, beverages and tobacco wll be dvded nto Food products and beverages and Tobacco products, Chemcals and chemcal products wll be dvded nto Chemcals excludng pharmaceutcals and Pharmaceutcals, Basc metals wll be dvded nto Iron and steel and Non-ferrous metals, anufacturng nec; recyclng wll be dvded nto anufacturng nec and Recyclng, Electrcty, gas and water supply wll be dvded nto Electrcty and Gas and water supply. 2

3 Table 2 Industres n the 60-Industry database Industry Name ISIC rev 3 TOTAL ALL INDUSTRIES Agrculture 01 2 Forestry 02 3 Fshng 05 4 nng and quarryng Food, drnk & tobacco Textles 17 7 Clothng 18 8 Leather and footwear 19 9 Wood & products of wood and cork Pulp, paper & paper products Prntng & publshng neral ol refnng, coke & nuclear fuel Chemcals Rubber & plastcs Non-metallc mneral products Basc metals Fabrcated metal products echancal engneerng Offce machnery Insulated wre Other electrcal machnery and aparatus nec Electronc valves and tubes Telecommuncaton equpment Rado and televson recevers Scentfc nstruments Other nstruments otor vehcles Buldng and reparng of shps and boats Arcraft and spacecraft Ralroad equpment and transport equpment nec Furnture, mscellaneous manufacturng; recyclng Electrcty, gas and water supply Constructon Sale, mantenance and repar of motor vehcles and motorcycles; retal sale of automotve fuel Wholesale trade and commsson trade, except of motor vehcles and motorcycles Retal trade, except of motor vehcles and motorcycles; repar of personal and household goods Hotels & caterng Inland transport Water transport Ar transport Supportng and auxlary transport actvtes; actvtes of travel agences Communcatons Fnancal ntermedaton, except nsurance and penson fundng Insurance and penson fundng, except compulsory socal securty Actvtes auxlary to fnancal ntermedaton Real estate actvtes Rentng of machnery and equpment Computer and related actvtes Research and development Legal, techncal and advertsng Other busness actvtes, nec Publc admnstraton and defence; compulsory socal securty Educaton Health and socal work Other communty, socal and personal servces Prvate households wth employed persons 95 3

4 arables The followng varables are covered: alue added s current gross value added measured at producer prces or at basc prces, dependng on the valuaton used n the natonal accounts. It represents the contrbuton of each ndustry to total GDP. alue added s expressed n natonal currences, and the measures are therefore not corrected for dfferences n purchasng power of the currences. alue added n Euro-countres before the ntroducton of the euro are converted to euros wth the 1999 offcal fxed euro converson rates (see table 3). Deflator s the change n the value added deflator. It can be combned wth current value added to derve quantty ndces of real value added at ndustry level. Persons engaged comprses number of workers engaged n producton, ncludng employees as well as self-employed, workng propretors and unpad famly workers. Hours refers to average annual hours worked per employee or per person engaged. alue added volume ndces (1995 =100) (for more detals see below n text on aggregaton ), alue added per person employed and value added per hour worked are volume ndces (1995 =100) and are mputed from the varables reported above. Table 3: 1999 Euro converson rates (natonal currency/euro) Austra Ireland 0.79 Belgum Italy Fnland 5.95 Luxemburg France 6.56 Netherlands 2.20 Germany 1.96 Portugal Greece Span Sources The constructon of the 60-Industry Database was done n a two-step procedure to ensure natonal accounts compatblty. For most countres the basc pont of departure s the STAN database. STAN provdes data whch s generally based on the latest offcal natonal accounts data of ndvdual OECD member states. Due to changes n ndustral classfcatons and the ntroducton of the new 1993 System of Natonal Accounts (SNA) and the 1995 European System of Accounts (ESA), offcal seres often lack ndustry detal, especally for the perod before the 1990s. In some cases, STAN goes beyond the offcal publshed data and provdes data at more detaled levels based on addtonal sources, but mportant gaps stll reman. For a number of countres data were drectly obtaned from the natonal accounts nstead of from the OECD STAN database. For some countres there s no or only lmted OECD STAN 4

