EUROSTAT REVIEW ON NATIONAL ACCOUNTS AND MACROECONOMIC INDICATORS

Transcription

1 EUROSTAT REVIEW ON NATIONAL ACCOUNTS AND MACROECONOMIC INDICATORS 1/2017

2

3 EUROSTAT REVIEW ON NATIONAL ACCOUNTS AND MACROECONOMIC INDICATORS 1/2017

4 Prined by Imprimerie Cenrale in Luxembourg Manuscrip compleed in June 2017 Neiher he European Commission nor any person acing on behalf of he Commission is responsible for he use ha migh be made of he following informaion. Luxembourg: Publicaions Office of he European Union, 2017 European Union, 2017 Reuse is auhorised provided he source is acknowledged. The reuse policy of European Commission documens is regulaed by Decision 2011/833/ EU (OJ L 330, , p. 39). For any use or reproducion of phoos or oher maerial ha is no under he EU copyrigh, permission mus be sough direcly from he copyrigh holders. For more informaion, please consul: hp://ec.europa.eu/eurosa/abou/policies/ copyrigh The informaion and views se ou in his publicaion are hose of he auhors and do no necessarily reflec he official opinion of he European Union. Neiher he European Union insiuions and bodies nor any person acing on heir behalf may be held responsible for he use which may be made of he informaion conained herein. Prin ISSN Ca.: KS-GS EN-C PDF ISSN X Ca.: KS-GS EN-N

5 Conens Conens Ediorial 5 Oupu growh and inflaion across space and ime W. Erwin Diewer and Kevin J. Fox 7 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure Fris Bos 41 Processing scanner daa by an augmened GUV index Ludwig von Auer 73 Big daa ypes for macroeconomic nowcasing Dario Buono, Gian Luigi Mazzi, George Kapeanios, Massimiliano Marcellino and Fois Papailias 93 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 3

6 Aims and scope Aims and scope EURONA is an open access, peer-reviewed, scholarly journal dedicaed o Naional Accouns and Macroeconomic Indicaors. EURONA aims a providing a plaform for researchers, scholars, producers and users of macroeconomic saisics o exchange heir research findings, hereby faciliaing and promoing he advancemen of Naional Accouns and Macroeconomic Indicaors. EURONA publishes empirical and heoreical aricles wihin he scope of Naional Accouns and Macroeconomic Indicaors, as well as aricles on imporan policy uses of hese saisics. They may relae o boh users and producers ineress, presen subjecs of general relevance or invesigae specific opics. EURONA is non-parisan and applies he highes sandards o is conen, by emphasising research inegriy, high ehical sandards, validiy of he findings and cuing edge resuls. EURONA gives room o all viewpoins. The aricles published in EURONA do no necessarily reflec he views or policies of he European Commission. Websie: hp://ec.europa.eu/eurosa/web/naional-accouns/publicaions/eurona Conac: ESTAT-EURONA@ec.europa.eu Ediors Paul Konijn ( 1 ), Eurosa Domenico Sarore ( 2 ), Ca Foscari Universiy, Venice Ediorial board Silke Sapel Weber, Eurosa Alber Braakmann, Saisisches Bundesam Gerard Eding, Cenraal Bureau voor de Saisiek Rosmundur Gudnason, Saisics Iceland Rober Inklaar, Universiy of Groningen, he Neherlands Sanjiv Mahajan, Office for Naional Saisics Gabriel Quiros, Inernaional Moneary Fund Philippe Sauffer, Federal Saisical Office Peer van de Ven, Organisaion for Economic Co operaion and Developmen ( 1 ) Paulus.Konijn@ec.europa.eu ( 2 ) sarore@unive.i 4 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

7 Ediorial Ediorial This firs EURONA issue of 2017 includes four imporan aricles ha ake us hrough hisory, presen and fuure of macro-economic saisics. Erwin Diewer and Kevin Fox kick-off his issue wih a horough discussion of mehods o aggregae real oupu and inflaion for a group of counries, as well as he long-sanding issue of creaing real GDP daa ha are consisen across ime and across counries. The resuls underline he imporance of regular cross-counry price comparisons, as he auhors, using acual OECD daa, show ha reliance on infrequen comparisons can creae significan uncerainy. Forunaely, he Inernaional Comparison Program, responsible for he inernaional price and volume comparisons, is now aiming for a fuure of regular and frequen comparisons. In he second aricle of his issue, Fris Bos akes he reader back in hisory in order o learn for he fuure. He describes how user needs have shaped naional accouns since he 17h cenury unil he presen. He shows ha fiscal and moneary policy needs have been imporan drivers of he developmen of naional accouns and heir respecive inernaional sandards, as well as he need o measure and compare (growh of) maerial living sandards. However, policy issues such as welfare, inequaliy and susainable developmen were less refleced in he sandards. The auhor ends wih a number of suggesions o ensure ha naional accouns remain relevan in he fuure. Many people believe ha he fuure belongs o big daa. In he hird aricle, Ludwig von Auer coninues he discussion on he use of mulilaeral mehods for processing scanner daa, one ype of big daa, for consumer price indices ha was sared by Anonio Chessa in EURONA 1/2016. He argues ha he mehod proposed by Chessa can be simplified in wo ways: by using a simpler index formula and a simpler aggregaion mehod. I appears ha he las word on mulilaeral mehods for consumer price indices has no ye been wrien. Regarding big daa in general, Dario Buono, Gian Luigi Mazzi, George Kapeanios, Massimiliano Marcellino and Fois Papailias presen, in he las paper of his issue, a caegorisaion of big daa sources and analyse heir poenial for macro-economic nowcasing, for example for employmen, GDP and inflaion. They also include a discussion of he pros and cons of big daa for nowcasing and noe ha he cons should no be ignored. I hink ha hese aricles ogeher give a lo of food for hough and, of course, for coninued discussion. In is role as a plaform for such discussions, EURONA welcomes any reacions o he aricles on ESTAT-EURONA@ec.europa.eu. Paul Konijn Edior of EURONA EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 5

8

9 1 Oupu growh and inflaion across space and ime W. ERWIN DIEWERT ( 1 ) AND KEVIN J. FOX ( 2 ) Absrac: I is common for comparisons o be made of oupu growh and inflaion across groups of counries, ye such comparisons can resul in inconsisencies. We address wo issues: (i) how o measure aggregae real oupu and inflaion for groups of counries and (ii) how o consruc measures of real GDP for a group of counries where he counry measures of real GDP are consisen across ime and space. A mehod is proposed for harmonizing conflicing esimaes of OECD member-counry real GDP, ensuring consisency over space and overall group consisency over ime. A new measure of OECD inflaion is also proposed. JEL codes: C43, C82, E01 Keywords: Purchasing power pariies, ICP, inflaion, price and volume indexes, Fisher indexes ( 1 ) Vancouver School of Economics, Universiy of Briish Columbia and UNSW Sydney, Ausralia. ( 2 ) School of Economics, UNSW Sydney, Ausralia.

10 1 Oupu growh and inflaion across space and ime Economericians have an ambivalen aiude owards economic daa. A one level, he daa are he world ha we wan o explain, he basic facs ha economiss purpor o elucidae. A he oher level, hey are he source of all our roubles. Zvi Griliches (1985, p. 196) 1. Inroducion Providing consisen esimaes of real GDP across counries and ime is imporan for many policy-relevan purposes, such as assessing convergence of living sandards; see Eurosa (2012) and he World Bank (2013). The OECD publishes esimaes of Purchasing Power Pariies (PPPs) on an annual basis and hese PPPs can be used o generae esimaes of real GDP for member counries ha are comparable across counries for he given year. However, he resuling esimaes of relaive real GDP are ypically inconsisen wih naional esimaes of real GDP growh for he member counries. We use OECD daa for he years in order o sudy wo issues. Firs, wha is he opimal approach o consrucing esimaes of OECD aggregae real GDP and associaed measures of aggregae OECD inflaion? Index number heory is used o decompose naional nominal GDP ino price and quaniy (or volume) componens, bu consrucing esimaes of inflaion and real GDP growh for a group of counries ha use differen currencies is a more complicaed operaion ( 3 ). Second, how can PPP informaion be opimally used in conjuncion wih counry daa on real GDP growh o consruc esimaes of OECD member counry real GDP ha are in principle simulaneously comparable across space and ime? Using our proposed soluion as a benchmark, we show ha if PPP daa are available only infrequenly, as is he case for he World Bank provided PPP daa used in he Penn World Tables, hen esimaes can differ considerably as new PPP informaion becomes available. This is raher inconvenien: sudies of compeiiveness and living sandards convergence across counries will wan o use real GDP series ha are no subjec o violen revision. In Secion 2, we sudy he firs measuremen issue using jus naional daa and exchange rae informaion. Secions 3-6 use he OECD PPP daa and sudy he second measuremen issue (and revisi he firs issue). In Secion 3, we propose a harmonized mehod for consrucing esimaes of member counry GDP volumes ha are comparable across ime and space. In Secions 4 and 5, we compare our harmonized esimaes wih esimaes of member counry real GDP ha are generaed by using PPP daa for only one year. These base year esimaes are hen projeced o all oher years using each counry s own growh raes of GDP. Secion 4 uses he PPPs for 2000 and Secion 5 uses he PPPs for We find ha he resuling wo panel ses of real GDP esimaes are very differen from each oher and hey also are very differen from our harmonized esimaes developed in Secion 3. Secion 6 considers he conex where PPPs ( 3 ) A he naional level, curren price (value) daa can ypically be decomposed ino a volume (or quaniy) series and price series. A he inernaional level, a second price componen eners he picure in he form of a conversion rae from he domesic o a common currency. The implicaion is ha values can be expressed a curren marke exchange raes (or curren inernaional prices, if purchasing power pariies PPPs are used); and a consan exchange raes (or consan inernaional prices). OECD (2001, p. 6). 8 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

11 Oupu growh and inflaion across space and ime 1 are only available infrequenly, as is he case wih World Bank provided PPPs ( 4 ). We use PPP informaion for 2000 and 2012 o provide inerpolaed esimaes of volumes for each counry, finding ha he curren inerpolaion mehod implemened in he widely used Penn World Tables did no work well wih our OECD daa base. In conras, we find ha our proposed mehod produces esimaes ha are much closer o our preferred harmonized esimaes of Secion 3. Secion 7 concludes his paper. 2. OECD growh and inflaion using marke exchange raes Our firs measure of aggregae GDP growh over he member counries in he OECD during he years uses naional growh raes of GDP and domesic prices convered ino US dollar a marke exchange raes. The aggregaion principle used o form OECD aggregae GDP volumes and prices in his secion is he same ha is used o aggregae prices and quaniies across differen regions in a counry: each commodiy in each region is regarded as a separae commodiy in he index number formula. In wha follows, we use he OECD ordering of counries, which is as in Table 1. Table 1: OECD counry codes 1 = Ausralia 10 = France 19 = Korea 28 = Slovenia 2 = Ausria 11 = Germany 20 = Luxembourg 29 = Spain 3 = Belgium 12 = Greece 21 = Mexico 30 = Sweden 4 = Canada 13 = Hungary 22 = Neherlands 31 = Swizerland 5 = Chile 14 = Iceland 23 = New Zealand 32 = Turkey 6 = Czech Republic 15 = Ireland 24 = Norway 33 =U.K. 7 = Denmark 16 = Israel 25 = Poland 34 = Unied Saes 8 = Esonia 17 = Ialy 26 = Porugal 9 = Finland 18 = Japan 27 =Slovak Republic The counry values for nominal GDP in he naional currencies for he years can be obained from he OECD online daa base, OECD.Sa ( 5 ). We conver hese esimaes ino billions and denoe he esimae for counry n in year by V n. The corresponding volume esimaes can be obained from he same source ( 6 ), and we similarly conver hese esimaes ino billions and denoe hese volumes (or quaniies) by Q n for n = 1,,34 and = 2000,,2012. The corresponding counry price level for counry n in year is defined as P n V n /Q n for n = 1,,34 and = 2000,,2012. These naional price levels and volumes are lised in he Annex; see Tables A1 and A2. Since he counry volumes Q n are measured in domesic currency unis (which are no comparable across counries), we need o conver he annual domesic nominal values of GDP ( 4 ) For example, he Inernaional Comparison Program (ICP) of he World Bank provided PPPs for 155 counries for he year 2005 and has more recenly provided a new se of PPPs for 2011 covering 177 counries. The ICP mehodology is explained in World Bank (2013). How can hese wo benchmark ses of PPPs be used in conjuncion wih naional daa in order o provide esimaes of counry real GDP ha are comparable across all years from 2005 o 2012? The inerpolaion mehod explained in Secion 6 could be used in his conex. ( 5 ) OECD.Sa Table B1_GE: Gross domesic produc (expendiure approach); Naional currency, curren prices, millions, annual daa. ( 6 ) OECD.Sa Table B1_GE: Gross domesic produc (GDP); Naional currency, consan prices, naional base year, millions, annual daa. EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 9

12 1 Oupu growh and inflaion across space and ime ino common currency unis using he average exchange raes for he corresponding year. In principle, he numeraire counry could be any of he 34 OECD counries bu i seems reasonable o choose he larges counry. The OECD has convenienly done his for us, convering each counry s nominal GDP ino US dollar a he average marke exchange raes for he given year ( 7 ). We conver hese esimaes ino billions and denoe he US dollar esimae for nominal (curren price) GDP for counry n in year by v n. The year, counry n US dollar price level for GDP, p n, is iniially defined as v n /Q n where he counry volumes (or real oupus) Q n have already been defined using naional daa. The resuling p n were normalized so ha p n 2000 = 1 for n = 1,,34. The Q n were hen normalized in he opposie direcion so ha US dollar values were preserved. Denoe he resuling normalized Q n as q n for n = 1,,34 and = 2000,,2012 ( 8 ). These US dollar price levels p n and he corresponding volumes q n are lised in Tables A3 and A4 in he Annex. We are now in a posiion o calculae annual aggregae OECD real oupu and he corresponding price level for each of he years using he price and volume daa, p n and q n as inpus ino an index number formula. I will be useful o consider he choice of index number formula in he conex of providing index levels for wo periods, say periods 0 and 1 ( 9 ). Suppose here are n commodiies o be aggregaed. Denoe he price and quaniy vecors for period by p [p 1,, p N ] and q [q 1,,q N ] for = 0,1. The value of ransacions in he N commodiies during period is defined as v n =1 N p n q n p q ( 10 ). The problem of choosing funcional forms for he price and quaniy indexes is usually phrased as follows: find wo suiable funcions of 4N variables (i.e. prices and quaniies for each good in each period), a price index funcion P(p 0, p 1,q 0,q 1 ) and a quaniy index funcion Q(p 0, p 1,q 0,q 1 ), such ha he produc of hese wo funcions is equal o he value raio, v 1 /v 0. Thus he funcions P and Q are o saisfy he following equaion: (1) p 1 q 1 /p 0 q 0 = P(p 0, p 1,q 0,q 1 )Q(p 0, p 1,q 0,q 1 ). I can be seen ha if he funcional form for eiher he price or quaniy index is deermined hen he funcional form of he corresponding quaniy or price index is also deermined using equaion (1) ( 11 ). Two naural choices for he funcional form for he price index are he well-known Laspeyres and Paasche price indexes, P L and P P, defined as follows ( 12 ): (2) P L (p 0, p 1,q 0,q 1 ) p 1 q 0 /p 0 q 0 ; (3) P P (p 0, p 1,q 0,q 1 ) p 1 q 1 /p 0 q 1. Using (1), i can be seen ha quaniy indexes ha correspond o P L and P P are Q P and Q L, respecively, defined as follows: ( 7 ) OECD.Sa Table B1_GE: Gross domesic produc (expendiure approach); US dollar, curren prices, curren exchange raes, millions, annual daa. ( 8 ) Noe ha v n = p n q n for n = 1,,34 and = 2000,,2012. For each n, he US dollar volumes q n are proporional o he naional volumes Q n ; i.e., we have q n = λ n Q n for = 2000,,2012 for each counry n where λ n is he facor of proporionaliy for counry n. ( 9 ) For maerials on he hisorical developmen of index number heory, see Diewer (1993) and Balk (2008). ( 10 ) The inner produc of wo vecors x [x 1,,x N ] and y [y 1,,y N ] of he same dimension N is defined as x y n =1 N x n y n. ( 11 ) Once P and Q saisfying (1) have been chosen, he corresponding price levels for periods 0 and 1, say P 0 and P 1, and he corresponding quaniy (or volume) levels for periods 0 and 1, say Q 0 and Q 1, are generally deermined as follows: P 0 1; P 1 P(p 0, p 1,q 0,q 1 ); Q 0 v 0 = p 0 q 0 and Q 1 v 0 Q(p 0, p 1,q 0,q 1 ) = v 1 / P(p 0, p 1,q 0,q 1 ). Noe ha he price and quaniy indexes can be inerpreed as raios of aggregae price and quaniy levels; i.e., we have P(p 0, p 1,q 0,q 1 ) = P 1 /P 0 and Q(p 0, p 1,q 0,q 1 ) = Q 1 /Q 0. ( 12 ) I can be seen ha he Laspeyres price index uses he baske of period 0 quaniies, q 0, and prices ou his baske a he prices of period 0 (in he denominaor) and prices ou he same baske a he prices of period 1 (in he numeraor) and akes he raio of hese coss as he price index. The Paasche index is similar bu uses he baske of period 1 quaniies, q 1, as he common quaniy vecor in he numeraor and denominaor. 10 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

13 Oupu growh and inflaion across space and ime 1 (4) Q P (p 0, p 1,q 0,q 1 ) p 1 q 1 /p 1 q 0 ; (5) Q L (p 0, p 1,q 0,q 1 ) p 0 q 1 /p 0 q 0. The Paasche and Laspeyres price and quaniy indexes are equally plausible. The problem is ha hey can generae quie differen esimaes of inflaion and growh. A naural soluion o his problem is o ake a symmeric average of hese wo esimaes; aking he geomeric mean of hese wo price indexes (and of he wo corresponding quaniy indexes) leads o indexes ha have desireable axiomaic properies ( 13 ). These are Fisher (1922) ideal price and quaniy indexes, P F and Q F, defined as follows ( 14 ): (6) P F (p 0, p 1,q 0,q 1 ) [P L (p 0, p 1,q 0,q 1 )P P (p 0, p 1,q 0,q 1 )] 1/2 ; (7) Q F (p 0, p 1,q 0,q 1 ) [Q L (p 0, p 1,q 0,q 1 )Q P (p 0, p 1,q 0,q 1 )] 1/2. There is one more choice ha needs some discussion: namely, should fixed base or chained Fisher indexes be used when aggregaing over many periods? The chain sysem measures he change in prices going from one period o a subsequen period using a bilaeral index number formula involving he prices and quaniies peraining o he wo adjacen periods ( 15 ). These one period raes of change (he links in he chain) are hen cumulaed o yield he relaive levels of prices over he enire period under consideraion. If he bilaeral price index is P, he chain sysem generaes he following sequence of price levels for he firs hree periods: (8) 1, P(p 0, p 1,q 0,q 1 ), P(p 0, p 1,q 0,q 1 ) P(p 1, p 2,q 1,q 2 ). The fixed base sysem of price levels using he same bilaeral index number formula P simply compues he level of prices in period relaive o he base period 0 as P(p 0, p,q 0,q ). The fixed base sequence of price levels for periods 0, 1 and 2 is: (9) 1, P(p 0, p 1,q 0,q 1 ), P(p 0, p 2,q 0,q 2 ). There are wo major problems associaed wih he use of fixed base indexes in he conex of annual ime series daa: (i) over longer periods of ime, i becomes more difficul o mach up producs in he curren period wih he corresponding producs in a disan base period, leading o less accurae index numbers; and (ii) fixed base indexes are subjec o revisions (ha can be subsanial) when he base period is finally changed. When using fixed base Paasche or Laspeyres indexes, he revision problem can become massive ( 16 ). Thus a major advanage of he chain sysem in he conex of aggregaing annual daa is ha chaining will reduce he spread beween he Paasche and Laspeyres indexes ( 17 ). These wo indexes each provide an asymmeric perspecive on he amoun of price change ha has occurred beween he wo periods under consideraion and i could be expeced ha a single poin esimae of he ( 13 ) See Fisher (1922) and Diewer (1992) (1997). ( 14 ) I can be verified ha P F Q F = v 1 /v 0 ; i.e., he Fisher price and quaniy indexes saisfy equaion (1). ( 15 ) The chain principle was inroduced independenly ino he economics lieraure by Lehr (1885, pp ) and Marshall (1887, p. 373). Boh auhors observed ha he chain sysem would miigae he difficulies due o he inroducion of new commodiies ino he economy, a poin also menioned by T.P. Hill (1993, p. 388). Fisher (1911, p. 203) inroduced he erm chain sysem. ( 16 ) The US Bureau of Economic Analysis used o provide long erm esimaes of US GDP back o 1926 using fixed base Laspeyres volume indexes. When he base year was changed, he resuling Laspeyres esimaes of real GDP growh changed massively, promping he BEA o swich o chained Fisher indexes in he early 1990s. ( 17 ) See Diewer (1978, p. 895) and T.P. Hill (1988) (1993, pp ). EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 11

14 1 Oupu growh and inflaion across space and ime aggregae price change should lie beween hese wo esimaes. Chaining will usually lead o a smaller difference beween he wo and hence lead o esimaes ha are closer o he ruh ( 18 ). For year = 2001,,2012, denoe he chained Fisher aggregae OECD volume level for by Q and he corresponding US dollar price level by P, and define he OECD volume growh rae γ and he corresponding OECD US dollar inflaion rae ρ in percenage poins as follows: (10) γ 100[(Q /Q 1 ) 1] ; (11) ρ 100[(P /P 1 ) 1]. The chained Fisher OECD aggregae price and volume levels, P and Q, for he years are lised in Table 2 along wih he corresponding percenage poin annual growh raes, ρ and γ, for he years For comparison purposes, we also provide he aggregae OECD chained Laspeyres and Paasche indexes, P L and P P, over he same period. I can be seen ha he chained Fisher, Laspeyres and Paasche price levels are all very close o each oher so ha for his paricular applicaion, he choice of index number formula does no maer very much. Table 2: OECD annual aggregae volumes Q and price levels in US dollar P, P L and P P, price levels in euro P EU, PPP price levels P PPP and percenage poin changes, Year Q P P L P P γ ρ ρ EU P EU P PPP Source: Auhor's calculaions The sample average of he year o year growh raes for OECD real GDP using US dollar weighs, he γ, was 1.58 % per year. I can be seen ha here was only one year where OECD real growh was negaive: 2009 ( 3.64 %). The sample average of he OECD inflaion raes ρ (measured in US dollar a marke exchange raes) was 3.24 % per year. However, wha is sriking is he variabiliy of he US dollar inflaion raes. ( 18 ) There is a more elaborae jusificaion for chaining annual daa ha is based on aggregaing over observaions ha have he mos similar price srucures; see R.J. Hill (2001), (2004), (2009), Diewer (2009) and Diewer and Fox (2017). Typically, adjacen annual observaions will have more similar price srucures han a pair of observaions chosen from wo disan periods. However, i is no always bes o use chained indexes. T.P. Hill (1993, p. 388), drawing on he earlier research of Szulc (1983) and T.P. Hill (1988, pp ), noed ha i is no appropriae o use he chain sysem when prices oscillae or bounce o use Szulc s (1983, p. 548) erminology. This bouncing phenomenon can occur when aggregaing subannual daa when here are seasonal flucuaions or periodic sales (wih deeply discouned prices). However, in he conex of more or less smoohly rending prices and quaniies as is he usual case using annual daa, T.P. Hill (1993, p. 389) recommended he use of chained symmerically weighed indexes such as he Fisher ideal index. Thus in his paper, we will use chained Fisher indexes when aggregaing over counries. 12 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