5 data (.e. Ireland, Tawan), whereas for a few other countres natonal accounts data was more up-to-date than the OECD STAN data. As a second step the STAN or natonal accounts data were complemented wth nformaton from detaled ndustry and servces statstcs and addtonal (hstorcal) natonal accounts data for ndvdual countres. Use was made of nternatonal statstcs such as the OECD Structural Statstcs for Industry and Servces, the OECD Servces Statstcs on alue Added and Employment, Eurostat Structural Busness Statstcs and the Eurostat Labour Force Survey. Furthermore natonal sources have been used ncludng natonal accounts and economc census and survey materal. The spreadsheets assocated wth each of the country fles provde a full account of the sources used for each country, each year and each varable. Fllng procedure In case of mssng data there are bascally two procedures for estmatng value added, employment and compensaton data by ndustry: (1) applyng shares from addtonal data to hgher level aggregates or (2) applyng hgher-level growth rates to more detaled levels. The frst s most useful when for a partcular sub-sector there s no data avalable for any year. In that case, the share of the sub-sector n some hgher level aggregate s derved from addtonal secondary data sources and appled to the aggregate n the basc source. In case data s avalable n the basc source for some years, secondary data shares are used for mssng years provded they correspond closely to the basc source. If not, growth rates from secondary data are appled to the orgnal basc data for mssng years. To mantan natonal accounts compatblty a normalsaton procedure s used so that subsectors add to the correspondng hgher-level ndustry aggregates provded n the basc source. If there s a summaton dscrepancy, the subsectors absorb the resdual. Each sub-sector does so n proporton to ts weght n the parent ndustry. Ths procedure ensures that output and employment measures are natonal accounts compatble and, mportantly, have the same economy-wde coverage. 6 For seres on hours worked per person and deflators for value added, gaps n the basc source (STAN or natonal accounts data) are flled n a smpler way. If no addtonal data could be found, hgher level aggregates have been used. For example, n case no separate fgure on hours worked per person for sub-sector 351 could be found, the fgure for 35 (when avalable) s used nstead. The same was done for deflators where necessary. 6 Often addtonal data s taken from surveys. Samplng coverage and defntons n survey data can dffer wthn and across countres. 5

6 The source descrptons n each of the country fles provde a detaled account of the fllng procedures used for each country, year and varable. Alternatve deflators for ICT producton At present there are only a few countres that have an adequate system n place for measurng prces of computers and semconductors whch take nto account the fast ncrease n qualty of these goods. To acheve nternatonal comparablty, harmonsed US deflators are appled for two ICT-producng manufacturng ndustres (ISIC 30 and 321) n all countres. US value added deflators are corrected for dfferences n overall nflaton between each country and the US. Inflaton s measured as the change n the deflator of all ndustres, excludng the ICTproducng manufacturng ndustres. 7 Snce the ICT-producng ndustres are not separately dstngushed n the US Natonal Income and Product Accounts, these deflators were constructed usng prce changes n output and ntermedate nputs of ICT goods as follows: 1) alue of shpments deflators for manufacturng ndustres and gross output deflators for non-manufacturng ndustres were obtaned from the BEA data sets on Gross Output by Detaled Industry and Shpments of anufacturng Industres. A Törnqvst ndex was appled to obtan gross output (value of shpments) deflators for each of the ndustres. 2) The Input/Output (I/O) tables from the Bureau of Labor Statstcs (BLS) 8 for are then used to calculate an ntermedate nput deflator for each ISIC ndustry. For each of the 190 I/O ndustres a gross output deflator seres was calculated. 9 These deflators are used to calculate an ntermedate nput prce ndex for each ndustry. For ndustry ths s done n the followng way: ln P = s, ln P, (1) In equaton (1) P, s the prce of the th ntermedate nput used n ndustry. The prce change for ths nput s weghted by the average share of nput n total ntermedate nputs n current prces of the ndustry over the two perods:, t t, t 1 t 1 s 1 P,, P,,, = +, t t, t 1 t 2 P P 1,,,,. (2) 7 Ths procedure s based on Schreyer (2000, 2002). 8 Specfcally from the Offce of Occupatonal Statstcs and Employment Proectons ( These tables are used snce they are avalable for each year n the sample and because the ndustry detal s much greater (190 versus 96 ndustres). 9 If there s a many-to-one correspondence, a Törnqvst ndex s used to aggregate to the level of aggregaton of the I/O table. 6