15 Oupu growh and inflaion across space and ime 1 The principles used o consruc our OECD aggregae measures of real GDP, Q, are he same principles ypically used o consruc counry wide esimaes of real GDP wihin a counry. Esimaes of real GDP aggregae oupu growh over regions wihin he counry use regional price levels as weighs for he regional volumes. In consrucing naional esimaes of real oupu, he naional saisician does no assume ha he quaniies (or volumes) in each region are comparable across regions; all ha is assumed is ha whaever is being measured a he regional level is comparable over ime. This is he same principle ha is being used o consruc he above OECD real oupu measures Q ; here is no assumpion ha he counry unis of measuremen are comparable across counries. The one difference in our suggesed mehod for consrucing OECD real GDP as opposed o mehods used o consruc naional esimaes of real GDP is ha we needed o conver naional values of GDP ino a common currency using annual average marke exchange raes. We chose o make his conversion using US dollar as he numeraire currency. If we chose anoher currency o be he numeraire currency, he uni of measuremen would change, bu he overall OECD growh raes for real GDP would remain he same; i.e., he γ lised in Table 2 do no change if we convered all counry nominal GDP esimaes ino a differen common currency a annual average marke exchange raes and hen applied he same mehodology o consruc he overall OECD volume esimaes ( 19 ). However swiching o a differen numeraire currency dramaically affecs he inflaion raes ρ ; he OECD aggregae price level esimaes P and he resuling inflaion raes ρ defined by (11) change depending on he chosen numeraire currency. In order o illusrae he dependence of he above OECD GDP inflaion raes on he choice of he numeraire counry, we compued he aggregae OECD price and volume levels, P EU and Q EU, using Germany as he numeraire counry. Thus insead of using he US dollar esimaes for nominal GDP for counry n in year defined earlier by v n, for euro zone counries we used he repored naional value esimaes of GDP. For non-euro zone counries, we convered he v n ino euro using he implied OECD exchange rae ha can be obained by dividing he naional value esimae of GDP for Germany (or any oher euro zone counry) by he corresponding US dollar measure. The same Fisher index number mehodology was hen used o consruc P EU and Q EU. The resuling euro based price index P EU and inflaion growh raes ρ EU 100[(P EU / P EU 1 ) 1] are lised in Table 2. Comparing he inflaion measures using he US and hen Germany as he numeraire counries shows ha he resuling price levels, P and P EU, and inflaion raes, ρ and ρ EU, are compleely differen. The price levels are ploed in Figure 1 o clearly show he exen of heir divergence; P rends upward from 1.00 in 2000 o end up a 1.45 in 2012 whereas he euro based OECD price level rends downward o 0.89 in 2008 and hen rend upward o end up a 1.04 in The explanaion for hese diverging resuls is simple: hey are driven by large exchange rae movemens over he sample period ( 20 ). Our conclusion a his poin is ha our firs approach o measuring OECD real oupu and inflaion using naional GDP daa and marke exchange raes is (perhaps) saisfacory for measuring real oupu bu ha i is no saisfacory for measuring inflaion. A saisfacory inflaion measure will be inroduced in he following secion when we inroduce our second approach o measuring aggregae OECD inflaion. The preferred price levels from his approach (P PPP ) are also lised in Table 2 and ploed in Figure 1 for ease of comparison wih he corresponding exchange rae based series ( 21 ). ( 19 ) In order for his saemen o be rue, we need our chosen bilaeral index number formula o saisfy he following wo ess: Q(λp 0, p 1,q 0,q 1 ) = Q(p 0, p 1,q 0,q 1 ) for all scalars λ > 0 and Q(p 0,λp 1,q 0,q 1 ) = Q(p 0, p 1,q 0,q =) for all scalars λ > 0. The Fisher, Laspeyres and Paasche bilaeral quaniy indexes all saisfy hese homogeneiy-in-prices properies. ( 20 ) US prices in erms of euro declined markedly from 2000 o 2008 and his explains he large number of negaive ρ EU over his period; he number of euro i ook o buy one US dollar in 2000, 2008 and 2012 was 1.085, and 0.778, respecively. ( 21 ) These will be discussed in he following secion, where he corresponding inflaion rae series are ploed in Figure 2. EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 13

16 1 Oupu growh and inflaion across space and ime Figure 1: Alernaive OECD price levels in US dollar, euro and PPPs (Index) USD: P PPPs : P PPP EUR : P EU Source: Auhor's calculaions The analysis presened in his secion made no assumpion ha he goods and services produced in any counry were comparable o he goods and services produced in any oher counry. In he following secion, i will be assumed ha he goods and services produced in each counry are comparable across counries and differen measures of OECD growh and inflaion will be derived. 3. OECD growh and inflaion measuremen using annual PPP informaion The OECD (in close cooperaion wih Eurosa) produces an annual series of Purchasing Power Pariies (PPPs) ha enable he comparison of real GDP of member counries wih each oher ( 22 ). For each OECD counry n and each year, PPP n is an esimae of he number of unis of he naional currency of counry n ha is required o purchase one dollar of US GDP in year ( 23 ). We divide he counry n nominal value of GDP in year in domesic currency, V n, by he corresponding PPP n in order o obain an esimae, r n, of counry n s real GDP in year in unis ha are comparable across counries for year ( 24 ): (12) r n V n /PPP n ; n = 1,,34; = 2000,,2012. ( 22 ) The consrucion of hese PPPs is explained in he Eurosa and OECD PPP Manual; see Eurosa (2012). The Inernaional Comparison Program (ICP) of he World Bank consruced PPPs for over 150 counries for 2005 and The ICP mehodology is explained in World Bank (2013). ( 23 ) OECD.Sa, Table 4: PPPs and Exchange Raes; PPPGDP; Purchasing Power Pariies for GDP; Naional currency per US dollar; Annual; This Table is reproduced in he Annex as Table A5. ( 24 ) These relaive GDP volume measures for year are no comparable across years. 14 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

17 Oupu growh and inflaion across space and ime 1 Once he r n have been calculaed, hey can be summed so ha r n =1 34 r n and hen he year counry n share of OECD real oupu can be defined as follows ( 25 ): (13) s n r n /r ; n = 1,,34; = 2000,,2012. These counry shares of OECD real GDP are lised in Table 3, which enables he comparison of GDP volumes across all OECD counries for each year ( 26 ). Noe ha counry 34, he US, has he larges share (around %), followed by counry 18, Japan, (10-11 %) and counry 11, Germany (7 %). We can use his informaion o consruc esimaes of overall real GDP growh and inflaion across OECD counries. A naural mehod is o use he counry shares in Table 3 as weighs for naional raes of growh of real GDP. The year growh facor for counry n can be defined as Q n / Q n 1 where Q n is counry n s GDP volume in year, and he OECD Laspeyres ype growh facor (or chain link) for year, Γ L, as he following weighed average of he naional growh facors: (14) Γ L n =1 34 s n 1 (Q n /Q n 1 ) ; = 2001,,2012. The measure of OECD GDP volume growh defined by (14) is he mehod used by he OECD o calculae heir official measure of OECD volume growh. I cerainly is a sensible measure, using counry (one plus) growh raes going from year 1 o year, Q n /Q n 1, weighed by he counry real volume shares s n 1 for year 1, which were derived using PPPs. However, he above formula suffers from he same problem ha he sandard Laspeyres formula has: namely, i does no rea he periods in a symmeric fashion. The counerpar o he Laspeyres-ype formula defined by (14) is he following Paasche-ype formula ( 27 ): (15) Γ P [ n =1 34 s n (Q n /Q n 1 ) 1] 1 ; = 2001,,2012. The corresponding Fisher-ype formula for OECD volume growh for year is defined as follows ( 28 ): (16) Γ F [Γ L Γ P ] 1/2 ; = 2001,,2012. ( 25 ) Noe ha he counry shares s n can be consruced wihou using counry exchange raes (in principle). Using definiions (12) and (13), i can be seen ha he s n can be wrien in he following form: s n = [V n /PPP n ]/[ i =1 N (V i /PPP i )] for all n and. Compare hese real shares s n wih he corresponding counry US dollar shares S n = [V n /e n ]/[ i =1 N (V i /e i )] defined in he Annex. All of he measures derived in his secion are independen of counry exchange raes. ( 26 ) Row n+1 in he Table gives he shares for counry n where we use he sandard ordering of OECD counries lised in he previous secion. Since he PPPs used by he OECD are invarian o he choice of he numeraire counry (up o a scalar facor), i can be verified ha he counry shares lised in Table 3 are also invarian o he choice of numeraire counry. ( 27 ) Suppose ha here is only one homogeneous commodiy in each counry s GDP. Then he volume for counry n in year, Q n, should be equal o he number of unis of his homogeneous commodiy. Under hese condiions, i can be seen ha boh Γ L and Γ P equal n =1 34 Q n / n =1 34 Q n 1. ( 28 ) If he PPPs are independen of he choice of he numeraire counry (up o a scalar facor), hen he growh facors, Γ L Γ P and Γ F will no depend on he choice of he numeraire counry. EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 15

18 1 Oupu growh and inflaion across space and ime The growh facors (or chain link indexes) defined by (14)-(16) can be muliplied ogeher o generae OECD aggregae volume levels. The growh facors can also be ransformed ino growh raes, γ L, γ P and γ F (in percenage poins), by using he following definiions for = 2001,,2012: (17) γ L 100[Γ L 1] ; γ P 100[Γ P 1] ; γ F 100[Γ F 1]. Table 3: Counry shares of OECD real GDP n s n 2000 s n 2001 s n 2002 s n 2003 s n 2004 s n 2005 s n 2006 s n 2007 s n 2008 s n 2009 s n 2010 s n 2011 s n Noe: n denoes he counry code, given in Table 1. Source: Auhor's calculaions The annual OECD volume growh measures defined by (17) as well as our earlier US dollar weighed measures γ are lised in Table 4. I can be seen ha he Laspeyres, Paasche and Fisher measures of OECD growh explained in his secion are virually idenical. Therefore, moving 16 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

19 Oupu growh and inflaion across space and ime 1 from he OECD Laspeyres-ype measure of overall volume growh o he Fisher measure did no make much difference for his daa se ( 29 ). I can also be seen ha our preferred Fisher measure of OECD growh in comparable unis across counries, γ F, grew on average abou 1/10 of a percenage poin more rapidly per year han our preferred measure of OECD GDP growh using US dollar weighs, γ. Alhough his is no a large difference in growh raes, i is significan and so users need o decide which measure, γ F or γ, bes suis heir needs. The measure γ can be defined using jus naional informaion on domesic price and quaniy (or volume) indexes and exchange raes while he measure γ F requires informaion on domesic values, domesic volume indexes and PPPs. PPPs are no likely o be nearly as accurae as naional measures of price and volume change due o he difficulies in maching producs across counries. There are addiional difficulies wih he reamen of inernaional rade in he consrucion of PPPs. The γ measure has he problem ha large flucuaions in exchange raes can lead o flucuaions in he γ while he PPP based γ F measures are heoreically independen from exchange rae movemens ( 30 ). Thus one has o weigh he disadvanage of possibly less reliable PPPs agains he advanage of having aggregae growh measures ha are independen from exchange rae movemens ( 31 ). The OECD real oupu shares, s n defined by (13), can also be used as weighs for naional GDP inflaion raes. We define he OECD Laspeyres, Paasche and Fisher PPP based chain link price indexes, Π L, Π P and Π F for = 2001,,2012, as follows ( 32 ): (18) Π L n =1 34 s n 1 (P n /P n 1 ) ; Π P [ n =1 34 s n (P n /P n 1 ) 1 ] 1 ; Π F [Π L Π P ] 1/2. These chain link indexes can be muliplied ogeher o generae he corresponding OECD aggregae price levels. The resuling Fisher OECD price level index for year is denoed by P PPP and i is lised in he las column of Table 2 ( 33 ). The inflaion growh facors can also be ransformed ino growh raes, ρ L, ρ P and ρ F in percenage poins, by using he following definiions for = 2001,,2012: (19) ρ L 100[Π L 1] ; ρ P 100[Π P 1] ; ρ F 100[Π F 1]. These PPP based inflaion raes (in percenage poins) are lised in he las 3 columns of Table 4. The sample averages of he ρ L, ρ P and ρ F are 2.17, 2.10 and 2.14 percenage poins. Viewing Table 4, i can be seen ha here are some significan differences beween he hree measures of OECD inflaion ha are PPP based ( 34 ). Comparing he inflaion numbers in Tables 2 and 4, ( 29 ) Recall ha he official OECD measure of real GDP growh is he Laspeyres measure, γ L. Our esimaes differ slighly from he official measures due o rounding. The exchange-rae-weighed growh raes γ should be somewha lower han he PPP-weighed growh raes γ F due o he Balassa-Samuelson effec and his expecaion is realized for he OECD daa. We would expec he divergence o grow as less rich counries are added o he lis of counries. ( 30 ) Exchange rae movemens do no direcly affec he domesic raes of growh (he Q n /Q n 1 ) bu as we have seen, hey do affec he weighs used o aggregae he counry real growh raes ino he overall OECD Laspeyres, Paasche and Fisher growh raes. Exchange rae flucuaions are large enough o maerially affec he weighs, which in urn lead o maerial flucuaions in he overall OECD volume growh raes. ( 31 ) I will generally be he case ha he s n will be greaer han he corresponding S n for counries n ha are relaively poor and hus he index of OECD aggregae real GDP growh defined in Secion 2 will end o be a more pluocraic index (since rich counries ge larger share weighs in his index) compared o he more democraic index of OECD aggregae real GDP growh defined in Secion 3. Thus one could choose beween he wo indexes based on one s preferences over weighs. We owe his poin o Marshall Reinsdorf. ( 32 ) The official OECD measure of household inflaion over member counries is he Laspeyres measure defined in (18) where household consumpion replaces GDP; see he OECD (2014). ( 33 ) This price index saisfies he ime reversal es whereas is Laspeyres and Paasche counerpars do no saisfy his imporan es. Hence he Fisher PPP based inflaion index P PPP is our preferred measure of OECD aggregae inflaion. ( 34 ) In view of hese differences in he hree indexes of OECD GDP inflaion, i may be preferable for he OECD o replace heir Laspeyres ype indexes of OECD household inflaion by heir Fisher counerpars. EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 17

20 1 Oupu growh and inflaion across space and ime as ploed in Figure 2, i can be seen ha he PPP based esimaes of OECD inflaion are much more reasonable han he esimaes ha were based on counry exchange raes ha were lised in Table 2, he ρ and ρ EU. Our conclusion is ha he OECD Fisher price index P PPP is a much beer measure of OECD inflaion han he indexes ha used exchange raes insead of PPPs. Table 4: Annual percenage poin changes in OECD PPP based Laspeyres, Paasche and Fisher volume measures γ L, γ P and γ F, US dollar weighed volume measures γ and Laspeyres, Paasche and Fisher PPP based inflaion measures, ρ L, ρ P and ρ F : Year γ L γ P γ F γ ρ L ρ P ρ F Average Source: Auhor's calculaions Figure 2: Alernaive OECD inflaion raes using US dollar, euro and PPPs (Percenage poin changes) EUR : ρ EU PPPs : ρ L PPPs : ρ p PPPs : ρ F USD: ρ Source: Auhor's calculaions 18 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

21 Oupu growh and inflaion across space and ime 1 Now we come o he mos difficul problem: how can we use PPP informaion and naional growh raes o obain esimaes of member counry GDP volumes ha are comparable across ime and space? The Eurosa (2012) Manual offers he following advice: To race he evoluion of relaive GDP volume levels beween counries over ime, i is necessary o selec one of he reference years as a base year and o exrapolae is relaive GDP volume levels over he oher years. Exrapolaion is done by applying he relaive raes of GDP volume growh observed in he differen counries. This provides a ime series of volume indices a a consan uniform price level ha replicaes exacly he relaive movemens of GDP volume growh of each counry. Eurosa (2012, p. 18). We implemen his sraegy in Secions 4 and 5 below, where we choose he relaive counry GDP volumes given by he counry shares of OECD aggregae GDP for 2000 (Secion 4) and for 2012 (Secion 5) and we use naional growh raes for counry GDP volumes o exrapolae hese base shares o all ime periods. However, i will be seen ha he resuling comparable counry volumes over ime and space differ considerably, depending upon which base year is chosen. This is raher inconvenien: sudies of compeiiveness of OECD counries and living sandards convergence across counries will wan o use counry volume series ha are no subjec o violen revision ( 35 ). Our suggesed soluion o he problem of harmonizing naional growh raes of GDP wih he counry shares of OECD aggregae real GDP ress on wo principles. Firs, he resuling harmonized esimaes of counry volumes mus be consisen wih he real annual cross counry volume shares s n lised in Table 3 above. Second, OECD aggregae real GDP growh mus be equal o he raes of aggregae growh generaed by our recommended Fisher indexes Γ F defined by (16). Using hese principles, he counry GDP volumes are uniquely deermined (up o a scalar unisof-measuremen facor). To see his, firs define he OECD volume index ha chains he Γ F defined by (16) ino a ime series index, Q H. Define Q H as follows: (20) Q H ; Q H Q H 1 Γ F ; = 2001,,2012. Now use he counry shares of OECD real GDP s n lised in Table 3 and he aggregae index Q H o define he following preliminary harmonized counry volumes for counry n in year, q Hn, as follows: (21) q Hn Q H s n ; n = 1,,34; = 2000,,2012. Noe ha for each year, n =1 34 q Hn = n =1 34 Q H s n = Q H ( n =1 34 s n ) = Q H and so he harmonized volumes saisfy he wo principles lised above. In principle, he counry volumes defined by (21) are independen of counry prices and exchange raes ( 36 ). ( 35 ) McCarhy (2013, pp ) explains in some deail why esimaes of real GDP based on naional growh informaion do no mach up exacly wih relaive GDP esimaes based on PPP benchmark informaion. The PPP informaion is generally no as accurae as naional price index informaion due o he difficuly of maching represenaive producs across counries. However, counry mehodology for consrucing naional price indexes (such as he Consumer Price Index) differs considerably across counries; e.g., some counries may use ou of dae reference expendiure baskes, some counries use Carli indexes a he elemenary level while ohers use he Jevons or Duo indexes which generae lower esimaes of inflaion a he elemenary level and some counries may use qualiy adjusmen mehods more exensively han ohers. All of hese mehodological differences lead o inconsisencies beween he ime series and cross secional esimaes. Finally, he index number formulae used a he naional levels and in he consrucion of he benchmark PPPs are in general no ransiive and so i is impossible o achieve perfec consisency. ( 36 ) However, in pracice, he PPPs do no do a perfec job in eliminaing exchange rae effecs (since adjused exchange raes are used in place of rue PPPs o deflae inernaional rade flows). If he relaive PPPs are independen of he choice of he numeraire counry, hen he relaive volumes defined by (20) will also be independen of his choice. EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 19

22 1 Oupu growh and inflaion across space and ime I is of ineres o define US dollar prices for real GDP for each counry. Recall ha he value of counry n s nominal GDP convered ino US dollar a marke exchange raes for year was defined as v n. The corresponding harmonized US dollar price of a uni of (comparable across counries) real GDP for counry n in year is defined as follows: (22) p Hn v n /q Hn ; n = 1,,34; = 2000,,2012. Table 5: Harmonized OECD counry GDP volumes in comparable US dollar unis of measuremen q Hn n 2000 q Hn 2001 q Hn 2002 q Hn 2003 q Hn 2004 q Hn 2005 q Hn 2006 q Hn 2007 q Hn 2008 q Hn 2009 q Hn 2010 q Hn 2011 q Hn 2012 q Hn Source: Auhor's calculaions In order o make he harmonized volumes and prices defined by (21) and (22) comparable o he counry prices and volumes expressed in US dollar ha are lised in he Annex in Tables A3 and A4, we impose a normalizaion on he prices defined by (22) ha makes he price level for 20 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

23 Oupu growh and inflaion across space and ime 1 he US in 2000 equal o uniy; i.e., we divide all prices defined by (22) by a consan ha ses he resuling p H equal o 1 and he quaniies or volumes defined by (21) are all muliplied by his consan. The resuling normalized q Hn and p Hn are lised in Tables 5 and 6. Noe ha q H = q and p H = p = 1 so ha counry GDP volumes are measured as muliples of a bundle of US GDP in he year Thus he price levels in Table 6 measure he US dollar value of consan bundle of GDP ha is (in heory) comparable across counries. The price levels in Table 6 are comparable across space and ime, whereas he price levels pn lised in Table A3 of he Annex are only comparable across ime for each counry. Table 6: Harmonized OECD counry GDP price levels in comparable US dollar unis of measuremen p Hn n p Hn 2000 p Hn 2001 p Hn 2002 p Hn 2003 p Hn 2004 p Hn 2005 p Hn 2006 p Hn 2007 p Hn 2008 p Hn 2009 p Hn 2010 p Hn 2011 p Hn Source: Auhor's calculaions EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 21

24 1 Oupu growh and inflaion across space and ime From Table 6, i can be seen ha he counries wih he lowes price levels (in US dollar) in 2012 are counries 13, 21, 25 and 32 (Hungary, Mexico, Poland and Turkey) wih price levels in he 0.76 o 0.77 range. Counries wih relaively high price levels in 2012 are counries 1 (Ausralia, p H = 2.00), 4 (Canada, p H = 1.62), 7 (Denmark, 1.77), 9 (Finland, 1.58), 18 (Japan, 1.76), 20 (Luxembourg, 1.56), 23 (New Zealand, 1.55), 24 (Norway, 2.01), 30 (Sweden, 1.69) and 31 (Swizerland, 1.96). These price level esimaes are ( 37 )(imperfec) indicaors of he compeiveness of he counry on inernaional markes, wih lower price levels indicaing greaer compeiveness. A problem wih he volume esimaes lised in Table 5 is ha hey do no respec naional growh raes of GDP by counry, in ha hey are no required o agree wih he naional growh raes ha are produced by he individual counries; only he aggregae OECD growh rae is respeced. In he following wo secions, we will derive alernaive counry volume esimaes ha are comparable over ime and space. These alernaive esimaes will respec counry growh raes bu hey will no reproduce he real OECD counry expendiure shares lised in Table 3 for all ime periods. 4. OECD growh and inflaion using counry annual GDP volume growh raes and base period shares of OECD real GDP We generae comparable counry GDP volume esimaes for OECD counries covering he period by using he real GDP counry volume shares for 2000, he s n 2000 lised in Table 3 above, along wih he naional growh raes of counry real GDP relaive o 2000, he Q n / Q n 2000 lised in Table A2 of he Annex. This is a ypical sraegy in forming esimaes of real GDP ha rely on PPPs ha are only produced infrequenly. Our purpose in lising hese esimaes is o evaluae how differen he resuling esimaes are from our preferred harmonized volume esimaes, q Hn, lised in Table 5 above. Define preliminary base period esimaes of counry GDP volumes for year and counry n, q Bn, as follows: (23) q Bn s n 2000 (Q n /Q n 2000 ) ; n = 1,,34 ; = 2000,,2012. The above esimaes are obviously based on he counry shares of real OECD GDP ha prevailed in 2000 (he s n 2000 ) and he long erm counry growh raes of real GDP (he Q n /Q n 2000 ). The companion counry US dollar price levels for counry n and year, p Bn, are defined as follows: (24) p Bn v n /q Bn ; n = 1,,34 ; = 2000,,2012 where v n is he nominal value of GDP for counry n in year convered ino US dollar a marke exchange raes for ha year. In order o make he volumes and prices defined by (23) and (24) comparable o he counry prices and volumes expressed in US dollar ha are lised Tables 5 and 6 in he previous secion, ( 37 ) The price levels p Hn are imperfec indicaors of compeiveness because no all componens of GDP are inernaionally raded. Moreover, hese price levels are no independen of he choice of he numeraire currency (US dollar in his case). They are also imperfec because hey depend heavily on he accuracy of he underlying PPPs and hese PPPs are subjec o considerable error variances due o he difficulies involved in maching produc prices (and quaniies) across counries. 22 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

25 Oupu growh and inflaion across space and ime 1 we impose a normalizaion on he prices defined by (24) ha makes he price level for he US in 2000 equal o uniy; i.e., we divide all prices defined by (24) by a consan ha ses he resuling p B equal o 1 and he volumes defined by (23) are all muliplied by his consan. The resuling normalized p Bn are lised in Table 7 ( 38 ). Table 7: OECD counry GDP price levels in comparable US dollar unis of measuremen p Bn based on counry growh raes of GDP volumes and year 2000 counry shares of OECD oupu n 2000 p Bn 2001 p Bn 2002 p Bn 2003 p Bn 2004 p Bn 2005 p Bn 2006 p Bn 2007 p Bn 2008 p Bn 2009 p Bn 2010 p Bn 2011 p Bn 2012 p Bn Source: Auhor's calculaions The differences beween he enries in Tables 6 and 7 are very large. If we ake each column in Table 6, subrac he corresponding enries in he same column of Table 7 and hen ake he ( 38 ) The enries in Tables 5 and 6 enable one o recover he US dollar values of GDP, equal o v n = p Hn q Hn for n = 1,,34 and = 2000,,2012. Then he q Bn can be recovered as v n /p Bn. EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 23

26 1 Oupu growh and inflaion across space and ime absolue value of he differences, we find ha he average absolue difference grows from 0 in 2000 o 9.4 percenage poins in 2012 ( 39 ). The maximum absolue difference grows from 0 in 2000 o 54.0 percenage poins in These are massive differences in price levels, which ranslae ino massive differences in GDP levels. This problem of he inconsisency wih naional growh raes is well known bu mos users of PPP adjused counry real volume esimaes are no aware of he magniude of hese inconsisencies ( 40 ). In he following secion, we underake a compuaion ha is similar o hose of he presen secion excep ha we use he real volume shares of 2012 as he benchmark shares insead of he shares of OECD growh and inflaion using counry annual GDP volume growh raes and curren period shares of OECD real GDP We generae comparable counry GDP volume esimaes for OECD counries covering he period by using he real GDP counry volume shares for 2012, he s n 2012 lised in Table 3 above, along wih he naional growh raes of counry real GDP relaive o 2000, he Q n /Q n 2000 lised in Table A2 of he Annex. This mehod for forming comparable counry GDP volumes is used by he World Bank when he Inernaional Comparisons Projec produces a new se of PPPs ( 41 ). The mehodology is sraighforward and follows he approach used in he previous secion excep ha he 2012 counry volume shares are used in place of he 2000 shares. The preliminary end of sample period esimaes of counry GDP volumes for year and counry n, q En, is defined as follows: (25) q En s n 2012 (Q n /Q n 2012 ) ; n = 1,,34 ; = 2000,,2012. The above esimaes are obviously based on he counry shares of real OECD GDP ha prevailed in 2012 (he s n 2012 ) and he levels of real GDP in year relaive o he corresponding counry n level in 2012 (he Q n /Q n 2012 ). The companion counry US dollar price levels for counry n and year, p En, are defined as follows: (26) p En v n /q En ; n = 1,,34 ; = 2000,,2012 where v n is he nominal value of GDP for counry n in year convered ino US dollar a marke exchange raes for ha year. In order o make he volumes and prices defined by (25) and (26) comparable o he harmonized counry prices and volumes expressed in US dollar ha are lised Tables 5 and 6 in Secion 3, ( 39 ) The sequence of average absolue differences in percenage poins over he 13 years is as follows: 0, 0.8, 1.6, 2.0, 2.2, 3.8, 5.2, 6.4, 7.6, 6.8, 8.0, 9.3, 9.4. The sequence of maximum absolue differences in percenage poins over he 13 years is as follows: 0, 2.9, 5.2, 6.0, 5.4, 14.0, 25.1, 28.2, 46.9, 24.8, 34.3, 48.5, ( 40 ) See Eurosa (2012, p. 18) on his poin. ( 41 ) See Chaper 18 in he World Bank (2013). The Penn World Tables use he exrapolaion mehodology described in his secion and he previous secion o consruc esimaes of comparable real GDP for periods subsequen o he las available ICP round and prior o he firs available ICP round; see Summers and Heson (1991) and Feensra, Inklaar and Timmer (2013). 24 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

27 Oupu growh and inflaion across space and ime 1 we impose a normalizaion on he prices defined by (26) ha makes he price level for he US in 2000 equal o uniy; i.e., we divide all prices defined by (26) by a consan ha ses he resuling p E equal o 1 and he quaniies or volumes defined by (25) are all muliplied by his consan. The resuling normalized p En are lised in Table 8 ( 42 ). Table 8: OECD counry GDP price levels in comparable US dollar unis of measuremen p En based on counry growh raes of GDP volumes and year 2012 counry shares of OECD oupu n 2000 p En 2001 p En 2002 p En 2003 p En 2004 p En 2005 p En 2006 p En 2007 p En 2008 p En 2009 p En 2010 p En 2011 p En 2012 p En Source: Auhor's calculaions I can be seen ha here are subsanial differences beween he price levels lised in Table 8 as compared o he price levels lised in Table 7 and he harmonized price levels lised in Table 6. If we ake each column in Table 6, subrac he corresponding enries in he same column of ( 42 ) As in he previous secion, he q En can be recovered as v n /p En. EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 25

28 1 Oupu growh and inflaion across space and ime Table 8 and hen ake he absolue value of he differences, we find ha he average absolue difference for 2000 over he 34 counries is 6.0 percenage poins, which increases o 7.9 percenage poins for 2005 and hen gradually decreases o 4.2 percenage poins in Over all observaions, he maximum absolue deviaion is 35.6 percenage poins ( 43 ). Again hese are large differences in price levels, which ranslae ino large differences in GDP levels. For comparing real GDP levels across ime and space, he resuls presened in his secion indicae ha he sraegy of using naional growh raes and a single cross counry comparison of real GDP levels will no lead o sable comparisons. The harmonizaion sraegy suggesed in Secion 3 will lead o sable comparisons and if he accuracy of he annual sequence of PPPs is roughly consan, he resuling harmonized esimaes seem o be preferable o he consisen naional growh rae esimaes ha are based on a single cross counry comparison. 6. OECD growh and inflaion using adjused counry annual GDP volume growh raes and OECD shares of real GDP for wo benchmark years The OECD provides annual PPPs so ha esimaes of relaive GDP volumes can be consruced for all member counries for each year. However, he World Bank s ICP PPPs are only available a infrequen inervals ( 44 ). We now consider using he benchmark GDP shares for he years 2000 and 2012 along wih informaion on naional GDP growh raes in order o inerpolae beween he benchmark years. We propose an inerpolaion mehod ha leads o counry shares of real GDP ha are exacly consisen wih he shares s n for he year 2000 and he shares s n for he year We begin by using he mehodology of Secion 3 o consruc counry measures of real GDP ha jump from he year 2000 o he year The long erm growh facor for counry n can be defined as Q n /Q n where Q n is counry n s GDP volume in year ( 45 ). We use hese long 2000 erm growh facors along wih he year 2000 counry shares of OECD real GDP, s n, in order o define he OECD Laspeyres ype long erm growh facor, Γ L, as he following weighed average of he naional long erm growh facors: (27) Γ L n =1 34 s n 2000 (Q n 2012 /Q n 2000 ) ; = 2001,,2012. The counerpar o he Laspeyres ype formula defined by (27) is he following Paasche ype formula ha uses he shares of 2012 and reciprocal long erm growh raes: (28) Γ P [ n =1 34 s n 2012 (Q n 2012 /Q n 2000 ) 1 ] 1 ; = 2001,,2012. ( 43 ) The sequence of average absolue differences over he 34 counries in percenage poins over he 13 years is as follows: 6.0, 6.1, 6.2, 7.0, 7.8, 7.9, 6.2, 6.8, 5.9, 5.4, 5.1, 5.2, 4.2. The sequence of maximum absolue differences in percenage poins over he 13 years is as follows: 24.6, 25.9, 32.3, 35.6, 33.1, 27.1, 20.5, 22.0, 16.0, 24.7, 18.7, 12.6, 6.3. Recall ha we normalized he price level of he US o be 1 in 2000 for he p Hn and he p En. If insead of using he normalizaions p H = p E = 1 when consrucing Tables 6 and 9, we used he normalizaions p H = p E = 1, we would find ha he absolue differences beween he resuling p Hn and p En would equal 0 for all counries n for = Thus he choice of normalizaion (and hence of he unis of measuremen) can affec he resuls. ( 44 ) The World Bank has produced benchmark PPPs for over 150 counries for 2005 and 2011; see World Bank (2013). ( 45 ) These long erm counry growh facors are convenienly lised in he las column of Table A2 in he Annex. 26 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

29 Oupu growh and inflaion across space and ime 1 A symmeric average of he wo indexes leads o he following Fisher ype formula for OECD long erm volume growh going from he year 2000 o he year 2012: (29) Γ F [Γ L Γ P ] 1/2 ; = 2001,,2012. The long erm indexes defined by (27)-(29) urn ou o be , and respecively, so ha here is pracically no difference in he hree indexes for his daa se ( 46 ). We ake he Fisher measure as our preferred measure of OECD volume growh beween 2000 and 2012, and use his o define counry volumes for Preliminary esimaes of counry GDP volumes in comparable unis for he years 2000 and , q In and q In (he index I indicaes ha hese are inerpolaed esimaes), are defined as follows: (30) q In 2000 s n 2000 ; q In 2012 Γ F s n 2012 ; n = 1,,34. The volumes defined by (30) will be imposed as consrains on our inerpolaion scheme. Define he implied long erm growh facor over he years for counry n, g n, ha is implied by he esimaes of counry levels given by equaions (30): (31) g n q In 2012 /q In 2000 ; n = 1,,34. These growh facors are no necessarily equal o he naional growh facors G n ha are implied by he naional growh raes lised in Table A2 of he Annex: (32) G n Q n 2012 /Q n 2000 ; n = 1,,34. Thus for each counry n, here is an error facor or discrepancy, E n g n /G n beween he implied growh raes gn defined by (31) and he naional growh raes beween 2000 and 2012, G n defined by (32). We will disribue hese errors in a proporional manner and use he resuling adjused naional growh raes o inerpolae beween he wo benchmark observaions. Thus define he counry n proporional annualized discrepancy facor, α n, as follows ( 47 ): (33) α n [g n /G n ] 1/12 ; n = 1,,34. The q In for non-benchmark years can now be defined as follows ( 48 ): (34) q In q In 1 (Q n /Q n 1 )α n ; n = 1,,34 ; = 2001,.,2011. Once he q In have been defined, he corresponding US dollar price levels p In are defined in he usual way: (35) p In v n /q In ; n = 1,,34 ; = 2001,.,2011. In order o make he volumes and prices defined by (34) and (35) comparable o he harmonized counry prices and volumes expressed in US dollar ha are lised Tables 5 and 6 in Secion 3, we impose a normalizaion on he prices defined by (35) ha makes he price level for he ( 46 ) Noe ha (29) defines a direc comparison of he daa of 2000 wih he daa of 2012 whereas in Secion 3 above, we used chained Fisher ype indexes o go from 2000 o The chained Fisher index for 2012 relaive o 2000 is equal o , which is very close o is direc counerpar, ( 47 ) The average of he g n /G n was The maximum raio was 1.27 (Norway) and he minimum raio was 0.81 (Israel). The PPP based growh raes rea changes in he erms of rade differenly han he naionally based growh raes and so flucuaions in he price of oil probably explain he Norwegian divergence. For he hree larges counries, he raio was 1.05 (Germany), 0.94 (Japan) and 0.97 (US). ( 48 ) I can be verified ha if we apply definiions (34) for = 2012, we obain he q In 2012 defined by (30). EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 27

30 1 Oupu growh and inflaion across space and ime US in 2000 equal o uniy; i.e., we divide all prices defined by (35) by a consan ha ses he resuling p I equal o 1 and he quaniies or volumes defined by (34) are all muliplied by his consan. The resuling normalized pin are lised in Table 9 ( 49 ). Table 9: OECD counry GDP price levels in comparable US dollar unis of measuremen p In based on adjused counry growh raes of GDP volumes and year 2000 and 2012 counry shares of OECD oupu n p In 2000 p In 2001 p In 2002 p In 2003 p In 2004 p In 2005 p In 2006 p In 2007 p In 2008 p In 2009 p n 2010 p In 2011 p In Source: Auhor's calculaions Again, i can be seen ha here are some subsanial differences beween he price levels lised in Table 9 as compared o he price levels lised in Table 6 bu he discrepancies are much reduced as compared o he discrepancies when only one benchmark se of PPPs is used. The overall sample average absolue discrepancy is now only 1.9 percenage poins. The average ( 49 ) As usual, he q In can be recovered as v n /p In. 28 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

31 Oupu growh and inflaion across space and ime 1 absolue difference for 2000 over he 34 counries is 0, which increases o 3.2 percenage poins for 2005 and 2007 and hen gradually decreases o 0.3 percenage poins in Over all observaions, he maximum absolue deviaion is 12.5 percenage poins ( 50 ). Some enaive conclusions can be drawn from he ables in his secion and he previous secions. Firs, he inerpolaion mehod which is consisen wih benchmark expendiure shares for wo widely separaed years seems o work quie well. If he benchmark PPPs are of equal qualiy, he inerpolaion mehod is much beer han simply projecing he counry shares from a single benchmark using naional growh raes ( 51 ). Second, if i is oo expensive o prepare annual PPPs for a group of counries, hen he inerpolaion mehod will probably generae comparable counry real GDP volumes ha are close o our preferred harmonized volumes described in Secion 3, provided ha benchmark PPPs are calculaed every hree o five years ( 52 ). Feensra, Inklaar and Timmer (2013, pp ) (FIT) use a simple inerpolaion mehod o harmonize heir new Penn World Table esimaes of real GDP growh (in comparable unis of measuremen) using boh ICP informaion beween wo benchmarks and naional informaion on GDP growh. Their inerpolaion mehod is similar o our suggesed mehod, in ha heir inerpolaed esimaes are consisen wih he relaive GDP levels for he PPP benchmark years. The key o heir inerpolaion mehod is he consrucion of inerpolaed PPPs beween ICP benchmarks. We explain heir mehod using our noaion and adaping heir analysis o he problem of consrucing PPPs for he years , given ha we have PPPs for he benchmark years 2000 and Recall ha he domesic price level for counry n in year was defined as P n V n /Q n for n = 1,,34 and = 2000,,2012 (and hese price levels are lised in Annex Table A1). Recall also ha PPP n was defined as he number of unis of he naional currency of counry n ha is required o purchase one dollar of US (real) GDP in year (and hese OECD PPPs are lised in Annex Table A5). Using our noaion, heir inerpolaed PPP for counry n in year, PPP FITn, is defined as follows: (36) PPP FITn (1 w )(PPP n 2000 )(P n /P n 2000 ) + w (PPP n 2012 )(P n /P n 2012 ) ; n = 1,,34; = 2000, 2001,,2012 where he weigh funcion w is defined as follows: (37) w ( 2000)/12 ; = 2000, 2001,,2012. Thus w grows linearly in wih w 2000 = 0 and w 2012 = 1. Noe ha PPP FITn 2000 = PPP n 2000 and PPP FITn 2012 = PPP n 2012 so ha he inerpolaed PPPs coincide wih he acual PPPs for he wo benchmark years, 2000 and Thus he inerpolaed PPP for counry n in year, PPP FITn, is a simple weighed average of wo exrapolaed PPPs for counry n. The firs index in he weighed average uses he PPPs for 2000, PPP n 2000, and pushes hese PPPs forward using he normalized ( 50 ) The sequence of average absolue differences over he 34 counries in percenage poins over he 13 years is as follows: 0, 1.0, 1.8, 2.0, 2.3, 3.2, 3.0, 3.2, 2.8, 2.6, 1.7, 1.2, The sequence of maximum absolue differences in percenage poins over he 13 years is as follows: 0, 3.9, 9.7, 11.9, 9.4, 10.7, 8.3, 12.5, 12.3, 9.9, 6.1, 3.2, The reason why he differences are no all equal o 0 for 2012 is ha he direc aggregae Fisher index going from 2000 o 2012 differs slighly from is chained counerpar defined in Secion 3. ( 51 ) Beer means more consisen wih our preferred harmonized volumes ha could be calculaed if annual PPPs were available. ( 52 ) The inerpolaion mehod ha we described in his secion is no he only possible mehod ha could be used o calculae comparable real GDP series over ime and space when benchmark PPPs are only available infrequenly. In paricular, economericians may prefer o use an inerpolaion mehod ha is based on he Kalman filer; see Rao, Rambaldi and Doran (2010) (2011) for he descripion of such a mehod. For addiional mehods for harmonizing cross secional and ime series esimaes of real GDP, see Rao, Rambaldi and Balk (2013). Summers and Heson (1991, pp ) also used an economeric mehod o reconcile he differences beween naional growh raes and ICP generaed esimaes of relaive GDP levels. However, saisical agencies are generally relucan o adop mehods ha rely heavily on economerics so he simple mehod of inerpolaion described here is proposed as an aracive alernaive. EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 29

32 1 Oupu growh and inflaion across space and ime domesic inflaion raes P n /P n 2000 and he second index uses he PPPs for 2012, PPP n 2012, and pushes hese PPPs backwards using he normalized domesic inflaion raes P n /P n We do no consruc OECD real volumes for our sample period using he complee Feensra, Inklaar and Timmer mehodology bu we experimen wih heir mehod of weighing. Recall ha Tables 7 and 8 lised he OECD counry US dollar prices (in comparable unis across ime and space), p Bn and p En, where he prices p Bn (p En ) were based on counry growh raes of GDP volumes and year 2000 (2012) counry shares of OECD oupu. Using he weighs w defined by (37), define he FIT ype US dollar price levels, p FITn, as he following weighed averages of he 2000 US dollar prices p Bn 2000 and 2012 US dollar prices p En2012 : (38) p FITn (1 w )p Bn w p En 2012 ; n = 1,,34; = 2000, 2001,,2012. I can be seen ha p FITn 2000 = p Bn 2000 and p FITn 2012 = p En 2012 for n = 1,,34. Now compare he prices p FITn o our preferred Harmonized US dollar price levels p Hn lised in Table 6. Take he absolue value of he differences, p Hn p FITn. The sample average absolue difference (over ime periods and counries n) is 12.2 percenage poins. The wihin year absolue difference grows from 0 in 2000 o 25.3 percenage poins in 2008 ( 53 ). The maximum absolue difference grows from 0 in 2000 o 83.4 percenage poins in These are large differences in price levels, which ranslae ino large differences in real GDP levels. The relaive volumes generaed by dividing he US dollar GDP values by he corresponding US dollar prices defined by (38) are no longer independen of he choice of he numeraire counry. Thus insead of aking he weighed arihmeic means of he prices p Bn 2000 and p En 2012, ake he corresponding weighed geomeric means and denoe he resuling prices by p FITGn ( 54 ) Compare he prices p FITGn o our preferred Harmonized US dollar price levels phn lised in Table 6 and ake he absolue value of he differences, p Hn p FITGn. The sample average absolue difference is now 13.6 percenage poins, which is larger han he average differences using he weighed arihmeic means. The wihin year absolue difference grows from 0 in 2000 o 28.6 percenage poins in 2008 ( 55 ). The maximum absolue difference grows from 0 in 2000 o 84.7 percenage poins in Why do he above varians of he inerpolaion mehod suggesed by Feensra, Timmer and Inklaar generae US dollar price levels (and he corresponding counry real GDP levels) ha are so differen from he Harmonized counry US dollar price levels ha are lised in Table 6? The reason is ha he inerpolaed PPPs defined by (36) (and heir geomeric counerpars) depend on counry inflaion raes, which are quie variable ( 56 ). In order o eliminae he effecs of counry inflaion raes, we ried he following varian of he FIT inerpolaion mehod: insead ( 53 ) The sequence of wihin year average absolue differences in percenage poins over he 13 years is as follows: 0, 6.2, 6.2, 9.3, 14.5, 15.8, 14.7, 23.0, 25.3, 14.9, 9.4, 14.8, 4.2. The differences are nonzero in 2012 even hough he corresponding PPPs for 2012 are exacly consisen wih he OECD PPPs for Thus while he US dollar prices p Hn 2012 equal λp FITn 2012 for n = 1,,34, he facor of proporionaliy λ is no equal o one and hus he differences are nonzero in ( 54 ) Usually, aking geomeric means (raher han arihmeic means) of wo indexes leads o indexes ha have beer invariance and homogeneiy properies. For examples of his phenomenon, see Diewer (1997) and Hill and Fox (1997). ( 55 ) The sequence of wihin year average absolue differences in percenage poins over he 13 years is as follows: 0, 5.2, 5.0, 10.8, 16.7, 18.6, 17.5, 26.3, 28.6, 16.2, 11.6, 16.0, 4.2. ( 56 ) More precisely, he FIT mehod ineracs counry inflaion raes wih he linear in ime weighs in equaions (36) and hese weighs are independen of he magniude of economic price and quaniy daa ha perain o he counries whereas our inerpolaion mehod depends only on counry volume indexes over he sample period and he relaive volumes generaed by he PPPs a he beginning and end of he sample period. If he α n defined by (33) were all equal o uniy, hen he marix of counry real volumes generaed by exrapolaing he base period relaive GDP volumes forward by naional growh raes would be equal o he marix of counry real volumes generaed by exrapolaing he final period relaive GDP volumes backwards (afer normalizaion o a common base) and our inerpolaion mehod would generae his common marix of comparable over ime and space real GDP volumes. Under he same condiions, he FIT mehod would no generae he same marix (excep by chance). 30 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