7 3) Usng the deflators for gross output and ntermedate nputs, gross output and ntermedate nputs at constant prces was calculated. These are combned to calculate real value added growth : ( ln Q s ln ) 1 ln = (3) s Here,, Q and are the quantty ndces of value added, gross output and ntermedate nputs respectvely. Furthermore: s t t, t P Q P Q = 1 s = (4) P, t t t Q s s the share of current value added n current gross output. The average over the two perods s taken and used n equaton (3). Fnally, the value added deflator s derved as the dfference between the growth rate of current and real value added. Whle ths procedure does not exactly replcate the BEA procedure, t serves as a good approxmaton snce the aggregate deflators are close to the orgnal value added deflator from the natonal accounts. 10 Aggregaton Some countres at present stll use fxed-weght (Laspeyres) ndces to calculate aggregate value added at constant prces. Ths can lead to serous substtuton bas f the structure of the economy s changng over tme. For example, when fxed weghts are used the prce declne for computers wll be overstated because of the relatvely large weght n the base year compared to successve years (Landefeld and Grmm, 2000). To correct for ths problem, chan-weghted ndces lke chaned Laspeyres or chaned Fsher, or Törnqvst ndexes are needed. In (re)calculatng real value added aggregates for ndustry groups and for the aggregate economy chan-weghted (Törnqvst) deflators for value added were used. For ndustry ths s done n the followng way. Let P the deflator for value added n ndustry and aggregate deflator, then the change of the deflator n perod t s gven by: P the 10 Dfferences occur for a number of reasons: the BEA uses the detaled source materal of the I/O tables, whch s more dsaggregated than the 190-ndustry table used here. Furthermore a sngle deflator was used for all nputs from a certan commodty category, whle the BEA dstngushes between domestcally produced and mported goods. Also, the prce deflators used here correspond to the value of shpments of an ndustry not of the commodtes that are used as nputs. Fnally, the BEA uses Fsher aggregaton nstead of Törnqvst. 7

8 , t, t, t ln P = s ln P (5) The prce change n ndustry s weghted by the average share of ndustry n total value added over the two perods defned as: 1 = 2 P, t t, t 1 t 1 s, t +, t t, t 1 t P P 1 P. (6) Fnally, real aggregate value added growth s calculated as the growth rate of aggregate current value added 11 mnus the growth rate of the deflator as follows: t t, t, t ln = ln P ln P (7) It should also be stressed that the use of Törnqvst aggregaton dverts from smply addng up value added estmates at constant prces from the lowest ndustry level to hgher aggregates. In partcular when underlyng ndustres show strongly dfferent growth rates (such as ICTproducng ndustres relatve to non-ict ndustres), a summed result may devate from a Törnqvst weghted result. Hence ntermedate aggregates, such as a seres for say total manufacturng, can only be obtaned by agan applyng the Törnqvst aggregaton procedure. European Unon Aggregaton In order to arrve at totals for the European Unon, current prce measures n natonal currences are converted nto euro s usng sector specfc purchasng power partes (PPPs). 12 The PPPs for all countres n ths database stem from the ultlateral 1997 Industry-of-orgn PPP set, whch has been descrbed n more detal n Tmmer, Ypma and van Ark (2006, forthcomng) 13. For agrculture (see van Ark, Ypma and Rao, 2006, forthcomng), manufacturng (O ahony and van Ark, 2003, chapter 7), trade (Tmmer and Ypma, 2006, forthcomng) and transport and communcaton (Ypma, Tmmer and van Ark, 2006, forthcomng) dfferent methods for calculatng PPPs have been used. These publcatons provde extensve analyss of the development of the Unt alue Ratos. The ndustry PPPs were all converted to a 1997 bass, and then appled to obtan value added shares by ndustry 11 Aggregate current value added s the sum of ndustry current value added. 12 See van Ark and Tmmer (2003) for a dscusson of approaches to obtan sector-specfc PPPs. See also the dscusson on the Productvty and Unt Labour Cost Database n O'ahony and B. van Ark (2003), chapter 7. 8