33 Oupu growh and inflaion across space and ime 1 of using equaions (36) o inerpolae he PPPs beween he years 2000 and 2012, use he following equaions o define he inerpolaed PPP In for counry n and year : (39) PPP In (1 w )PPP n w PPP n 2012 ; n = 1,,34; = 2000, 2001,,2012 where he weigh w is sill defined by (37). Now reurn o our descripion for he consrucion of he harmonized counry esimaes for GDP and US dollar price levels ha is in Secion 3 bu replace PPP n in equaions (12) by heir inerpolaed counerpars PPP In defined by (39). Denoe he resuling US dollar price levels by pin. We compare he prices p In o our preferred US dollar price levels p Hn lised in Table 6 and as usual, ake he absolue value of he differences, p Hn p In. The sample average absolue difference (over ime periods and counries n) is only 2.48 percenage poins. The wihin year average absolue difference grows from 0 in 2000 o 4.3 percenage poins in 2005 ( 57 ). The maximum absolue difference is 20.6 percenage poins in The performance of his inerpolaion mehod is much beer han he previous inerpolaion mehod bu sill no quie as good as our suggesed inerpolaion mehod ha was described a he beginning of his secion (which generaed an average absolue difference of only 1.92 percenage poins). The above numerical experimens wih inerpolaion mehods ha are similar in spiri o he mehod used by Feensra, Inklaar and Timmer are no conclusive since i assumes ha he ruh is bes defined by he harmonized pariies p Hn defined earlier in Secion 3. However, a a minimum, he numerical experimens do show ha he mehod of inerpolaion beween benchmark Purchasing Power Pariy rounds does maer. Addiional research ino alernaive mehods of inerpolaion is required. 7. Conclusion A number of ineresing poins emerged from our invesigaions. If our focus is on measuring overall OECD GDP growh and PPP informaion is unavailable, hen he mehod ha is explained in Secion 2 may be used. The overall OECD growh measures, γ, do no depend on PPPs or he choice of he numeraire currency bu exchange rae flucuaions can cause perhaps unwarraned flucuaions. We compued γ using he US and hen Germany as he numeraire counry and found ha while he Fisher index of OECD real GDP growh remained invarian, he accompanying Fisher price indexes, P and P EU, exhibied wildly differen raes of growh. Thus hese price indexes are useless as indicaors of OECD inflaion. Three alernaive measures of overall OECD GDP growh were defined in Secion 3: he Laspeyres, Paasche and Fisher measures, γ L, γ P and γ F. These measures depended on he annual OECD PPP informaion. The Laspeyres measure is he official OECD measure for overall OECD growh bu he Fisher measure seems preferable on concepual grounds. However, for our daa se, all hree measures were very close o each oher. The Fisher measure of OECD growh, γ F, (which used real GDP share weighs consruced using PPPs) grew on average abou 1/10 of a percenage poin more rapidly per year over he period han our Secion 2 measure of OECD GDP growh γ, (which used exchange rae based share weighs). This resul was expeced since he Secion 3 PPP based share weighs S n ( 57 ) The sequence of wihin year average absolue differences in percenage poins over he 13 years is as follows: 0, 1.1, 1.9, 2.9, 4.0, 4.3, 3.7, 3.7, , 2.4, 1.9, 0.3. The sequence of wihin year maximum absolue differences in percenage poins over he 13 years is as follows: 0, 6.7, 13.8, 20.6, 20.3, 18.6, 15.0, 13.3, 15.8, 8.2, 6.8, 5.8, 0.5. EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 31

34 1 Oupu growh and inflaion across space and ime for rich counries (which generally have lower raes of GDP growh) are generally smaller han he corresponding Secion 2 exchange rae based share weighs S n. Secion 3 also inroduced hree measures of OECD aggregae GDP price inflaion, he Laspeyres, Paasche and Fisher measures ρ L, ρ P and ρ F defined by (19). These inflaion measures used PPP based counry share weighs o weigh he counry inflaion raes and were much more saisfacory han he Secion 2 measures of OECD aggregae inflaion. The hree measures differed somewha so he choice of index maers. Our preference is for he Fisher measure ρ F since i saisfies a ime reversal es whereas he oher wo indexes do no. We used wo principles in Secion 3 o generae our harmonized esimaes of real GDP for OECD counries: (i) The resuling harmonized esimaes of counry volumes q Hn mus be consisen wih he real volume shares s n lised in Table 3 and (ii) OECD aggregae real GDP growh mus be equal o he raes of aggregae growh generaed by our recommended Fisher indexes γ F. Once he harmonized esimaes of real GDP q Hn have been generaed, companion US dollar counry price levels p Hn can be generaed as p Hn v n /q Hn where v n is he exchange rae convered US dollar nominal value of GDP for counry n in year. These counry price levels are useful (bu imperfec) indicaors of a counry s compeiiveness in year. In Secions 4 and 5, alernaive measures of comparable levels of real GDP and he accompanying US dollar price levels were consruced. The measures consruced in Secion 4 used he PPP informaion for 2000 and naional growh raes for real GDP by counry whereas he esimaes consruced in Secion 5 subsiued he PPP informaion for We found remendous discrepancies in hese esimaes as compared o he harmonized esimaes consruced in Secion 3. The resuls lised in Secions 3 o 5 show ha i is very hazardous for analyss ineresed in comparaive levels of GDP across counries o use naional growh raes and a single cross counry comparison of real GDP levels. Evenually, he single cross counry comparison is replaced by anoher single cross counry comparison bu he new se of comparable GDP levels across ime and space can be vasly differen from he earlier se of GDP levels, paricularly for small counries. These resuls reinforce he case for using he harmonized series ha were defined in Secion 3. Using he Secion 3 mehodology, he previously consruced harmonized esimaes of relaive GDP levels remain unchanged as anoher year of daa is added. If PPP compuaions for a group of counries are only done on an infrequen basis (raher han on an annual basis as is he case for he OECD), hen he inerpolaion mehod explained in Secion 6 may prove o be a useful mehod for obaining comparable GDP levels ha are consisen wih he GDP relaive levels for he wo benchmark years. The resuls in Secion 6 also indicae ha differen inerpolaion mehods can generae very differen resuls. In paricular, he presen inerpolaion mehod used in he Penn World Tables did no work well wih our OECD daa base. Of course, he harmonizaion mehods ha have been suggesed in his paper can be applied o any oher value aggregae, such as consumpion, invesmen or domesic absorpion. The resuls in his paper show ha if counries wan o compare he size of heir economies or measure expendiure growh or price inflaion for a group of counries, i is absoluely essenial ha hose counries underake regular cross counry comparisons of prices. 32 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

35 Oupu growh and inflaion across space and ime 1 Acknowledgemens The auhors hank wo anonymous referees, Ber Balk, Yuri Dikhanov, Dennis Fixler, Anne-Sophie Fraisse, Rober Inklaar, Paul McCarhy, Prasada Rao, Marshall Reinsdorf and Paul Schreyer for helpful commens on an earlier draf. Financial suppor from he Social Sciences and Humaniies Research Council of Canada and he Ausralian Research Council s Discovery Program (DP ) is graefully acknowledged. References Balk, B.M.(2008), Price and Quaniy Index Numbers, New York: Cambridge Universiy Press. Diewer, W.E. (1978), Superlaive Index Numbers and Consisency in Aggregaion, Economerica Vol. 46, issue 4, pp Diewer, W.E. (1992), Fisher Ideal Oupu, Inpu and Produciviy Indexes Revisied, Journal of Produciviy Analysis, No 3, pp Diewer, W.E. (1993), The Early Hisory of Price Index Research, pp in Essays in Index Number Theory, Vol. 1, W.E. Diewer and A.O. Nakamura (eds.), Amserdam: Norh-Holland. Diewer, W.E. (1997), Commenary on Mahew D. Shapiro and David W. Wilcox: Alernaive Sraegies for Aggregaing Prices in he CPI, The Federal Reserve Bank of S. Louis Review, Vol. 79:3, pp Diewer, W.E. (2009), Similariy Indexes and Crieria for Spaial Linking, pp in Purchasing Power Pariies of Currencies: Recen Advances in Mehods and Applicaions, D.S. Prasada Rao (ed.), Chelenham UK: Edward Elgar. Diewer, W.E. and K.J. Fox (2017), Subsiuion Bias in Mulilaeral Mehods for CPI Consrucion using Scanner Daa, Discussion Paper 17-02, Vancouver School of Economics, Universiy of Briish Columbia, Canada. Eurosa (2012), Eurosa-OECD Mehodological Manual on Purchasing Power Pariies, 2012 Ediion, Luxembourg: Publicaions Office of he European Union. Feensra, R.C., R. Inklaar and M.P. Timmer (2013), The Nex Generaion of he Penn World Table, NBER Working Paper, No , Cambridge MA: NBER. Fisher, I. (1911), The Purchasing Power of Money, London: Macmillan. Fisher, I. (1922), The Making of Index Numbers, Boson: Houghon-Mifflin. Griliches, Z. (1985), Daa and Economericians The Uneasy Alliance, American Economic Review, Vol. 75, pp Hill, R.J. (2001), Measuring Inflaion and Growh Using Spanning Trees, Inernaional Economic Review, Vol. 42, pp Hill, R.J. (2004), Consrucing Price Indexes Across Space and Time: The Case of he European Union, American Economic Review, Vol. 94, pp EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 33

36 1 Oupu growh and inflaion across space and ime Hill, R.J. (2009), Comparing Per Capia Income Levels Across Counries Using Spanning Trees: Robusness, Prior Resricions, Hybrids and Hierarchies, pp in Purchasing Power Pariies of Currencies: Recen Advances in Mehods and Applicaions, D.S. Prasada Rao (ed.), Chelenham UK: Edward Elgar. Hill, R.J. and K.J. Fox (1997), Splicing Index Numbers, Journal of Business and Economic Saisics, Vol. 15, pp Hill, T.P. (1988), Recen Developmens in Index Number Theory and Pracice, OECD Economic Sudies, 10, pp Hill, T.P. (1993), Price and Volume Measures, pp in Sysem of Naional Accouns 1993, Eurosa, IMF, OECD, UN and World Bank, Luxembourg, Washingon, D.C., Paris, New York, and Washingon, D.C. Lehr, J. (1885), Beiräge zur Saisik der Preise, Frankfur: J.D. Sauerlander. Marshall, A. (1887), Remedies for Flucuaions of General Prices, Conemporary Review, Vol. 51, pp McCarhy, P. (2013), Exrapolaing PPPs and Comparing ICP Benchmark Resuls, pp in Measuring he Real Size of he World Economy, Washingon D.C., World Bank. OECD (2001), Inroducion of he Euro in OECD Saisics, Saisics Direcorae, Naional Accouns, STD/NA(2001)8, Paris: OECD, Sepember 14. OECD (2014), Mehodological Noes: Compilaion of G20 Consumer Price Index, Paris: OECD. Rao, D.S.P., A. Rambaldi and B.M. Balk (2013), Purchasing Power Pariies, Price Levels and Measures of Regional and Global Inflaion, preliminary paper presened a he Economic Measuremen Group Workshop 2013, November 28, Universiy of New Souh Wales, Sydney, Ausralia. Rao, D.P., A.N. Rambaldi, and H.E. Doran (2010), Exrapolaion of Purchasing Power Pariies using Muliple Benchmarks and Auxiliary Informaion: A New Approach, The Review of Income and Wealh 56: Supplemen, pp. 59 S98. Rao, D.S.P., A.N. Rambaldi, and H.E. Doran (2011), An Economeric Approach o he Consrucion of Complee Panels of Purchasing Power Pariies: Analyical Properies and Empirical Resuls, School of Economics, Universiy of Queensland, S Lucia, QLD 4072, Ausralia. Summers, R. and A. Heson (1991), The Penn World Table (Mark 5): An Expanded Se of Inernaional Comparisons, , Quarerly Journal of Economics, Vol. 106, pp Szulc, B.J. (1983), Linking Price Index Numbers, pp in Price Level Measuremen, W.E. Diewer and C. Monmarquee (eds.), Oawa: Saisics Canada. World Bank (2013), Measuring he Real Size of he World Economy, Washingon D.C.: World Bank. 34 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

37 Oupu growh and inflaion across space and ime 1 Annex This Annex liss he underlying OECD daa, some supplemenary ables and noes. The source for all of he daa lised in his Annex is OECD.Sa. The counry price levels P n using domesic currencies (normalized o equal uniy in 2000) are lised in Table A1 and he corresponding volumes are lised in Table A2. Table A1: OECD counry price levels in naional currencies P n n P n 2000 P n 2001 P n 2002 P n 2003 P n 2004 P n 2005 P n 2006 P n 2007 P n 2008 P n 2009 P n 2010 P n 2011 P n Source: Auhor's calculaions EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 35

38 1 Oupu growh and inflaion across space and ime I can be seen ha Counry 18, Japan, had he lowes rae of domesic inflaion, which was acually a deflaion. Counry 32, Turkey, had he highes rae of domesic inflaion, which was 533 % over he sample period. The counries ha exhibied he fases raes of real GDP growh over he sample period were counries 5 (Chile), 27 (Slovak Republic), 32 (Turkey), 19 (Korea) and 25 (Poland) wih growh raes equal o 168 %, 167 %, 162 %, 159 % and 156 % respecively. Table A2: OECD Counry GDP volumes relaive o he corresponding 2000 volumes, Q n /Q n 2000 n Source: Auhor's calculaions In order o obain he counry volume levels Q n ha mach up wih he price levels P n in Table A1, he enries in he rows labeled 1-34 need o be muliplied by he counry volume levels for 2000, he Q n 2000 for n = 1,,34. These year 2000 levels are as follows: , , , , , , , 6.16, , , , , , 36 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

39 Oupu growh and inflaion across space and ime , , , , , , 22.00, , , , , , , 31.18, 18.57, , , , , , The unis are in billions of year 2000 domesic currency unis. Table A3 liss he counry n, year US dollar price levels for GDP, p n, and Table A4 liss he corresponding volume levels, q n. Noe ha p n q n equals v n, he year value of counry n s GDP in curren US dollar. Table A3: OECD counry price levels in US dollar a marke exchange raes p n n p n 2000 p n 2001 p n 2002 p n 2003 p n 2004 p n 2005 p n 2006 p n 2007 p n 2008 p n 2009 p n 2010 p n 2011 p n Source: Auhor's calculaions EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 37

40 1 Oupu growh and inflaion across space and ime Table A4: OECD counry GDP volumes in US dollar unis of measuremen q n n q n 2000 q n 2001 q n 2002 q n 2003 q n 2004 q n 2005 q n 2006 q n 2007 q n 2008 q n 2009 q n 2010 q n 2011 q n Source: Auhor's calculaions I is useful o spell ou in a bi more deail wha he US dollar period share weighs for counry n, S n v n / i =1 N v i look like in erms of domesic values and exchange raes. Le e n be he number of unis of domesic currency i akes for counry n o purchase one dollar of US currency in year. Then counry n s domesic currency value for is GDP in year, V n, is relaed o he corresponding US dollar value for year, v n by he equaions V n = v n e n for all and n. Similarly, he domesic currency price level for counry n in year, P n, is relaed o he corresponding US dollar price level for year, p n by he equaions P n = p n e n for all and n. Subsiuing hese equaions ino he definiion for he counry US dollar shares of OECD GDP, we find ha S n = [V n /e n ]/[ i =1 N (V i / e i )] for all n and. Noe ha hese shares do no depend on which counry is chosen as he numeraire currency. The counry GDP values in domesic currencies, V n, will move smoohly 38 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

41 Oupu growh and inflaion across space and ime 1 over ime bu exchange rae flucuaions will inroduce erraic movemens in he shares S n over ime. Table A5: Annual purchasing power pariies PPP n for OECD counries , naional currencies per US dollar n PPP n 00 PPP n 01 PPP n 02 PPP n 03 PPP n 04 PPP n 05 PPP n 06 PPP n 07 PPP n 08 PPP n 09 PPP n 10 PPP n 11 PPP n Source: Auhor's calculaions The exchange rae based Laspeyres and Paasche price indexes P L and P L ha appear in Table 2 are buil up using he following chain link indexes: P L,link n =1 N S n 1 (p n /p n 1 ) = n =1 N [V n 1 /e n 1 ] [(P n /e n )/(P n 1 /e n 1 )]/[ i =1 N (V i 1 /e i 1 1)] and P P,link [ n =1 N S n (p n /p n 1 ) 1 ] 1 = { n =1 N [V n /e n ] [(P n /e n )/ (P n 1 /e n 1 )] 1 ]/[ i =1 N (V i /e i )]} 1. I can be seen ha exchange rae flucuaions affec no only he domesic share weighs in he above expressions bu hey also inerac direcly wih he counry inflaion raes, P n /P n 1. Thus ineremporal exchange rae variaion noise will end o drown ou he counry inflaion rends. EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 39

42 1 Oupu growh and inflaion across space and ime Define he exchange rae based Laspeyres, Paasche and Fisher chain link volume indexes as Q L,link n =1 N S n 1 (q n /q n 1 ) = n =1 N S n 1 (Q n /Q n 1 ), Q P,link [ n =1 N S n (q n /q n 1 ) 1 ] 1 = [ n =1 N S n (Q n / Q n 1 ) 1 ] 1 and Q F,link [Q L,link Q P,link ] 1/2. I can be seen ha exchange rae flucuaions affec only he US dollar counry shares (he S n ) and no he growh raes of counry real GDP (he Q n /Q n 1 ) and hus hese volume link indexes will be much more sable han heir price counerpars. We noe ha he chained Fisher Q lised in Table 2 can be defined as follows: Q ; Q Q 1 Q F,link ; =2001,, EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

43 2 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure FRITS BOS ( 1 ) Absrac: This paper provides an overview of he roles and uses of naional accouns since he seveneenh cenury. I shows ha naional accoun saisics, including heir key-indicaors like GDP and GDP-volume growh, have mainly been developed for fiscal and moneary policy purposes, for measuring progress in maerial well-being and for showing heir relaionship wih he supply and use of goods and services in a marke economy. Naional accouns saisics were no developed for measuring welfare, qualiy of life, inequaliy or susainable developmen. This reflecs ha for such policy issues measuremen in moneary erms is less meaningful and more difficul o compile reliably. In order o mee he complicaed measuremen issues and many daa needs of he fuure, a sizable par of naional accouns saisics should be slow-saisics, i.e. compiled regularly bu no very frequenly, e.g. every wo or five years. To reduce misinerpreaion of GDP-volume growh saisics and o simulae susainable growh, such slow-saisics should also include well-being accouns and ne-ne saisics on domesic produc and naional income, i.e. including a correcion for he exhausion of naural resources. JEL codes: B00, E01, E60, Q56 Keywords: Hisory of naional accouning, guidelines on naional accouning, macroeconomic modeling and forecasing, measuremen in economics 1 ( 1 ) CPB Neherlands Bureau for Economic Policy Analysis.

44 2 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure 1. Inroducion The roles and uses of he naional accouns have developed over ime and have been simulaed by major evens, like he economic crisis in he hiries, he Second World War, European unificaion and he recen financial crisis. In his paper, an overview is provided of hese developmens in roles and uses over ime ( 2 ). Three differen periods are disinguished. The period is labelled as he early esimaes (Secion 2). This period of nearly hree cenuries sars wih he firs naional income esimaes in England and France. All hese early esimaes were incidenal and direced a concree policy issues. They were herefore no only he sar of measuring he naional economy, bu implied also he sar of quaniaive economic policy analysis. The second period is very shor, only wo decades: Bu hey were revoluionary decades for naional accouning (Secion 3). Major feaures of his ransiory period are: The sar of official and regular naional accouns saisics in a limied number of counries. These saisics were no direced a specific policy issues, bu were inended o provide informaion relevan more o general policy and analysis; Invenion of many new applicaions and uses of he naional accouns approach, e.g. he Keynesian revoluion in economic heory and policy, inpu-oupu analysis and economeric modelling; Revoluion in naional accouning conceps and mehods, e.g. he birh of he firs modern naional accouning sysems; Poliical and economic circumsances favouring a naional accouns approach in policy and analysis (economic crisis of he hiries, he end of he gold sandard, he Second World War and he need for recovery aferwards); We are sill living in he hird period (1950 presen). Considering heir very imporan role, i could be labelled he era of he inernaional guidelines (Secion 4). Major feaures of his period are; A rapid exension of he number of counries for which official and regular naional accouns saisics are available. This is accommodaed by he appearance of privae and public forecass of naional accouns saisics and heir conceps. Decision-makers and researchers become gradually accusomed o using naional accouns saisics; The publicaion of inernaional guidelines. This was very imporan for accumulaing and ransferring of knowledge on naional accouning. I was also imporan for harmonizing he conceps used by individual counries in compiling naional accouns saisics.; ( 2 ) The bes source on early naional income esimaes is sill Sudenski (1958); abou early English esimaes, see also Sone (1997). For he more recen developmens, see e.g. Bos (2009a, chapers 2, 3 and 4), Bos (2011), Vanoli (2005), Vanoli (2014) and Kenessey (1994). Maddison (2003 and 2005) provides a hisorical overview mainly resriced o he measuremen of economic growh. A major new insigh on naional accouns in Germany was provided by Tooze (2001). He suggess ha Wagemann s work during he Weimar republic was a pioneering effor in empirical macro-economics. Afer he Firs World War, Wagemann was head of he German Saisical Office and of he German Insiue for Business-Cycle Research. The Insiue esimaed and prediced he economic oulook for policy-makers based on an explicily macro-economic model of he economy using ime-series daa. This work was no menioned in he German hisory of naional accouns provided by Reich (1994). 42 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

45 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure 2 Major concepual developmens, e.g. abou how o measure prices and volumes. In fac, onlyrelaively recenly deailed guidelines on measuring prices and volumes in he naional accouns were published; The developmen of saellie accouns, e.g. on he environmen or healh care. This implies ha he naional accouns developed from no only a ool for macro-economic managemen bu also for such specific policy areas; Poliical and economic circumsances, e.g. globalisaion, European unificaion, he collapse of communism in Easern Europe and China, he financial crisis and he growing role for inernaional organisaions (IMF, World Bank, Unied Naions, European Union and OECD), favoured a harmonized naional accouns approach in policy analysis. Despie hese rends owards harmonizaion, here are sill subsanial differences beween he roles and uses of naional accouns saisics in counries all over he world and even wihin he European Union. These differences apply o he role in public decision-making, he relaive imporance of specific uses and he division of asks beween he naional accouns, official policy repors and research ouside he official naional accouns. This reflecs ha he roles and uses of he naional accouns depend also on he specific naional circumsances, e.g. he poliical and insiuional environmen. This issue is illusraed by he developmen of he Duch naional accouns and he ineracion wih he budge, moneary policy and naional exper insiues like he Duch Cenral Bank, he Neherlands Bureau for Economic Policy Analysis and he Duch Social and Culural Planning Bureau (see Bos (2006) and Bos (2008a), pp ). This paper serves various purposes. The firs purpose is o serve as an anidoe agains simple one-dimensional views on he naional accouns, e.g. naional accouns is GDP, GDP is no a good measure of welfare and should herefore no be used for policy-making and analysis. The second purpose is o inform economiss abou he imporance of naional accouning and abou he major recen developmens in naional accouning. This is necessary, as naional accouning has become much more complex and a profession separae from economic heory and economeric model building. Beer undersanding of he naional accouns may also simulae economiss o use naional accouns saisics and conribue o is developmen. The hird purpose is o sugges a redirecion of naional accouning owards is major uses. Wha are he major uses of he naional accouns in view of differen economic, poliical and insiuional circumsances? Wha is he link wih he various ypes of economic analysis? Wha is he link wih major policy debaes? Clarifying his can help o make naional accouns guidelines and (iner)naional saisical programmes much more balanced and effecive in view of hese uses. A he end of his paper (Secion 5), hree suggesions are provided o improve he relevance of naional accouns saisics in he fuure: More slow saisics for difficul o measure opics; Fewer rapid saisics ha do no conain much informaion; Beer explaining value added, use and misuse of naional accouns saisics. EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 43

46 2 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure 2. The early esimaes ( ) The esimaes of naional income and wealh by Pey, King and Davenan in England and Boisguilleber and Vauban in France sared Poliical Arihmeick (see Table 1). Table 1: Major evens in he early esimaes of naional accouning Year Even Firs naional income esimaes; in England by Pey, King and Davenan; in France by Boisguilleber and Vauban 1707 Firs index-numbers by Fleewood 1760 Tableau économique by Quesnay: economic accouns used as a primiive growh and general equilibrium model; precursor of inpu-oupu ables 1770 The concep of value added invened by Young Firs naional income esimaes in Russia Firs naional income esimaes in he Neherlands 1805 Firs naional income esimaes in Germany 1823 Firs naional income esimaes in consan prices by Lowe 1843 Firs naional income esimaes in he USA 1886 Firs official naional income esimaes by he governmen (Ausralia, Coghlan) Firs naional income esimaes in Ausria, Ausralia, India and Greece 1915 W.I. King (USA): one of he las naional income esimaes combined wih clear policy conclusions 1920 The economic consequences of he peace by Keynes: using naional accouns saisics o assess he dramaic economic consequences of a major poliical agreemen Privae insiuions sar publishing naional income sudies, e.g. universiy insiues in Sweden and Norway, USA: Brookings Insiuions, NBER, Naional Indusrial Conference Board; Ausria WIFO More official naional income esimaes (e.g. Greece, Canada, Sovie Russia, Germany, Neherlands, New Zealand, USA and Turkey) 2.1. England A he end of he seveneenh cenury, Pey waned [firsly] o prove mahemaically ha he Sae could raise a much larger revenue from axes o finance is peace and warime needs, and ha i could do so by more equiable and less burdensome forms of axaion... Secondly, Pey waned o disprove once and for all he noion ha England had been ruined by he Revoluion and foreign wars and was no mach, eiher miliarily or commercially, for Holland and France (Sudenski (1958), p. 27, 28). King also draws clear poliical conclusions from his esimaes: he Warr canno well be susain d beyond he year 1698 upon he Foo i now sands, unlesse (1) The Yearly Income of he Naion can be Increas d. (2) Or he Yearly Expence Diminish d. (3) Or a Forreign or Home Credi be Obain d or Esablish d. (4) Or he Confederacy be Inlarg d. (5) Or he Sae of he Warr Aler d. 44 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

47 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure 2 (6) Or a General Excise, in effec Inroduced. (King (1936), p. 47). Neverheless, Pey as well as King recognised also he more general advanages of esimaing naional wealh and income. King saes ha informaion on a counry s wealh and populaion is a Piece of Poliical Knowledge, of all ohers, and a all imes, he mos useful, and Necessary (King (1936), p. 13). Pey wen even furher by advocaing ha in socio-economic discussions no word migh be used bu wha marks eiher number, weigh, or measure (Sudenski (1958), p. 27). The esimaes by King can be regarded as improvemens on hose of Pey. We will shorly discuss he four main feaures of King s esimaes. Like he esimaes of Pey and he earlies esimaes in France, King employs a comprehensive concep of producion and income. This concep is sill used in he curren inernaional guidelines. According o his concep he producion of goods as well as services generaes income. The second imporan feaure of he esimaes by King is ha hey already represened he hree ways of esimaing domesic produc: ne producion, expendiure and disribuion by ype income. The esimaes of Pey and mos esimaes up ill he 1930 s only covered one or wo ways. The hird imporan feaure of he calculaions by King was heir remarkable coverage. He presened no only he oal annual naional income, expendiure, and saving, bu also heir disribuion by social and occupaional groups, a breakdown of naional income by ype of income and an esimae of wealh (gold, silver, jewels, furniure, livesock, ec.). Like Pey, King provided a comparison of he naional incomes and wealh of England, Holland and France. Inernaional comparison, which is a major objecive of he inernaional guidelines, was herefore already presen in Pey s and King s pioneering esimaes. King s esimaes conained also ime series of he period of naional income, expendiure and axes received. The fourh imporan feaure of King s work is ha he used his ime series o forecas income, expendiure and ax revenue for he years 1696, 1697 and This ype of use of naional accouning figures daes herefore also back o he earlies esimaes of naional income. From a modern perspecive, he major limiaion of King s work is ha all esimaes are in curren prices. The firs price index numbers were invened only some years laer by Fleewood (1707) ( 3 ). However, he idea of deflaing naional income is much more recen and originaes from Lowe in 1823 (see Sudenski (1958), pp ). Lowe used his naional income esimaes o calculae he ax burden. He relaed hem o he axable income remaining afer deducing he subsisence incomes ha could no bear any axaion. He did his for he naional incomes of specific years in he preceding hree decades, calculaed firs a curren prices and nex a prices of Lowe devoed considerable space o he effecs of inflaion on fiscal policy. He was paricularly concerned over he increase in he burden of public deb on he shrinking money value of naional income. He proposed ha he governmen enac a abular sandard ha would conrol he value of money and he paymens ( 3 ) The issue addressed by Fleewood was he inerpreaion of values in old saues, enilemens, laws and regulaions (see Sone (1997), ). For example, according o fifeenh cenury college saues in Oxford, a fellow should vacae his fellowship if he came in possession of an esae by inheriance or of a perpeual pension worh of 5 pounds per year. According o Fleewood, correcions should be made for all price changes during he pas wo cenuries. He also menioned similar problems wih he fines for sealing menioned in old laws and he level of income which eniled o voe in parliamenary elecions. To ranslae such figures ino curren prices of several cenuries laer, he provided an overview of price changes for five commodiies (whea, oas, beans, ale and cloh) and recommended o calculae an unweighed index of average price change. EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 45

48 2 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure of ineres and wages. This abular sandard was o be based on he power of purchase of money in erms of basic commodiies and could be applied on a volunary basis. For calculaing naional income, King used ne produc esimaes for agriculure, i.e. he deduced from he revenues of he crop he expendiure on he seeds. However, he did no sysemaically use value added as a concep. This concep originaes from Young (1770, see Sone (1997), pp ). For calculaing value added in agriculure, Young did no only deduc he coss of seed, bu also coss of mainenance and repair, e.g. of buildings, vehicles and horses France In France, a he sar of he eigheenh cenury, Vauban proposed a major ax reform (see Sudenski (1958), pp ). Mos of curren axes were o be abolished, as hey were inequiable he small incomes carried nearly all he ax burden and responsible for povery. They should be replaced by an income ax, levied a a uniform rae of 5 % o 10 %, depending on general economic condiions. The income ax should consis of wo pars: a levy in kind on agriculural produce and a levy on all money incomes, i.e. on he ren of houses, profis of businesses and grain mills, he operaion of public properies, wages, pensions and he fees of governmen offices. In addiion o he income ax, curren axes on sal, posal charges, cusoms duies and axes on luxury goods like ea, coffee, chocolae, brandy and gil coaches should be reained and a ax on wine, cider and beer consumed in public houses should be added. He hen esimaed he poenial revenue of his ax reform. To his end, he made an esimae of he major elemens of French naional income, bu did no presen an esimae of oal French naional income. He was also no aware ha disinguishing beween he gross and ne value of agriculural producion is necessary for ensuring fairness for ax purposes. Vauban was a miliary engineer world famous for his forificaions and acics for aacking and defending hem. He was decoraed by he French king Louis XIV wih every conceivable order of disincion. He made his proposals for ax reform afer his reiremen. However, hese proposals were no all appreciaed by he ruling class. A he cour, Vauban fell herefore ino disgrace and he died some monhs laer. Abou half a cenury laer, Quesnay published his zigzag-diagram, which is a major precursor of modern inpu-oupu ables. Is purpose was o consruc a fundamenal Tableau of he economic order for he purpose of displaying expendiure and producs in a way which is easy o grasp, and for he purpose of forming a clear opinion abou he organizaion and disorganizaion which he governmen can bring abou (Translaion by Meek (1962), p. 108). Quesnay s Tableau économique was used as a primiive growh model ha served o promoe he idea ha agriculure and no merchandising or manufacuring is he engine of economic growh ( 4 ). According o he physiocras, invesmens in agriculure should herefore be promoed, axes and ineres raes for agriculure should be reduced and olls and oher resricions on rade in agriculural producs should be abolished. The able also served o argue ha he mos efficien way of axing is o direcly ax he group ha ulimaely pays he ax, i.e. he landlords insead of he farmers or he arisans ( 5 ). The able was herefore also a simple general equilibrium model. ( 4 ) On he inerpreaion of Quesnay s Tableau économique as a growh model, see Elis (1984), in paricular he firs wo chapers. ( 5 ) However, he physiocras were cerainly no enemies of he landlords, as hey also argued ha he ne surplus of agriculure should be sufficienly high (e.g. by raising he prices of agriculural producs) o pay he axes and o give hem sufficien income. 46 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

49 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure 2 Quesnay s Tableau économique is a clear economic accouning model. As such i can be regarded as he firs precursor of boh he inpu-oupu ables and he secor accouns. The esimaes by King, Pey and Vauban were sysemaic, bu did no sress he circular flow of income and expendiure and he ineracions beween socio-economic groups The Neherlands Poliical arihmeic had a surprisingly slow sar in he Neherlands ( 6 ), by far he riches counry in he world ( 7 ) in he seveneenh cenury. In 1798, on reques of he naional assembly of he new Baavian republic, Hora Siccama and van Rees made a plan for revising he ax sysem (Hora Siccama and van Rees (1789)). Their ask was o invesigae how axes could be levied efficienly, in proporion o personal wealh, wih as low as possible ax raes ha were sufficien for financing governmen expendiure. Esimaes of naional income, governmen expendiure and curren and poenial ax revenue were included in he plan. The plan, including a proposal for inroducing personal income and wealh ax, was no execued due o a revol and a change in governmen. In 1804, Meelerkamp published a sudy of 600 pages on he economic and miliary power of he Baavian republic over he las fify years and in comparison wih seven oher counries (Grea Briain, France, Russia, Prussia, Sweden, Denmark and Saxony). The sudy was published in Duch and also ranslaed in French (Meelerkamp (1804)). He compared he size of he counry, he number of inhabians, naional income, naional wealh, impors and expors, income and oulay of he governmen, governmen deb and he number of soldiers and warships. However, he comparison of naional income was limied o he Neherlands (including Belgium), Grea Briain, France and Saxony. Meelerkamp s sudy was a reacion o he pessimisic mood in he Neherlands: people hough heir counry was in decline, full of deb, losing morale and fighing spiri and wih shrinking revenue from inernaional rade and colonies (Meelerkamp, 1804, Inroducion, par I). He waned o invesigae wheher facs corroboraed hese feelings and opinions and wha policies should be pursued. He concluded ha: The example of Russia shows ha governmen should inerfere as lile as possible wih domesic rade (huning, fishing, farming, crafsmanship and commerce); his simulaes employmen, income and he supply of producs; Duch welfare did no decline during he pas fify years; some sources of income declined indeed, bu were compensaed by he increase in oher sources of income; Welfare in oher European counries increased. This implied a relaive decline of Duch welfare. However, Duch welfare sill surpassed welfare in oher European counries; During he las cenury, governmen revenue increased in oher European counries much more han in he Neherlands. Relaively low axes may have been he reason for he much more rapid increase in Duch governmen deb. However, he ax burden in he Neherlands was now excessive and would in he long run cerainly ruin he economy. ( 6 ) On he hisory of he Duch naional accouns, see Bos (2006) and Bos (2008a). ( 7 ) In 1700, Duch GDP per capia was 70 % higher han ha of England and 130 % higher han ha of France (see Maddison (2003), p ). EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 47

50 2 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure 2.4. Major feaures of he early esimaes All early esimaes of naional income were pracical and direced a concree policy issues; his was a common feaure of naional income sudies up o he 1920 s. The policy issues addressed were naional economic power and performance, povery, unfair and inefficien axaion and susainabiliy of public finance. Ofen, several of hese issues were discussed and he naional accouns approach was essenial for demonsraing ha he various issues were inimaely linked. War, subsanial economic decline and wide-spread povery were circumsances ha simulaed he early esimaes. Also he availabiliy of census daa or income ax daa was imporan; his parly explains he dominan role of he English esimaes. The ruling class was ofen no very happy wih he naional income esimaes and he accommodaing proposals for reform. Early naional accounans were someimes exiled (e.g. Radishchev in Russia) or fell ino disgrace (e.g. Vauban in France); ohers may have feared he consequences of publishing heir work and lef i herefore unpublished. However, in he wenieh cenury -probably linked o advancemens made in democracy- esimaing naional income became gradually o be perceived as a ask of he governmen. In many counries, also privae insiuions ook he responsibiliy of regularly compiling naional accouns saisics and producing naional income sudies. Examples are universiy insiues (e.g. in Sweden and Norway), economic research insiues (e.g. NBER, Brookings Insiuion and Naional Indusrial Conference board in USA, Misubishi Economic Research Insiue in Japan and WIFO in Ausria) or privae and cenral Banks (e.g. he Bank of Nova Scoia in Canada). Such insiuions reflec he ineres of researchers, business, banks, pension funds and rade unions in naional income esimaes. They need such informaion for analyzing and monioring he economy, calculaing marke shares and deciding on invesmen, loans, morgages and wage negoiaions. Naional accouning had a brillian sar wih he work by Pey and King. Since hen, progress in naional accouning was ofen slow and small and here were major cases of regress, like he producion boundary used. For hree quarers of a cenury, Adam Smih was very influenial in his argumen ha labourers in agriculure as well as in manufacuring, commerce and he ransporaion of goods were o be regarded as producive. However, unlike King, he sill raed he whole civil and miliary personnel of governmen, he professions, he domesics, and ohers engaged in he performance of personal services and he services of dwellings as unproducive labourers. He considered he naional produc o be consiued solely of commodiies, and he naional income... o be composed of wages, ren and profi (including ineres) derived from he producion of hese aricles (Sudenski (1958), p. 19). Smih s view was suppored by among ohers Ricardo, Malhus, James Mill and John Suar Mill, bu became increasingly subjec o criicism by, e.g., Say, McCulloch, Senior, Walras and Marshall. Despie such unforunae inellecual deours, a he beginning of he wenieh cenury, he common sock of knowledge on naional accouning had already become considerable. I included for example a comprehensive producion boundary reaing e.g. agriculure and governmen services as producive, he noion of hree basic ways o esimae domesic produc and he conceps of value added and consan prices. 48 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

51 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure Keynes: The economic consequences of he peace Direcly afer he firs world war, Keynes wroe The economic consequences of he peace (Keynes (1920)). This polemic made him insanly world-famous. He was he official Briish represenaive a he sar of he Paris Peace Conference, bu resigned when i became eviden ha here was no hope for subsanial modificaion of he draf erms of peace. He aacked he peace-reay and he reparaion paymens imposed on Germany for no aking ino accoun facs and economic logic. On he basis of all kinds of saisics on naional wealh in e.g. Belgium, France, Grea Briain and Serbia, he esimaed ha he reay implied ha Germany had o pay he various allied powers 2 billion pound for direc war damage ( 8 ) and 5 billion for pensions and allowances. He also calculaed ha Germany s capaciy o pay was no sufficien for his. Transferable wealh (e.g. gold, silver, ships and foreign securiies) was only abou 0.5 billion pound and he res had herefore o be paid over a number of years ou of he ne revenue from expors. However, considering he loss of major pars of former Germany (e.g. Alsace-Lorraine) and he loss in capial sock (e.g. ships, livesock, wear and ear due o lack of repair), he ransferable par of naional income would be less han 0.1 billion pound per year. He proposed o limi Germany s reparaion paymens well wihin her capaciy o pay, o ensure a fair disribuion of coal and iron and o esablish a free rade union under he auspices of he League of Naions. Only in such a way, he European economy could be revived, avoid he perils of mass inflaion and ensure a proper sandard of living for he whole populaion. Keynes polemic could be regarded as a revol of economics agains poliics and in his revol naional accouns saisics played a major role. I is sill a major example of he use of naional accouns saisics for (economic) policy analysis. I marked a radical shif in Keynes hough and is propheic lessons were learned afer he Second World War, e.g. he Marshall aid and he European Union. All hese developmens urned ou o be very imporan for naional accouning (see Secions 3 and 4). 3. Revoluionary decades ( ) 3.1. Major changes in uses In he period , naional accouning was drasically ransformed. I was a revoluion in erms of he use of he naional accouns (see Table 2). Colin Clark (see e.g. Clark (1937 and 1940)) inroduced purchasing power pariies and showed how o make inernaional comparisons of real income. He also demonsraed how o make ineremporal comparisons. He even made a comparison beween he level of well-being in he ancien world (Egyp, Greece and he Roman Empire a he peak of heir powers) and ha in he nineeenh cenury and he firs half of he wenieh cenury! On reques of he US Senae, Simon Kuznes ( 9 ) measured he impac of he financial crisis of 1929 on he US Economy (Kuznes (1934)). He made esimaes of naional income for he period broken down by ype of income (labour income, propery income and ( 8 ) Considering ha much guesswork was included, he argued ha he amoun would cerainly be beween 1.6 billion pound and 3 billion pound. ( 9 ) Abou Kuznes and his work, see e.g. Carson (1975) and Lundberg (1984). EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 49

52 2 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure enrepreneurial income) and by indusry (e.g. agriculure, mining, manufacuring, ec.). These esimaes showed a conracion of %. Taking ino accoun he decline in wholesale prices, his indicaed ha he volume of naional income was reduced by % in he period He reconsruced naional income and produc accouns for he USA firs back o 1919 and evenually back o Such impressive measuremen exercises were he inpu for invesigaing business cycles and long erm economic growh. For example, wha was he role of he various indusries, wha was he role of echnology and innovaions, wha is he relaionship beween economic growh and inequaliy (he U-shaped Kuznes-curve) or beween economic growh and urbanizaion? Table 2: New applicaions and uses of he naional accouns during New applicaion or use Who? Purchasing power pariy: inernaional comparison of real income Clark (1937 and 1940) Sysemaic analysis of long erm growh by using naional accouns ime series Kuznes (1934) Inpu-oupu analysis Leonief (1936) Economeric modelling of naional economies Frisch (1933), Tinbergen (1936) Keynesian revoluion and he birh of macro-economics Keynes (1936) Analysing public finance in a macro-economic framework Meade and Sone (1941) Moneary policy linked o naional income insead of gold & inflaionary gap analysis Keynes (1940) Analysing balance of paymens in a macro-economic framework Meade (1951) Leonief (see e.g. Leonief (1936)) developed inpu-oupu analysis, esimaed deailed inpuoupu ables for he USA and applied his new ool o all kinds of problems. For example, he esimaed he resource coss of conversion o peace ime producion in 1945 and by calculaing he relaive facor inensiies of impors and expors he discovered he Leonief paradox: why are American expors labour inensive, while American impors are capial inensive? Tinbergen and Frisch ( 10 ) pioneered in economeric model building covering he whole naional economy (Morgan (1990)). These models need naional accouns daa as inpu and can be used o forecas he naional economy or o analyse he economy or alernaive policy proposals. As a consequence, he developmen of macro-economeric model building was a srong simulus for he developmen of naional accouns saisics. For example, in 1936, Tinbergen consruced he firs economeric model of he business cycle covering he whole economy (Tinbergen (1936)) o assess he bes policy measure o improve he business cycle: devaluaion, public works or reducing wages. His conclusion was ha devaluaion was he bes policy. This was very much agains he poliics of he ime, i.e. no only agains he official policy of he governmen bu also agains he opinion of he socialis pary. Neverheless, in Sepember 1936, he guilder was devalued wihou he disasrous consequences ha had been prediced by many. This conribued enormously o he repuaion of Tinbergen and his new mehod. Tinbergen s revoluionary model of 1936 was no based on a naional accouning scheme, e.g. conceps like naional income and final consumpion by he governmen were absen. Neverheless, i gave he developmen of naional accouning in he Neherlands a head sar (Bos (2008a)). In order o provide a beer empirical grounding o he economeric model, new and longer ime series were needed and he qualiy of exising esimaes was o be improved ( 10 ) On he impac of Frisch on naional accouning in Scandinavian counries, see Aukrus (1994). 50 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