9 and by country n the total EU aggregate for the whole perod These were then used to Törnqvst aggregate the deflators for ndvdual ndustres to an EU aggregate. German unfcaton In order to deal wth German unfcaton n 1990 two data sets have been constructed: one for Western Germany (coverng the perod ) and one for unfed Germany (coverng the perod ). Growth rates for Western Germany are lnked to 1991 data for unfed Germany. US reclassfcatons As the US Standard Industral Classfcaton (SIC) dffers consderably from the ISIC rev 3 (and NACE Rev.1) classfcaton, several maor adustments were made to the US Natonal Income and Product Accounts at detaled ndustry levels. The man reclassfcatons were carred out for motor vehcle trade and repars (ISIC 50), retal trade (ISIC 52), hotels and restaurants (ISIC 55), post and telecommuncatons (ISIC 64), all ndustres n busness servces (ISIC 71-74) and Other communty, socal and personal servces (ISIC 90-93). See the sources and methods descrpton n the US fle for more detals. Prce deflators for these ndustres have been developed n a smlar way as for ICT producng ndustres usng a double deflaton procedure (see above). Startng n 2002 the BEA swtched from SIC based data to NAICS based data. In the summer of 2004, BEA publshed ndustry estmates for the perod based on the NAICS classfcaton. 14 However, at present there s no straghtforward reconclaton possble between SIC and NAICS at the level of the 60-Industry database. For the tme beng we use the NAICS numbers as startng pont and lnk the trend of the SIC seres for the prevous perod n A more detaled descrpton of dfferences n the US data between the present and prevous verson of the 60 Industry Database s avalable from the Gronngen Growth and Development Centre upon request (ggdc@eco.rug.nl). 13 Tmmer,.P., G.Ypma and B. van Ark (2006), Industry-of-orgn PPPs for the Total Economy: A New Benchmark for 1997, GGDC Research emorandum GD-82 (forthcomng) 14 Per December 15, 2005, BEA released seres back to 1987, but these have not yet been ncorporated. 9

10 References Ark, B. van and.p. Tmmer (2003), easurng Productvty Levels. A Reader, OECD (draft). Ark, B. van, G. Ypma and D.S. Prasada Rao (2005), Agrcultural Purchasng Power Partes and Internatonal Comparsons of Agrcultural Output and Productvty, GGDC Research emorandum GD-84 (forthcomng), Unversty of Queensland/Gronngen Growth of Development Centre Eurostat Labour Force Survey, varous ssues, Luxembourg O ahony,. and B. van Ark (2003), EU Productvty and Compettveness: An Industry Perspectve Can Europe Resume the Catchng-up Process?, Offce for Offcal Publcatons of the European Communtes, Luxembourg. OECD (1996, 1997 & 2001), Servces Statstcs on alue Added and Employment, Pars OECD (2002), Purchasng Power Partes and Real Expendtures, Pars. OECD (2003), Structural Statstcs for Industry and Servces , Pars OECD (2005), OECD STAN Database 2005 Schreyer, P. (2000), The Contrbuton of Informaton and Communcaton Technology to Output Growth: A Study of the G7 Countres, STI Workng Papers 2000/2, OECD, Pars. Schreyer, P. (2002), Computer Prce Indces and Internatonal Growth and Productvty Comparsons, Revew of Income and Wealth, 48 (1), pp , arch. Tmmer,.P., G.Ypma and B. van Ark (2006), Industry-of-orgn PPPs for the Total Economy: A New Benchmark for 1997, GGDC Research emorandum GD-82 (forthcomng) Tmmer,.P., G.Ypma and B. van Ark (2006), Productvty Levels n Dstrbutve Trades: A New ICOP Dataset, GGDC Research emorandum GD-83 (forthcomng) Ypma, G.,.P. Tmmer and B. van Ark (2006), Productvty levels n Transport, Storage and Communcaton: A new ICOP Dataset, GGDC Research emorandum GD-85 (forthcomng) 10