53 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure 2 Keynes published his General Theory in 1936 (Keynes (1936)). This launched he Keynesian revoluion and gave birh o macroeconomics. This revoluion in economic heory had an enormous impac on naional accouning. The Keynesian ype of analysis esablished a direc link beween economic heory and naional accouning as boh came o use he same macroeconomic ideniies. The Keynesian ype of analysis also hrew a new ligh on he role of he governmen: a new responsibiliy for sabilising he economy was added. Accouning for his role of he governmen became necessary for economic policy analysis. This induced he inroducion of accouning per secor, in paricular he inroducion of a governmen secor. As a consequence of he Keynesian revoluion, he imporance of naional accouning figures for economic heory and economic policy increased and was more widely recognised. On Keynes insigaion, in 1941 Sone and Meade prepared UK esimaes on naional income and expendiure (Meade and Sone (1941)). These esimaes were used o presen governmen expendiure and revenue as par of a sysem of balanced ables describing he whole naional economy. In his way, hey became a ool in planning he Briish war economy (Sone (1951), p. 84; Painkin (1976), p. 1109). I also implied a revoluion in he pracice of governmen budgeing: since hen, in mos counries he governmen budge is presened in a macro-economic framework wih explici saisics or forecass abou economic growh and inflaion ( 11 ). In he hiries, counries lef he gold sandard. As a consequence, moneary policy needed a new anchor and was linked o naional income. In order o avoid excessive inflaion, he supply of money should grow in line wih he nominal growh of domesic produc correced for changes in he velociy of circulaion () 12. During and direcly afer he Second World War, several counries (e.g. USA, UK, France and he Neherlands) used he naional accouns o esimae he inflaion gap and invesigae he size and effeciveness of various policy measures o be aken o avoid massive inflaion, e.g. resricions on he use of bank deposis or a big incidenal capial ax. These sudies include also Keynes famous repor How o pay for he war? (Keynes (1940)) ( 13 ). In 1950, Meade wroe his book The balance of paymens (see Meade (1951)). In he firs hree chapers, he basic conceps of he balance of paymens are clarified and linked o hose in he naional accouns. The imporance of he various ypes of consisency is sressed. For example he consisency beween ne expors and a counry s indebedness o foreigners, beween he balance of paymens of various counries or beween he balance of paymens and he naional accouns, e.g. privae and public expendiure and saving, moneary circulaion and inflaion. In he remainder of he book, he used hese accouning ideniies o analyse he consequences of various ypes of moneary and fiscal policy on he balance of paymens. His exposiion of he balance of paymens and he naional accouns is sill he basis of modern ex book exposiions on inernaional economics (see e.g. Krugman and Obsfield, 2011). Mos of hese new uses also reinforced each oher. These uses were also closely linked o he economic circumsances: he economic crisis of he hiries, he Second World War and he need for recovery aferwards simulaed an acive role of he governmen. Naional accouns saisics urned ou o be very useful in such circumsances for analyzing, monioring, forecasing, discussing and planning he naional economy. ( 11 ) However, in he official budge of he UK his relaionship was only inroduced in 1967 (see Jones (2000)). ( 12 ) This follows from Fisher's equaion of exchange MV =PT in which M is money supply, V is velociy of circulaion (he number of imes money changes hands), P is average price level and T is he volume of ransacions of goods and services (see Fisher (1911)). ( 13 ) This repor also simulaed he work by Meade and Sone (1941), see above. EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 51

54 2 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure 3.2. Major changes in conceps and compilaion mehods These decades were also a revoluion in erms of he developmen of naional accouning conceps and compilaion mehods. Boh revoluions were no independen: hey reinforced each oher and ofen also he same persons were involved. The works by Clark and Kuznes consised of profound and deailed esimaes ha were accompanied by elaborae moivaions of he conceps and saisical mehods used. In he firs chaper of Naional Income and Oulay (Clark (1937)), Clark expounds he purposes of naional income measuremen and is basic conceps. Examples of he laer are his discussion of he inclusion of he services of owner-occupied dwellings, he exclusion of he services of consumer durables, he exclusion of holding gains and losses and a possible deducion for any demonsrable exhausion of naural resources (Clark (1937), p. 9). Kuznes had also a famous discussion wih Hicks in Economica on subjecs like he relaion beween changes in naional income and welfare, he valuaion of governmen oupu and he concep of inermediae and final produc (Hicks (1940 and 1948), Kuznes (1948b)). Commodiy-flow accouning can be regarded as he saisical counerpar of inpu-oupu analysis, as commodiy-flow accouns are a ype of inpu-oupu able. Commodiy-flow accouning was more or less simulaneously developed in Sweden (Lindahl e al (1937)), Denmark (Kampmann (1942)) and he Unied Saes (Kuznes (1938)). Clark and Kuznes did no play a role in developing naional accouning sysems, i.e. a sysem in which secors as well as accouns are used in presening daa. Kuznes saw i even as a dubious addiion o he heoreical equipmen (Kuznes (1948a), p. 154)! The developmen of naional accouning sysems occurred simulaneously in Briain, he Neherlands and he Scandinavian counries. In 1939, he League of Naions had requesed a repor wih guidelines in order o improve inernaional comparabiliy of naional accouning figures, bu he war delayed he progress on he repor. In Sepember 1944, represenaives of he UK, he USA and Canada me in order o exchange views... and, if possible, o bring abou uniformiy in erminology and he reamen of conroversial iems (Denison (1947), p. 3). As a resul of his meeing, he naional accouns of he Unied Saes and Canada were revised, which made hem more compaible wih he Sone/Meade proposals of 1941 and he Briish naional accouns (see Carson (1975), p. 177). Immediaely afer he war, in December 1945, consulaions on he Unied Naions repor were resumed. This ime also represenaives from counries occupied during he war by Germany, like he Neherlands and Norway, could be presen. The repor was published in 1947 by he UN (UN (1947)) and consised mainly of an appendix by Sone (Sone (1947)). This appendix can be regarded as he firs fully worked ou and deailed naional accouning sysem (see Aukrus (1986) and Carson (1975), p. 178). Furhermore, he repor was also pahbreaking in ha i conained for he firs ime inernaional recommendaions on naional accouning. However, he repor should no be regarded as he firs official guidelines of he Unied Naions: i was no approved by he saisical commission of he Unied Naions as official guideline, bu was referred o as a useful echnical repor; i was also no used as a guideline for submiing daa o he Unied Naions. 52 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

55 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure 2 4. The era of he inernaional guidelines ( ) 4.1. The successive guidelines ( 14 ) On reques of he Organisaion for European Economic Co-operaion(OEEC), in 1951 a guideline on naional accouning was wrien which was o be used in planning he Marshall-aid. This guideline and is wo immediae successors (OEEC (1952) and UN (1953)) can be regarded as he firs generaion of official inernaional sandards. In conras o he 1947 UN repor, raher simple accouning sysems were aimed a. In fac, a sysems approach was nearly absen as only some aggregaes and heir composing pars are o be compiled; he financial flows in he naional economy are even nearly fully ignored. Table 3: The successive guidelines on naional accouning Year Even Technical repor by he UN conaining recommendaions; including he famous annex by Sone: he firs deailed and fully worked naional accouning sysem Firs generaion of inernaional guidelines: OEEC guidelines of 1951 and 1952; UN guideline of 1953 (SNA53); very simple ables and accouns Second generaion of inernaional guidelines: UN guidelines of 1968 (SNA68), he European guidelines of 1970 (ESA70) and he Maerial Produc Sysem of 1969 (MPS69) for communis counries Third generaion of inernaional guidelines: join guidelines of 1993 by he inernaional organizaions (SNA93 by UN, IMF, World Bank, OECD and EC) and he European guidelines of 1995 (ESA95) Fourh generaion of inernaional guidelines: updaes of he join and European guidelines (SNA 2008 and ESA 2010) A second generaion of official guidelines was issued a he end of he sixies and beginning of he sevenies. I consised of wo guidelines by he UN: he Maerial Produc Sysem (MPS69) used by communis counries and he Sysem of Naional Accouns of 1968 (SNA68) used by he res of he World. For he special purposes of he European Communiies, also separae guidelines were issued for EC-counries, i.e. he European Sysem of economic Accouns of 1970 (ESA70). These are (almos) fully aligned wih he SNA, bu conain specific EU-deail. The SNA68 and ESA70 followed mainly he rails se by heir predecessors bu grealy expanded he accouning sysem, e.g. by including also inpu-oupu ables and consan prices. The MPS69 differed fundamenally from all he oher guidelines by is anachronisic concep of producion ha focuses on measuring only maerial producion ( 15 ). Neverheless, wih respec o he oher conceps and is scope, he MPS69 was someimes even more advanced han he SNA68 and ESA70 (see Arvay (1994)). The major example is he noion of oal consumpion of he populaion. In he SNA93 and ESA95, his is known as household acual individual final consumpion, i.e. he aggregae of household final consumpion expendiure plus social ( 14 ) A deailed comparison of he successive UN-guidelines (excluding he mos recen one) can be found in Bos (1993) and Bos (1994). ( 15 ) Sudenski (1958), p. 22, argues ha he producion boundary in MPS69 is based on he ideas of Marx, and more in paricular on a misaken inerpreaion of i. However, also Adam Smih advocaed a maerial producion concep (see Secion 2). So, i could be argued ha he communis guidelines were no acually based on he ideas of Marx, bu on hose of he inellecual faher of capialism, i.e. Adam Smih! EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 53

56 2 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure ransfers in kind governmen and non-profi insiuions, e.g. wih respec o educaion and healh care. In 1985 Richard Sone was awarded he Nobel prize in economics for his conribuions o he naional accouns, in paricular for his leading role in he developmen of he firs and second generaion of inernaional guidelines. Table 4: Major changes in he scope of he successive universal guidelines Scope SNA53 SNA68 SNA93 SNA 2008 Scope Simple se of ables and accouns in curren prices Exended accouning sysem, including inpu-oupu ables, general principles on prices and volumes and financial accouns Inclusion of balance shees, employmen and purchasing power pariies. More deailed accouning srucure (more accouns, more sub-secors and deailed supply and use ables). Separae chapers on saellie accouns and flexible adjusmens for naional circumsances. Deailed discussion of general principles on prices and volumes (e.g. chaining and index formulae) More deailed discussions of many opics, e.g. governmen accouns, he informal secor and capial services (imporan for produciviy measuremen). Bu no deailed discussion of price and volumes by indusry/produc and no separae chapers on quarerly naional accouns and regional accouns (unlike ESA95 and ESA 2010) In he mid nineies, a hird generaion was issued, he SNA93 and ESA95. The SNA93 and ESA95 have again grealy expanded he scope of he inernaional guidelines by including e.g. balance shees, purchasing power pariies, saellie accouns and Social Accouning Marices. A revised version of he MPS69 was no necessary due o he collapse of communism in Easern-Europe. All counries in ransiion in Easern Europe implemened he SNA93 and, for hose who applied for membership of he EU, he ESA95. In recen years, a fourh generaion was published, he SNA 2008 and ESA Basic conceps were hardly changed ( 16 ). Much aenion was paid o measuring he growing complexiy of economic realiy, e.g. all kinds of new financial insrumens, mixes of public and privae insurance and an even more global producion process. By adding several new chapers, e.g. on governmen accouns, links o moneary and financial saisics, res of he world accoun, quarerly accouns, saellie accouns, European accouns and regional accouns, he focus has been shifed subsanially oward a much beer link wih he major uses of he naional accouns. The old guidelines focused on merely presening a basic naional accouns framework in erms of secor accouns and inpu-oupu ables. Nex o he official guidelines on naional accouns, over he years many oher repors, handbooks and guidelines have been published on specific pars of he naional accouns, for example on price and volume measuremen, seasonal adjusmen, environmenal-economic accouning and governmen deb and defici. A very welcome novely is ha also he users perspecive (how o use he naional accouns for analysis?) has begun o be addressed, e.g. by he EU-KLEMS-projec for deailed and inernaionally comparable produciviy daa and analysis by indusry (Jäger (2016)), by he repor on naional accouns and policy analysis (UN (2002)) and by an inroducory book on naional accouning by Lequiller and Blades (2014). ( 16 ) There were some specific changes, e.g. including also R&D as capial formaion. 54 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

57 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure 2 By heir rapid expansion of scope since he guidelines of he fifies, he mos recen se of guidelines have incorporaed mos of he major innovaions in naional accouning since he Second World War. If we disregard he MPS69, basic conceps in he guidelines have shown a remarkable consisency, e.g. excluding he services of unpaid household services. Some of he criicism on unchanged basic conceps has been me by inroducing saellie accouns, e.g. on he link beween environmenal indicaors and naional accouns saisics. Neverheless, some imporan changes in concep have also occurred, e.g. he inroducion of chain indices or he expansion of he concep of capial formaion o include sofware, R&D and mineral exploraion. For decades, he inernaional guidelines on naional accouns saisics were parly inconsisen wih he inernaional guidelines on hree specific ypes of macro-economic saisics: balance of paymens, governmen finance saisics and saisics on employmen and populaion. However, nearly all hese inconsisencies have now been resolved. Also he links wih inernaional saisics on specific policy areas, like healh care, social securiy and environmen, have been improved. Inernaional guidelines on hese specific saisics explicily refer o he naional accouns conceps and classificaions, discuss similariies and differences and propose saellie accouns o bridge he gap beween he naional accouns and hese specific policy areas The role of he inernaional guidelines Inernaional guidelines have been influenial for many reasons. Firsly, he leading inernaional expers of he profession have developed he sysems in he inernaional guidelines. They are herefore relaively well hough ou and i is cosly, ime consuming and no easy o inven an alernaive sysem. Secondly, by keeping in line wih he inernaional guidelines, naional figures can be compared wih figures from oher counries. This is imporan, as inernaional comparison is a major use of naional accouning figures. Thirdly, in many counries, he naional accouns have been se up by or improved wih help from he inernaional organisaions issuing he guidelines (UN, OECD, EU) or wih help from counries advanced in naional accouning (Sweden, France). In he laer case, following he inernaional guidelines is usually simulaed o he exen ha he helping counries follow hem. Fourhly, we menion ha all counries are obliged o compile some figures on he basis of he inernaional conceps. In he EU, due o some imporan adminisraive uses, he guidelines are even legally binding; he same applies o he saisical programme linked o hese guidelines. The fifh reason is ha he daa submied o he inernaional organizaions play a cenral role in inernaional policy discussions and decision-making, e.g. abou accession o he European Union, graning loans or paying conribuions o he inernaional organizaions. To link naional discussion and decision-making o his inernaional conex, he inernaional conceps have o be used for naional purposes as well. The inernaional guidelines are very successful in sandardising he conceps and classificaions used in compiling naional accouns figures. The achievemen of he guidelines is ha all over he world official figures came o be based on uniform noions of he producion boundary, asse boundary, he disincion beween inermediae and final consumpion, ec. This is evidenced by some of he earlier counry pracices. They all differed fundamenally from he basic conceps in he successive guidelines. In Sweden in 1937, Lindahl published wo alernaive esimaes of naional produc, one including he services of unpaid household services and one excluding hem (Lindahl e al EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 55

58 2 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure (1937)). However preference was given o he laer. In Norway in 1946, he value of unpaid household services was included in oupu and naional produc. However, since 1951 hey are excluded (see Aukrus, 1994). In some official Scandinavian sudies (1937, 1951, 1953), he services of consumer durables like cars were included in oupu (see Aukrus (1994)). In France, unil 1975 when he ESA70 was implemened in he French naional accouns, he value of he oupu by banks, insurance companies and general governmen was no included in oupu and domesic produc (1971 base, see Demoes-Mainard and Bournay (1994)). Due o he influence of he inernaional guidelines, counry pracice o include in he naional accouns esimaes of unpaid household services and he services of consumer durables was gradually exinguished. However, as he French case shows, some drasic differences beween counry pracice and he inernaional guidelines exised uno he sevenies. Furhermore, if we look a he changes in he inernaional conceps and classificaions, some imporan differences coninued o exis for decades, as counry pracice ofen suck o old conceps and classificaions or coninued wih ypical naional versions of he naional accouns. For example, in he USA, he naional income and produc accouns and flow-of-funds accouns did no comply fully wih he previous inernaional guidelines (SNA1993), e.g. expendiure on miliary weapons sysems was reaed as gross fixed capial formaion. In 2006, he USA s Bureau of Economic Analysis (BEA) launched a se of Inegraed Macroeconomic Accouns based on he SNA 2008 and inegraing he informaion from he naional income and produc accouns and he flow-of funds accouns and adding various oher pieces of informaion. Even in his se of accouns, here are several clear differences wih he SNA 2008, like governmen enerprises included in he governmen secor, bu hey do generally no affec GDP (see Yamashia (2013) and McCulla e al (2015)) ( 17 ). Finally, European unificaion has launched in European counries an enormous effor o increase comparabiliy of heir naional accouns saisics and heir compliance o European guidelines (see below). Neverheless, even in Europe, some clear differences in counry pracices coninue o exiss, e.g. wih respec o price- and volume changes of healh care and educaion services Welfare and he guidelines The successive guidelines all agreed ha naional accouns should no aim a measuring welfare, bu focus on serving as a pracical ool for macro-economic policy issues (e.g. public finance, balance of paymens, disribuion of income in erms of profis and wages). In he lae sixies and he beginning of he sevenies, naional income was frequenly criicised for no being a welfare measure (e.g. Mishan (1969)). However, he auhors of he inernaional guidelines did no inend o provide a measure of economic welfare. For example, Jaszi even regards as one of his principal conribuions o have resised successfully o he will-o -he-wisp of forging naional oupu ino a measure of economic welfare. I was a minoriy of one in a company ha included such menal gians as Simon Kuznes and John Hicks, and a one poin I had o defy a forceful Secreary of Commerce who had insruced he BEA o prepare a measure of welfare (Jaszi (1986), p. 411). According o Okun, [he] beauy of... presen pracice is ha no sensible person could seriously misake he GNP for [a measure of oal social welfare] (Okun (1971), p. 133). In 1972, Nordhaus and Tobin (1972) illusraed in an impressive way wha accouning aimed a measuring welfare would imply. They calculaed a Measure of Economic Welfare (MEW) by modifying radiional naional income figures in several respecs. For example, hey deduced an esimaed value of he disameniies of urbanisaion and hey added enaive esimaes for ( 17 ) However, in ransmiing daa on he USA o he inernaional organizaion, he BEA makes correcions in order o comply o he inernaional guidelines. 56 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

59 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure 2 he value of unpaid household services. Since hen, many measures similar o MEW have been calculaed (see Eisner (1988)). Frequenly, hese measures were presened as par of exended or oal accouns. Measuring he conribuion of economic aciviy o welfare is only one of he reasons for drawing up such accouns. Some oher moives are o obain: more inclusive and relevan measures of capial formaion and oher facors in economic growh, and beer and/or addiional daa o fi conceps of consumpion, invesmen, and producion relevan o economic heory and srucural economeric relaions (Eisner (1988), p. 1612). The increased use of social indicaors like he Human Developmen Index (UNDP (1991)) is a somewha relaed developmen. In hese social indicaors, naional income (per capia) is only one of he variables, oher variables being e.g. infan moraliy, life expecancy and adul lieracy raes. In conras o measures like MEW and Naional Income, social indicaors are no measures in money erms; hey serve solely as indexes. In 2009, on reques of he French governmen prominen inernaional expers invesigaed how o measure economic and social progress (Sigliz e al (2009)). They provide wo ypes of recommendaions: hose ha say wihin he exising framework of naional accouns and hose ha clearly go beyond ha. Major recommendaions wihin he exising framework of he naional accouns are: Emphasize well-esablished indicaors oher han GDP in he naional accouns. For example, a) Ne Domesic produc adjused o record naural resource depleion as depreciaion, b) ne naional disposable income, which accouns for income received from and paid abroad (as business income and household income o and from abroad can be very imporan) and c) real disposable household income, which is much more relevan for living sandards and well-being. Compiling naional balance shees for financial and non-financial asses, including naural resources, is also essenial for measuring curren well-being and susainabiliy. Improve he empirical measuremen of key producion aciviies, in paricular he provision of healh care and educaion services and changes in he qualiy and ype of services, e.g. due o digializaion. Curren naional accouns measuremen pracice focuses on measuring changes in inpus or a very rough indicaor of oupu of healh care and educaion. Changes in he qualiy of goods and services is mosly ignored, e.g. he imporan role of new medicine and medical reamens. This reflecs ha his measuremen ask is challenging as he link beween healh and educaion services and healh and educaion oucomes is complicaed. For example, people s lifesyles will also affec healh oucomes, and he ime parens spend wih heir children will affec exam scores. A household saellie accoun should become an major par of he regular saisical programme of counries and inernaional organisaions ( 18 ). Such a saellie accoun is imporan for showing exended measures of income, consumpion and wealh of households and heir disribuion. Household income measures should be adjused for governmen services in kind, like healh and educaion services and subsidized housing, museums and parks. When here are large changes in inequaliy, GDP or any oher aggregae compued per capia does no provide an accurae assessmen of he siuaion in which mos people find hemselves. The household saellie accouns may also serve o incorporae in household income he implici income derived from non-marke aciviies (e.g. cleaning, cooking and care for children, sick people and he elderly), o rea ime and money spen ( 18 ) In he inernaional guidelines SNA 2008 and ESA 2010, saellie accouns are discussed in separae chapers, bu hey are no par of he cenral accouning framework, like he secor accouns and he supply and use ables, and hey are also absen or of secondary imporance in he regular saisical programme of mos counries and inernaional oragnisaions. EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 57

60 2 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure on ranspor services o work as inermediae consumpion and o show major changes in leisure ime. Major recommendaions o beer measure economic and social progress ha clearly go beyond he naional accouns framework are: Objecive and subjecive informaion should be provided abou he various dimensions of well-being, including no only maerial living sandards, healh and educaion, bu also poliical voice and governance, social relaionships, environmen and insecuriy. A dashboard of indicaors should be used o measure susainabiliy. The economic componen of susainabiliy should be capured by moneary aggregaion on iems for which reasonable valuaion echniques exis, such as physical capial, human capial and cerain naural resources. For he environmenal aspecs of susainabiliy, here is a need for a clear indicaor of our proximiy o dangerous levels of environmenal damage, like climae change or he depleion of fishing socks. This relaively recen repor was a game-changer and has already exered an enormous influence on public debae abou measuring welfare, qualiy of life and susainable developmen. I has also changed measuremen and publicaion pracice by inernaional organisaions, like Unied Naions, OECD and European Commission. For example, since 2011, OECD is publishing regularly for all OECD counries muli-dimensional saisical indicaors on well-being (see OECD (2015)). Eleven differen dimensions of well-being are disinguished, e.g., income and wealh, jobs and earning, healh saus, educaion and skills, social connecions, environmen qualiy and subjecive well-being; for each dimension, several saisical indicaors are presened. Recenly, also hisorical ime series on he developmen of well-being in OECD counries since 1820 have been published. This includes no only GDP per capia, bu also indicaors like lieracy, life expecancy, people s heigh, crime raes (homicide) and income inequaliy (see van Zanden e al (2014)). Serious effors have also been underaken o sar compiling naional accouns saisics on human capial (see UNECE (2016)) and on he disribuion of household income, consumpion and saving (see Zwijnenburg e al (2017)) Differences in scope of naional accouns all over he world Under he influence of he inernaional guidelines and he inernaional organisaions, naional accouns saisics are now available for pracically all counries. For mos counries, hey have also gradually exended in scope and deail. Neverheless, sill subsanial differences in scope, deail, qualiy and frequency exis beween he naional accouns saisics published by counries. For example, since he fifies counries like Norway, Denmark, he Neherlands and France ( 19 ) publish annual inpu-oupu ables. Inpu-oupu ables were incorporaed for he firs ime in he inernaional guidelines in he SNA68 and ESA70. Decades laer, only a few more counries published annual inpu-oupu ables. For a somewha larger group incidenal bu usually raher oudaed inpu-oupu ables exised. Nowadays, annual supply and use ables are par of he sandard delivery programme for OECD and especially EU counries, bu in mos oher counries such daa are sill no available. A similar sory can be old for he deailed secor accouns proposed by he SNA68, he ESA70 and he mos recen inernaional guidelines. Even ( 19 ) A very ineresing survey of he pos-war developmens in France can be found in Demoes-Mainard and Bournay (1994). A more exended overview is given by Vanoli (2005). 58 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

61 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure 2 now, a grea majoriy of he counries in he world only apply raher simple accouning sysems of he SNA53-syle. Subsanial differences in counry pracices exis also wih respec o specific naional accouns saisics, like regional accouns, quarerly accouns, saellies and balance shees. In some counries, all of hem are regularly published (e.g. in he Neherlands, France and Canada). In a somewha larger group, some of hem are regularly published, while in many counries hardly any daa are regularly published on any of hese opics. Our remarks wih respec o he inpu-oupu ables, deailed secor accouns, balance shees and saellie accouns reveal ha mos of he inernaional guidelines have been much more ambiious and encompassing han he naional accouning pracice of heir ime. This parly reflecs heir role as a pedagogical device and innovaive insrumen ( 20 ). However, in he case of he USA i may also parly reflec a fundamenally differen view on he role and design of he naional accouns. In line wih USA naional accouns pracice up o he 1970s and 1980s ( 21 ), wo quoes from American economiss sress he imporance of a relaively simple se of accouns and criicize he complexiy and cos-inefficiency of universal accouning framework: These accouns are designed o answer Who does Wha by means of Wha for Wha purpose wih Whom in exchange for Wha wih Wha changes in socks? Given his level of complexiy, here is a disinc danger ha when he revised SNA is acually pu in place, i, like he Hubble elescope, may no be successful in bringing ino focus a clear view of wha i was designed o examine. Only professional naional accounans will be able o fahom he naional accouns. Furhermore, he esablishmen of such an elaborae sysem as he sandard o be adoped by naional and inernaional saisical offices may resul in he SNA becoming a saisical behemoh independen of is creaors and wih an illogic of is own-no unlike a Frankensein monser. One of he major virues of naional accouning sysems used by many counries is ha hey do provide a relaively simple macroeconomic overview of he economic sysem (Ruggles (1990), p. 419). For every series of any real ineres ha is developed, a leas a dozen series of rivial or no value mus be esimaed o fill ou he accouns. Because mos of he series called for are of no appreciable ineres, exising sysems of daa collecion do no provide he informaion required by he new SNA (F.B.: SNA68); eiher collecion of rivial daa would be required or he number would have o be imaginary. The new SNA has anoher weakness: i is so complicaed ha no even serious and exper users of naional income and produc daa (and few producers for ha maer) can be expeced o undersand i or he meaning of he numbers i is o conain. A very simple se of accouns, supplemened by supporing ables o provide analyically ineresing deail and alernaive breakdowns, is a far beer approach, in my opinion (Denison (1971), p. 38). ( 20 ) I also reflecs he oally differen amouns of resources in counries available for saisics in general, and for naional accouns in paricular. ( 21 ) This changed in he 1990s: in 1993, he US Bureau of Economic Analysis (BEA) announced ha i plans o move oward he SNA. (BEA, 1993) Since 1993, all of BEA s major revisions have incorporaed feaures of and improved consisency wih he laes versions of he SNA. Also erminology is adjused in line wih he successive versions SNA, e.g. implemening SNA 2008 involved also swiching from he erm flow-of-funds-accouns o financial accouns. EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 59

62 2 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure 4.5. European unificaion The ongoing European unificaion is revoluionising European naional accouning ( 22 ). Naional accouns figures, like naional income, governmen defici and GDP volume growh have been seleced o play a special role in monioring and managing he European unificaion (see also Table 5). This role in European policy has also drasically increased he imporance of naional accouns saisics in naional policy. In discussing and deciding on he naional budge, naional accouns saisics on he governmen defici have become he cenral figures in all EU counries. They have ofen aken over his role from specifically naionally defined conceps (see ex box). The EU member qaes and he European Commission have been aware ha he naional accouns saisics had o be srenghened for such usage. They herefore launched an ambiious programme for improving he qualiy and comparabiliy of naional accouns figures and for drasically exending he se of naional accouns saisics ha are available for all EU member saes. Table 5: Use of naional accouns saisics for European policy Policy area Moneary policy and public finance Produciviy and growh policy Social policy Regional policy Agriculural policy Developmen aid Defense policy Maximum oal expendiure by EU EU budge conribuion GNI Which naional accouns saisic? Governmen defici and deb as a percenage of GDP Financial accouns showing e.g. he size of new morgages and loans by corporaions Economic growh, expendiure on R&D as percenage of GDP, EU KLEMS Social proecion saisics closely linked o naional accouns conceps Regional produc per capia is yardsick for graning regional funds Agriculural accouns showing also he developmen of farmer income Low domesic produc per capia is a yardsick for graning aid Expendiure on defense as a percenage of GDP Percenage of GNI (abou 1 %) From a universal perspecive, hese European developmens are in wo respecs revoluionary. Firsly, he developmen of jurisprudence on he inerpreaion and applicaion of he inernaional guidelines is a oally new developmen for he naional accouns. Secondly, he European experience gives a concree example of how he qualiy and comparabiliy of naional accouns saisics can be improved, e.g. by audiing ( 23 ), issuing guidelines on inpus, official requiremens for compiling and publishing an exensive se of naional accouns saisics and by more specific guidelines. Some of his exra specificaion is also very useful for counries ouside Europe, e.g. wih respec o he delimiaion of he governmen secor, he definiion of price and volume changes for various ypes of indusries and produc, he definiion and measuremen of services of owner-occupied dwellings, compiling regional accouns and compiling quarerly accouns. As a consequence, he European guidelines on naional accouning could be regarded as a useful supplemen of he universal guidelines, even for non-european counries. ( 22 ) For a more exended discussion see Bos (2009a), pp and Bos (2011). ( 23 ) For some EU-counries, his resuled in upwards revisions of official domesic produc and naional income saisics of 10 o 20 %! 60 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

63 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure 2 Box 1: European norms for acual defici and deb The reay of Maasrich in 1992 implied ha moneary policy became a responsibiliy of he European cenral bank and ha naional fiscal policy should comply wih he European norms of acual defici and deb. Deficis should no exceed 3 % of GDP and deb mus be below 60 % of GDP or be declining owards he 60 % norm a a saisfacory rae. According o he Sabiliy and Growh Pac, he budge balance should be close o balance or in surplus in he long run. As a consequence, he naional conceps on public finance were replaced by he new European conceps based on he naional accouns. This had several pracical implicaions: A change in conceps. For example, according o he naional accoun s concep of budge balance, revenue and expendiure like axes and ineres paymens should be recorded on a ransacions basis. Financial ransacions like loans and he sale of equiy are irrelevan and he governmen includes no only he sae and social securiy funds, bu also municipaliies, provinces and many oher non-marke unis mainly financed and conrolled by he governmen. The conceps canno be changed anymore over ime by he governmen. A link o naional accouns saisics and herefore a new role for naional saisical offices and a more limied role of Minisries of Finance. The official figures repored o he European Commission and European Cenral Bank should be consisen wih hose repored by he naional saisical office. In he end herefore, saisical offices are responsible for ranslaing he general European conceps ino operaional conceps for heir counry and o make he bes esimaes for hese operaional conceps. The ransiion owards European conceps does no imply ha bookkeeping and bookkeeping ricks have become irrelevan. Like all naional conceps of axable income, he European conceps on public finance can affec acual behaviour (e.g. simulae leasing of capial goods o reduce he defici or simulae he sale of public equiy in order o reduce public deb) and he specific insiuional arrangemens chosen ( a ). Furhermore, hey are no opimal from an economicheoreic poin of view (e.g. no forward looking and ignores financial asses and implici liabiliies like fuure pensions) and may no well ake accoun of he curren economic siuaion in a counry. They are he oucome of poliical negoiaions in view of he circumsances in Europe in 1992 and he purposes of he crieria, i.e. o provide signals ha counries are willing and able o live wih he discipline required by EMU (see Bovenberg and De Jong (1996), p. 18). Following he seminal work by Auerbach, Gokhale and Kolikoff (1991), in 1997, he Neherlands Bureau for Economic Policy Analysis sared o calculae generaional accouns for he Neherlands (see Bos and Teulings (2013b)). All kinds of naional accouns informaion on governmen revenue, expendiure and deb and economic growh serve as inpu. These calculaions demonsraed ha policy arrangemens (axes, public expendiure on social securiy, educaion and healh care, subsidies, ec.) in he Neherlands were no susainable due o he coss of ageing and he exhausion of he naural gas resources. They also showed he consequences of alernaive policy measures o reduce his gap for differen generaions. This forward looking approach of generaional accouning is he new paradigm for Duch public finance. Similar calculaions are now made regularly for all EU member saes and play also a role in assessing European fiscal policy. ( a ) On he meris and limiaions of he EMU-arges of governmen defici and deb, see also Bos (2009a, chaper 8) and Bos (2008b). EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 61

64 2 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure 4.6. The impac of he financial crisis of The financial and economic crisis of has had a major impac on he demand for naional accouns saisics. I has sressed he imporance of much beer and imelier recording of financial asses and liabiliies, heir inernaional dimension and heir relaionship o he real economy. The suppliers of naional accouns saisics have responded in various ways. For example, in he USA, inegraed macro-economic accouns were developed (see Yamashia, 2013), e.g. showing he imporance of morgage deb and holding gains and losses in household wealh accumulaion. In he European Union, much effor was pu ino beer recording public inervenions in he financial sysem and heir ineracion wih he member saes public finance. Inernaional organizaions, like UN, IMF, World Bank, European Commission and OECD, cooperaed o idenify and remedy gaps in he inernaional supply of naional saisics. More comprehensive and imely daa were needed. In paricular improvemen was recommended of he inernaional reporing abou he financial healh of financial insiuions, abou crossborder financial relaionships (e.g. abou he imporance of Greek governmen deb for French and German banks and Duch pension funds), abou governmen finance saisics, balance shees of households, governmen and non-financial insiuions and abou real esae prices ( 24 ). Also more coordinaion in he publicaion and communicaion of official saisics was recommended. The financial crisis has also raised serious doubs abou he measuremen of value added and real oupu growh of banks. Their definiions in he inernaional guidelines may need o be remedied, e.g. in order o beer accoun for aking more or less risks and o measure bank service oupu on ransacion couns insead of balances of loans and deposis (see Inklaar and Wang (2013) and van de Ven (2015)) Long run perspecives on he economy As already shown by Kuznes, compiling long run naional accouns saisics can be very imporan o beer undersand fundamenal economic issues. The much more recen work by Maddison, Gordon, Reinhar and Rogoff and Pikey can illusrae his. Maddison (2003, 2005 and 2007) colleced and compiled real GDP per capia figures for counries all over he world during he las millennium. This comprehensive daa se fully focused on real GDP per capia and did no show any oher naional accouns saisics, e.g. no breakdown by indusry is presened. This daa se has been used inensively by many oher economiss for heir research o beer undersand economic growh and why some counries are rich and oher poor (see e.g. Acemoglu and Robinson (2012)). Gordon (2012) invesigaes he imporance of innovaion for economic growh. Is poin of deparure is growh in real GDP per capia in he fronier counry since 1300, he UK unil 1906 and he USA aferwards. Growh in his fronier gradually acceleraed afer 1750, reached a peak in he middle of he 20h cenury, and has been slowing down since. This sudy suggess ha he growh effecs of he informaion revoluion are no of he same order of imporance as hose of previous echnological revoluions and are, in any even, playing ou. Fuure growh per capia could fall below 0.5 percen per year for an exended period of decades. This sudy plays a major role in inernaional discussions abou our fuure living sandards. However, is analysis and conclusion depend o grea exen on how well new goods and services, e.g. compuers, inerne and mobile phones) are measured in real GDP figures. This problem was ( 24 ) This is he G20 Daa Gaps iniiaive, a se of 20 recommendaions on improving economic and financial saisics. 62 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

65 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure 2 already raised by Adam Smih more han wo cenuries ago and, despie a lo of research effor, has sill no been resolved (see Hulen (2015)). Up o he recen financial crisis, macro-economic forecass on he developed counries mosly ignored he possibiliy of a financial crisis (see Bos and Teulings (2013)). This reflecs ha hese forecass were based on naional accouns ime series spanning some decades. Hisorical and long run ime series on economic growh in he Neherlands and oher developed counries did no seem relevan for such forecass. As a consequence, he recen financial crisis came as a oal surprise. Following (Taleb (2007)), his was regarded an example of a black swan: a very unlikely even wih a huge impac. However, by compiling and collecing daa on financial crises all over he world during he pas wo cenuries (e.g. daa on economic growh, governmen deb, household deb and capial inflows), Reinhar and Rogoff (2011) showed ha from a long run perspecive such black swans were much less rare han was assumed. Furhermore, hey showed ha sound economic policy and insiuions can play an imporan role in avoiding financial crisis or miigaing is mos negaive effecs. Since he Second World War, despie he drasic expansion of he scope of naional accouns saisics, inequaliy of income and wealh was an issue absen in naional accouns saisics all over he world. Also in he inernaional guidelines, he issue was fully ignored. However, only some years ago, Pikey (2014) showed ha his is an imporan issue and ha i can be well addressed by compiling naional accouns saisics on he size and economic reurn on privae and public wealh and on capial income and labour income as a percenage of naional income. In order o reveal he amoun of inequaliy, capial income and labour income are broken down by level of income and wealh, e.g. he poores 50 %, he riches 10 %, he middle class of 40 % and he very rich 1 %. These saisics show ha 50 % of naional income in he USA in 2010 is for he riches 10 %, which is similar o he siuaion in France, Briain and Germany in 1910 ( 25 ). Saring from his new se of saisics, he advocaes more progressive axaion of income and wealh. 5. Conclusions and hree suggesions for he fuure Since he seveneenh cenury, esimaes of naional income and GDP and oher naional accoun saisics, have mainly been developed for fiscal and moneary policy purposes, for measuring progress in maerial well-being and for showing heir relaionship wih he supply and use of goods and services in a marke economy. Naional accouns saisics were no developed for measuring welfare, qualiy of life, inequaliy or susainable developmen. This reflecs ha for such policy issues measuremen in moneary erms is less meaningful and more difficul o compile reliably and ofen requires also oally differen daa sources. Over he years, he scope, frequency and imeliness of naional accouns saisics has been expanded grealy, recenly in response o he financial crisis. Bu he supply of curren naional accouns saisics is sill far from being capable of serving he many daa needs of he fuure, including hose on welfare, qualiy of life, inequaliy or susainable developmen. Furhermore, major changes, like globalisaion ( 26 ), digialisaion, exhausion of naural resources and growing ( 25 ) Pikey sresses ha comparisons on he basis of inequaliy indexes, like he Gini coefficien, do no suffice o undersand he developmen of he disribuion of wealh and income over ime and across counries. ( 26 ) See van de Ven (2017). EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 63

66 2 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure inequaliy, challenge he relevance of he curren naional accouning conceps. How should he daa needs of he fuure and he complicaed measuremen issues bes be me? Wha can we learn from our overview of naional accouning since he seveneenh cenury? We provide hree suggesions ( 27 ): More slow saisics for difficul o measure opics. Fewer rapid saisics ha do no conain much informaion. Beer explaining value added, use and misuse of naional accouns saisics 5.1. More slow saisics for difficul o measure opics In order o mee he complicaed measuremen issues and many daa needs of he fuure, a sizable par of naional accouns saisics should be slow-saisics, i.e. compiled regularly bu no very frequenly, e.g. every wo or five years. In order o allow for inernaional comparison and o increase public aenion for he saisic, he publicaion of such slow-saisics should be coordinaed inernaionally. To reduce misinerpreaion of GDP-volume growh saisics and o simulae susainable growh, such slow-saisics should include well-being accouns, informaion on inequaliy of income and wealh and ne-ne saisics on domesic produc and naional income, i.e. including a correcion for he exhausion of naural resources. For undersanding economic growh, a regional breakdown of saisics on he naional economy (regional accouns) is also very imporan, as economic growh is ofen locaion specific and can differ subsanially wihin a naion. Awarding he 2008 Nobel prize in economics o Krugman, in paricular for his work on new economic geography (see e.g. Krugman (1991)), reflecs he imporance of he issue, bu regional accouns were hardly menioned in he SNA 2008 ( 28 ). Anoher major opic is alernaive measuremen of price- and volume change, e.g. wih respec o he qualiy and new ypes of financial, educaion and healh care services and he impac of elecommunicaion revoluion Fewer rapid saisics ha do no conain much informaion In many developed counries, he frequency and iming of all kinds of naional accouns saisics has been increased enormously. This is a major asse. However, here seems o have been some overshooing, as no all hese saisics conain really new informaion. This may be revealed by compiling saisics on he measuremen errors in he successive esimaes (see Bos (2009b)). The saisical informaion on measuremen errors in naional accouns saisics may also be used o inform users abou he inaccuracy of he esimaes Beer explaining value added, use and misuse of naional accouning By providing key-indicaors on he srucure and developmen of he naional economy, naional accouns saisics play an imporan role in public debae, policy and economic ( 27 ) For more ideas, see Bos (1996), Bos (2009a, chaper 9 and 10), Vanoli (2014) and van de Ven (2017). ( 28 ) Some figures may illusrae he issue. In he USA and Germany he riches sae has abou wice he per capia income of he poores sae and his raio is five in China and India (see Sehia (2014)). For inernaional comparison of small counries, comparison wih specific regions in large counries may be even more relevan han comparison wih he oal of large counries. For example, he Neherlands is a small and highly urbanized counry wih 17 million inhabians and could be compared o he neighbouring German sae Norh Rhine Wesfalia or o US saes like New York and Florida. 64 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

67 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure 2 analysis. A proper use of hese saisics requires a good basic undersanding of he underlying conceps and he reliabiliy of he daa. Wihou such undersanding, public debae, policy and analysis will seer on he wrong compass. However, here is widespread illieracy in naional accouning, naional accouning is generally considered o be one of he mos boring opics in economics and knowledge of basic naional accouns definiions and discussions seems o be irrelevan for economic researchers (see Bos, 2009a). The new universal guidelines on naional accouning (SNA 2008) are so lenghy and voluminous ha hardly any naional accounan has read hem. From a users perspecive, he problem is ha mos of he ex is devoed o explaining he general framework of naional accouning and ha he links wih specific uses for policy and analysis are ofen very indirec. How we define and measure he world is how we see he world and ha can affec how ciizens, business and governmen will reac and wha we hink ha is realisic, imporan or how fair. Changes in naional accouning rules, corporaion and income ax rules, rules by bookkeeping and supervisory bodies on banking, pension funds, hey maer grealy for wha we see and how we ac. Explaining he value added of naional accouns saisics for differen ypes of uses and heir link o adminisraive conceps of axaion, bookkeeping and supervision should herefore be a major par in economic curricula and in he educaion of naional accounans. And his needs no o be boring. The financial crisis and new research on long run perspecives on he naional economy (e.g. by Gordon, Rogoff and Reinhar and Pikey) have again revealed he meris and limiaions of naional accouns saisics. Recenly, also several popular books debaing biases in GDP and heir consequences for policy ( 29 ) have been published. All his informaion can be used o discuss he meris of changing is basic conceps, adding new perspecives or of even abolishing naional accouns saisics. Would policy and life in developmen counries or Saudi Arabia have been differen when GDP included a correcion for he exhausion of naural resources? Would policy and life in he USA have been differen when GDP included a correcion for he increase in inequaliy? How would we have managed he financial crisis in he absence of naional accouns saisics? I is also imporan o make a link beween naional accouns saisics and ordinary life. For example, wha could naional accouns saisics ell abou he life of Becky of 10 years in he USA versus ha of Desa of 10 years in Ehiopia (see Dasgupa (2007)) ( 30 ). Or: wha does a four-fold increase in GDP per capia since 1950 mean in he Neherlands? Wha does i ell abou he life of a female economic journalis now 60 years old and ha of her grandmoher who died a ha age in 1950? (see Bergen (2014)). Acknowledgemens The auhor would like o hank he referees for heir subsanial and very useful suggesions. The views expressed are hose of he auhor and do no necessarily reflec he views of his curren or former employer. ( 29 ) See e.g. Fioramoni (2013), Coyle (2014), Lepenies (2016) and Philipsen (2017). ( 30 ) A similar approach is aken by Kay (2003) comparing he lives of four families in USA, Sweden, Swizerland, Russia and India. EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 65

68 2 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure References Acemoglu, D. and J.A. Robinson (2012), Why naions fail The origins of power, prosperiy and povery, Crown Business Profile Books, London. Arvay, J. (1994), The Maerial Produc Sysem (MPS): a rerospecive, in Kenessey (1994), pp Aukrus, O. (1986), On he occasion of a Nobel prize, Review of Income and Wealh, Vol. 32, issue 1, pp Aukrus, O. (1994), The Scandinavian conribuion o naional accouning, in Kenessey (1994), pp Auerbach, A.J., J. Gokhale and L.J. Kolikoff (1991), Generaional accouns A meaningful alernaive o defici accouning, NBER Working Paper No BEA (US Bureau of Economic Analysis) (1993), New inernaional guidelines in economic accouning, Survey of Curren Business, February 1993, pp Bergen, A. van, (2014), Gouden Jaren; hoe ons dagelijks leven in een halve eeuw onvoorselbaar is veranderd, Alas Conac Uigeverij. Bos, F. (1993), Sandard naional accouning conceps, economic heory and daa compilaion issues; on consancy and change in he Unied Naions Manuals on Naional Accouning (1947, 1953, 1968, 1993), Naional Accouns Occasional Paper, No 61, Saisics Neherlands, Voorburg. Bos, F. (1994), Consancy and Change in he Unied Naions Manuals on Naional Accouning (1947, 1953, 1968, 1993), in Kenessey (1994), pp Bos, F. (1996), The fuure of he naional accouns, Naional Accouns Occasional Paper No 84, Saisics Neherlands, Voorburg. Bos, F. (2006), The developmen of he Duch naional accouns as a ool for analysis and policy, Saisica Neerlandica, Vol. 60, issue 2, pp Bos, F. (2008a), Uses of Naional Accouns: Hisory, Inernaional Sandardizaion and Applicaions in he Neherlands, paper presened a he 2008 World Conference on Naional Accouning and Performance Measures across Naions, May, Washingon DC, MPRA paper No Bos, F. (2008b), The Duch fiscal framework; hisory, curren pracice and he role of he CPB, OECD Journal on Budgeing, Vol. 8, issue 1, pp Bos, F. (2009a), The naional accouns as a ool for analysis and policy; in view of hisory, economic heory and daa compilaion issues, VDM Verlag Dr. Müller, Saarbrücken. Bos, F. (2009b), The Ar and Craf of Compiling Naional Accouns Saisics and heir Implicaions for Reliabiliy, Review of Income and Wealh, Vol. 55, issue 4, pp Bos, F. (2011), Three Cenuries of Macro-Economic Saisics, MPRA paper No Bos, F. (2013), Meaning and measuremen of naional accouns saisics, paper for he World Economics Associaion s Online Conference on he Poliical Economy of Economic Merics, 28 January-25 February 2013, MPRA paper No EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

69 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure 2 Bos, F. and C. Teulings (2013a), Neherlands: Fosering Consensus on Fiscal Policy, in G. Kopis (ed.), 2013, Resoring Public Deb Susainabiliy: he Role of Independen Fiscal Insiuions, Oxford Universiy Press, Chaper 7, pp Bos, F. and C. Teulings (2013b), Shor and long-erm forecasing by he Neherlands Bureau for Economic Policy Analysis (CPB): science, wichcraf or pracical ool for policy?, OECD Journal of Budgeing, Vol. 13, issue 1, pp Bovenberg, L.J. and A. de Jong (1996), The ins and ous of European moneary union, CPB Repor 1996/3, pp Carson, C.S. (1975), The hisory of he Unied Saes naional income and produc accouns: he developmen of an analyical ool, Review of Income and Wealh, Vol. 21, issue 2, pp Clark, C. (1937), Naional income and oulay, MacMillan and Co, Ld, London. Clark, C. (1940), Condiions of economic progress, MacMillan and Co., Ld, London. Coyle, D. (2014), GDP: A brief bu affecionae hisory, Princeon Universiy Press. Dasgupa, P. (2007), Economics: a very shor inroducion, Oxford Universiy Press. Demoes-Mainard, M. and J. Bournay (1994), Revisions of he French naional accouns; pas experience as lessons for he fuure, paper presened a he 23h IARIW Conference, S Andrews, Canada. Denison, E.F. (1947), Repor on riparie discussions of naional income measuremen, Sudies in Income and Wealh, Vol. X, 3 22, NBER, New York. Denison, E.F. (1971), U.S. Naional Income and Produc Esimaes: Evaluaion and Some Specific Suggesions for Improvemen, Survey of Curren Business, 50h Anniversary Issue, pp Eisner, R. (1988), Exended Accouns for Naional Income and Produc, Journal of Economic Lieraure, XXVI (4), pp Elis, W. (1984), The classical heory of economic growh, MacMillan and Co., Ld, London. Eurosa (1995), European Sysem of Naional and Regional Accouns (ESA95), Office for Official Publicaions of he European Communiies, Luxembourg. Eurosa (2013), European Sysem of Accouns (ESA 2010), Publicaion Office of he European Union, Luxembourg. Fisher, I. (1911), The purchasing power of money and he heory of ineres. Is deerminaion and relaion o credi, ineres and crisis, McMillan, New York. Fleewood, W. (1707), Chronicon Preciosum: or An accoun of English money, he price of corn and oher commodiies for he las 600 years, 1s edn (anonymous) Harper, London, reprined in Frisch, R. (1933), Propagaion problems and impulse problems in dynamic economics, in Economic essays in honour of Gusav Cassel, Allen & Unwin, London. Fioramoni, L. (2013), Gross Domesic Problem: The poliics behind he world s mos powerful number, Zed Books. Gordon, R.J. (2012), Is U.S. Economic Growh Over? Falering Innovaion Confrons he Six Headwinds, NBER Working Paper No Hicks, J.R. (1940), The Valuaion of Social Income, Economica, Vol. 7, No 26, pp EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 67

70 2 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure Hicks, J.R. (1948), The Valuaion of Social Income A Commen on Professor Kuznes Reflecions, Economica, Vol. 16, No 59, pp Hora Siccama, J.H. and B. van Rees (1798), Algemeen belasingplan, Archieven B. van Rees, Koninklijke Biblioheek, The Hague. Hulen, C.R. (2015), Measuring he Economy of he 21s Cenury, NBER Reporer, No 4, pp Inklaar, R. and J.C. Wang (2013), Real oupu of bank services: wha couns is wha banks do, no wha hey own, Economica, Vol. 80, issue 317, pp Jäger, K. (2016), EU KLEMS Growh and Produciviy Accouns 2016 Release, saisical module. Jaszi, G. (1986), An Economic Accounan s Audi, American Economic Associaion Papers and Proceedings, pp Jones, R. (2000), Naional accouning, governmen budgeing and he accouning discipline, Financial Accounabiliy & Managemen, Vol. 16, issue 2, pp Kampmann, V. (1942), Danmarks naionalindkoms i rediverne (Denmark s naional income in he 1930s), lecure o Naionaløkonomisk Forening 19. Ocober In: NaionalOkonomisk Tidsskrif, Vol. 80, pp Kay, J. (2003), Culure and prosperiy; he ruh abou markes, why some naions are rich bu mos remain poor, Harper Business. Kenessey, Z (ed.) (1994), The Accouns of Naions, IOS Press, Amserdam. Keynes, J.M. (1920), The economic consequences of he peace, MacMillan and Co., Ld, London. Keynes, J.M. (1936), The general heory of employmen, ineres and money, MacMillan and Cambridge Universiy Press, London. Keynes, J.M. (1940), How o pay for he war, Harcour Brace & co, New York. King, G. (1936), Two Tracs, John Hopkins Press, Balimore. Krugman, P.R. (1991), Increasing reurns and economic geography, Journal of Poliical Economy, Vol. 99, issue 3, pp Krugman, P.R. and M. Obsfield (2011), Inernaional economics; heory and policy, Harper Collins College Publishers. Kuznes, S., (1934), Naional income, , NBER Bullein, No 49. Kuznes, S. (1938), Commodiy flow and capial formaion. NBER Books, New York. Kuznes, S. (1948a), Naional Income: a New Version (Discussion of he new Deparmen of Commerce Income Series), The Review of Economics and Saisics, Vol. 30, issue 3, pp Kuznes, S. (1948b), On he Valuaion of Social Income Reflecion on professor Hicks aricle, Economica, XV(57), pp (par I), XV(58), pp (par II). Lepenies, P. (2016), The power of a single number; a poliical hisory of GDP, Columbia Universiy Press. Lequiller, F. and D. Blades (2014), Undersanding naional accouns, OECD Publishing, Paris. Leonief, W.W. (1936), Quaniaive Inpu and Oupu Relaions in he Economic Sysem of he Unied Saes, Review of Economic Saisics, Vol. 28, issue 3, pp EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

71 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure 2 Lindahl, E., E. Dahlgren and K. Kock (1937), The Naional income in Sweden Pars one and wo. Universiy of Sockholm, Insiue for Social Sciences, London. Lundberg, E. (1984), Lundberg on Kuznes. In: H.W. Spiegel and W.J. Samuels, Conemporary economiss in perspecive, Par II, JAI Press Inc., London, pp Maddison, A. (2003), The World Economy: Hisorical saisics, OECD Developmen Cenre Sudies, OECD, Paris. Maddison, A. (2005), Measuring and inerpreing world economic performance , Review of Income and Wealh, Vol. 51, issue 1, pp Maddison, A. (2007), Conours of he World Economy, AD. Essays in Macro-Economic Hisory, Oxford Universiy Press McCulla, S.H., K.E. Moses and B.R. Moulon (2015), The Naional Income and Produc Accouns and he Sysem of Naional Accouns 2008, comparison and research plans, Survey of Curren Business, June 2015, pp Meade, J.A. and R. Sone (1941), The Consrucion of Tables of Naional Income, Expendiure, Savings and Invesmen, Economic Journal (51)202/203, pp Meade, J.A. (1951), The balance of paymens, Oxford Universiy Press, London. Meek, R.L. (1962), The Economics of Physiocracy: Essays and Translaions, Harvard Universiy Press. Meelerkamp, R. (1804), De oesand van Nederland in vergelijking gebrag me die van enige andere landen van Europa, Cornel en van Baalen, Roerdam. Mishan, E.J. (1969), The Coss of Economic Growh, Penguin Books Ld, New York. Morgan, M.S. (1990), The hisory of economeric ideas, Cambridge Universiy Press, Cambridge. Nordhaus, W. and J. Tobin (1972), Is Growh Obsolee? in: Naional Bureau of Economic Research, Economic Growh, Columbia Universiy Press, New York, pp OECD (2015), How s Life? 2015: Measuring Well-being, OECD Publishing Okun, A.M. (1971), Social Welfare has no Price Tag, Survey of Curren Business, Vol. 51, issue 7, pp Painkin, D. (1976), Keynes and Economerics: on he Ineracion Beween he Macroeconomic Revoluions of he Inerwar Period, Economerica, Vol. 44, issue 6, pp Pikey, T. (2014), Capial in he weny-firs cenury, Harvard Universiy Press. Philipsen, D. (2017), The lile big number: how GDP came o rule he world and wha o do abou i, Princeon Universiy Press. Reich, U. (1994), German naional accouns beween poliics and academics, in Kenessey (1994), pp Reinhar, C. and K. Rogoff (2011), From financial crash o deb crisis, American Economic Review, Vol. 101, issue 5, pp Ruggles, R. (1990), A Noe on he Revision of he Unied Naions Sysem of Naional Accouns Review of Income and Wealh, Vol. 36, issue 4, pp Sehia, D. (2014), Regional accouns of India: mehods, new esimaes and heir uses, Review of Income and Wealh, Vol. 62, issue 1, pp EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 69

72 2 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure Sigliz, J.E., A. Sen and J.P. Fioussi (2009), Repor by he Commission on he measuremen of economic performance and social progress. Sone, R. (1947), Definiion and Measuremen of he Naional Income and Relaed Toal. Appendix in UN (1947). Sone, R. (1951), The Use and Developmen of Naional Income and Expendiure Esimaes. In: D.N. Cheser (ed.), Lessons of he Briish war economy, Cambridge Universiy Press, pp Sone, R. (1997), Some Briish Empiriciss in he Social Sciences , Cambridge Universiy Press, Cambridge. Sudenski, P. (1958), The Income of Naions, New York Universiy Press, New York. Taleb, N.N. (2007), The Black Swan: The Impac of he Highly Improbable, Random House. Tinbergen, J. (1936), Prae-advies over de vragen: Kan hier e lande, al dan nie na overheidsingrijpen, een verbeering van de binnenlandsche conjuncuur inreden, ook zonder verbeering van onze exporposiie? Welke leering kan en aanzien van di vraagsuk worden gerokken ui de ervaringen van andere landen?, in: Vereniging voor de Saahuishoudkunde en de saisiek, Praeadviezen, Marinus Nijhoff, s Gravenhage, pp Tooze, J.A. (2001), Saisics and he German Sae: he making of modern economic knowledge, Cambridge Universiy Press. UN (1947), Measuremen of Naional Income and he Consrucion of Social Accouns, Sudies and Repors on Saisical Mehods No 7, UN, Geneva. UN (1953), A Sysem of Naional Accouns and Supporing Tables ( SNA58 ), Sudies in Mehods, Series F No 2, Rev. 1, UN, New York. UN (1968), A Sysem of Naional Accouns ( SNA68 ), Sudies in Mehods, Series F No 2, Rev. 3, UN, New York. UN (1993), Sysem of naional accouns ( SNA93 ), Join publicaion by he European Commission, IMF, OECD, World Bank and UN. UNDP (Unied Naions Developmen Programme) (1991), Human Developmen Repor 1991, Oxford Universiy Press, New York. UN (2002), Use of Macro Accouns in Policy Analysis, Sudies in Mehods, Series F, No 81, UN, New York. Unied Naions, European Commission, Inernaional Moneary Fund, OECD and World Bank (2008), Sysem of Naional Accouns 2008 ( SNA 2008 ), New York, USA. UNECE (2016), Guide on measuring human capial, New York and Geneva. Vanoli, A. (2005), A hisory of naional accouning, IOS Press, Amserdam. Vanoli, A. (2014), Naional accouning a he beginning of he 21s cenury: Wherefrom? Whereo?, Eurosa Review of Naional Accouns and Macroeconomic Indicaors (EURONA), 1/2014, pp Ven, P. van de (2015), New sandards for compiling naional accouns: wha s he impac on GDP and oher macro-economic indicaors?, OECD Saisics Brief, February 2015, No 20. Ven, P. van de (2017), Presen and Fuure Challenges o he Sysem of Naional Accouns: Linking Micro and Macro, Review of Income and Wealh, forhcoming. 70 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

73 Uses of naional accouns from he 17h cenury ill presen and hree suggesions for he fuure 2 Yamashia, T. (2013), A guide o he Inegraed Macroeconomic Accouns, Survey of Curren Business, April 2013, pp Young, A. (1770), A six monhs' our hrough he norh of England, 4 volumes, Srahan and Nicoll, London. Zanden, J.L. van, J. Baen, M.M.d'Ercole, A. Rijpma, C. Smih and M. Timmer (2014), How was life? Global well-being since 1820, OECD Publishing. Zwijnenburg, J., S. Bourno and F. Giovannelli (2017), Repor by exper group on dispariies in a naional accouns framework; resuls from he 2015 exercise, OECD Saisics Working Paper 2016/10. EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 71

74

75 3 Processing scanner daa by an augmened GUV index (*) LUDWIG VON AUER ( 1 ) Absrac: Saisics Neherlands is a pioneer in he use of scanner daa for price measuremen purposes. In his paper A New Mehodology For Processing Scanner Daa in he Duch CPI Anonio G. Chessa (2016a) of Saisics Neherlands proposes an approach for a more coheren and auomaed processing of scanner daa. This approach is denoed here as he C-approach. The C-approach uses a wo sage aggregaion procedure in which individual iems are bundled ino producs and hen producs ino segmens. On he lower sage (iems producs), for each produc and each monh, a uni value and an aggregae quaniy is compued. These produc daa are used in he upper sage (producs segmens), where for each segmen an average price change is compued. The compuaions on he upper sage uilize he Geary- Khamis index. The aggregaion of segmens ino he overall price change of he economy uses he radiional Laspeyres-ype procedure. The C-approach is no concerned wih his final sep of price aggregaion. The C-approach is an imporan advancemen in official price measuremen. A he same ime his paper argues ha in some siuaions he C-approach migh be unnecessarily complex. As an alernaive, i proposes o replace he Geary-Khamis index by a simpler formula like he Gerardi index or even an augmened Generalized Uni Value index. Furhermore, i recommends o examine wheher for some segmens he wo sage aggregaion of he C-approach can be replaced by a single sage aggregaion. JEL codes: C43, E31 Keywords: CPI, scanner daa, relaunch, assignmen bias, assormen bias, Lehr, GUV index ( 1 ) Universiä Trier, Germany ( * ) Companion paper o A New Mehodology For Processing Scanner Daa in he Duch CPI by Anonio G. Chessa (2016a).

76 3 Processing scanner daa by an augmened GUV index 1. Inroducion In radiional price collecion, employees of naional saisical offices visi shops and collec he prices of a sample of iems. Scanner daa are an aracive alernaive o his cosly and ime consuming approach. They provide he price analys wih informaion on (1) large samples of iems insead of he raher resriced samples available by he radiional price collecion procedures, (2) ransacion prices insead of offer prices, (3) quaniies sold, and (4) addiional feaures of he iems (e.g. some price-deermining qualiaive characerisics). Saisics Neherlands is one of he pioneers in he use of scanner daa for price measuremen purposes. Oher counries benefi from he experience and innovaions ha Saisics Neherlands accumulaed over wo decades. In a recen paper, Chessa (2016a) builds on his experience and proposes addiional innovaions. He describes a procedure for handling and using scanner daa in a more efficien and coheren way. For simpliciy, his procedure is denoed here as he C-approach. I is applicable o differen ypes of scanner daa and differen ypes of goods and services. Scanner daa provide he price analys wih an unusually rich se of informaion. The C-approach is a sophisicaed mehod ha is able o uilize his informaion. The price change beween he base monh and some comparison monh is compued as he raio of he average price level of he comparison monh and he average price level of he base monh. This is an alernaive o he orhodoxy of price index heory, ha is, o he averaging of he individual iems price raios ( 2 ). The C-approach is a valuable conribuion o he discussion on he mos aracive price measuremen mehodology. The proposed mehods have been esed wih scanner daa provided o Saisics Neherlands. Therefore, he C-approach is no jus an inellecual exercise, bu a esed procedure for a more efficien and coheren compuaion of reliable inflaion numbers. The presen paper aemps o conribue in wo ways. Firs, i provides a sreamlined exposiion of he C-approach ogeher wih a horough analysis of is srenghs and weaknesses. Second, a recommendaion for a simplified approach is developed. Such a simplificaion is desirable, because i would reduce he obsacles o a wider inegraion of scanner daa informaion ino official price measuremen. The C-approach is a wo sage procedure, where, on he lower sage, iems are aggregaed ino producs and, on he upper sage, hese producs are aggregaed ino a segmen. A he lower sage, for each produc and each monh a represenaive price is compued by he uni value formula. In addiion, he oal monhly quaniy for each produc is calculaed. A he upper sage, his informaion is used o compue he average price change beween a (fixed) base period monh and he mos recen monh. For his purpose he Geary-Khamis index is applied. This is a mulilaeral price index formula ha requires a simulaneous soluion of several equaions. Wih compuers, he soluion can be obained by ieraive compuaional procedures. The presen paper propose o replace he Geary-Khamis index by a simpler mulilaeral index formula or by an augmened bilaeral index formula. For example, he mulilaeral Gerardi index does no require ieraive soluion procedures. Alernaively, one may use an augmened ( 2 ) As described in Auer (2009, p.3), his orhodoxy can be raced back o Fisher s (1922, p. 451) seminal work. 74 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

77 Processing scanner daa by an augmened GUV index 3 version of some generalized uni value index described in Auer (2014). The generalized uni value indices are bilaeral. Therefore heir compuaion is much simpler han ha of he Geary- Khamis index. If he resuls generaed by such simplified procedures do no significanly deviae from he resuls from some suiable benchmark procedure, he simplified procedures may represen an aracive alernaive o he Geary-Khamis index. The paper is organized as follows. Secion 2 explains he C-approach in more deail. A discussion of ha approach can be found in Secion 3. Secion 4 proposes he simplified approaches and Secion 5 concludes. 2. Two sage aggregaion To undersand he basic idea of he C-approach, is srenghs and weaknesses, and he modificaions proposed in his paper, a few preliminary remarks on he processing of scanner daa a Saisics Neherlands are necessary. Suppose ha Saisics Neherlands received he scanner daa on ransaced men s T-shirs. The daa cover many differen men s T-shirs. Each specific T-shir is denoed as an iem (e.g. C&A package of wo whie coon men s T-shirs wih shor sleeves and round neck). Each iem covered by he scanner daa has is own idenificaion number, denoed as GTIN. Saisics Neherlands bundles iems (GTINs) ha are sufficienly similar ino producs. For example, all packages of wo whie coon men s T-shirs wih shor sleeves are grouped ino he produc package of wo whie coon men s T-shirs wih shor sleeves ; see Figure 1. This is he lower sage of aggregaion. This sage of aggregaion will be discussed in Secion 2.1. Producs ha are closely relaed o each oher are bundled ino a (consumpion) segmen. For example, all producs represening men s T-shirs are grouped ino he segmen men s T-shirs ; see Figure 1. This is he upper sage of aggregaion. I is sudied in Secion Lower sage of aggregaion: bundling iems ino producs Suppose ha we are employed by Saisics Neherlands and ha we have o compue he average price change of some segmen s (e.g. men s T-shirs). The provider of he scanner daa has supplied us wih sufficien informaion such ha we can compue for each iem j belonging o segmen s a monhly represenaive price, p j, and he monhly quaniy ransaced, q j. Furhermore, he daa conain aribues describing he iems (e.g. brand, fabric,...). The daa cover several monhs, ranging from he base monh = 0 o he mos recen monh = T. In principle, we could aggregae all daa in one sep o compue he average price change of segmen s beween monhs 0 and T. As we will argue below, his single sage procedure would be he preferable approach, if he se of iems remained consan over ime. However, many segmens wih unsable ses of iems exis. Therefore, he C-approach applies he wo sage procedure depiced in Figure 1. A he lower sage of aggregaion he segmen s iems are grouped ino producs. Which iems j of segmen s should be aggregaed ino a produc? For example, which iems represening men s T-shirs should be bundled ino a produc? In he C-approach, iems are bundled ino producs according o heir aribues (e.g. T-shirs per package, brand, fabric, colour, lengh of sleeves). EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 75

78 3 Processing scanner daa by an augmened GUV index Figure 1: The wo sage aggregaion procedure of Saisics Neherlands. e.g. men s T-shirs Upper sage of aggregaion Segmen e.g. package of wo whie coon men s T-shirs wih shor sleeves Lower sage of aggregaion Produc 1 Produc 2 Iem 1 Iem 2 Iem 3 Iem 4 e.g. C&A package of wo whie coon men s T-shirs wih shor sleeves and round neck The erm characerisic refers o a specific value ha an aribue can ake (Chessa (2016a, p 53)). For example, C&A and Hugo Boss are differen characerisics of he aribue brand. In he experimenal daa se used in he paper all packages of wo whie coon men s T-shirs wih shor sleeves are aggregaed ino he produc package of wo whie coon men s T-shirs wih shor sleeves. The choice of his definiion is based on a sensiiviy analysis: an aribue was deemed as imporan for defining a produc when he exclusion of ha aribue significanly changed he price index for he whole segmen men s T-shirs (Chessa (2016a, p 54)). This sensiiviy analysis idenifies a se of relevan aribues. Iems ha, wih respec o hese relevan aribues, show he same characerisics, are grouped ino one produc. Afer he grouping of iems ino producs is compleed, for each monh and each produc i of segmen s, some represenaive price, p i, and some aggregae quaniy, q i, is compued. Le A i denoe he se of iems j of segmen s ha in monh are bundled ino produc i. For each monh and each produc i, he C-approach compues he represenaive price, p i, as he uni value (Segniz (1870, p. 184)): (1) p i = j A i p j q j j A i q j. The aggregae quaniy, q i, is given by (2) q i = q j. j A i Noe ha a his lower sage of aggregaion no price changes or price indices are compued. Only in he upper sage of aggregaion he informaion on he uni values, p i, and aggregae quaniies, q i, is aggregaed ino he average price change of segmen s. To his upper sage we urn now. 76 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

79 Processing scanner daa by an augmened GUV index Upper sage of aggregaion: bundling producs ino segmens In his upper sage of aggregaion wo asks are execued simulaneously. Differen producs are aggregaed ino segmens and, for ha segmen, an average price change is compued. In his upper sage aggregaion he C-approach uses an index formula labeled as qualiy adjused uni value index (QAUV index). To undersand his index formula, i is necessary o undersand he uni value index (UV index) inroduced by Drobisch (1871a, p 39; 1871b, p 149) ( 3 ). Le G s denoe he se of producs i ha are assigned o segmen s. For each produc i and each monh, he corresponding uni value, p i, and he oal number of unis ransaced, q i, are known. They were compued a he lower sage of aggregaion by formulae (1) and (2). Also he aribues describing he produc (e.g. package of wo whie coon men s T-shirs wih shor sleeves) is known. We wan o compue for segmen s he average price change beween he base monh, = 0 (e.g. December 2014), and he mos recen monh, = T (e.g. December 2015). As one possibiliy, we could compue for monh = 0 and also for monh = T he segmen s uni value: (3) P = UV i Gs p i q i i G q s i for = 0, T. 0 T This gives us he segmen s wo uni values P and P. UV T/0 T 0 The UV index, P, is he raio of he wo uni values P and P. I measures he average price UV UV UV level change of segmen s beween monhs 0 and T: UV (4) P T/0 = UV P T UV P 0 UV = ( i G s T p it q i T ) / ( i Gs T q i T ) ( i G s 0 pi0 q i 0 ) / ( i G s 0 qi 0 ) = i G s T p it q i T i Gs 0 q i 0 i G s 0 pi0 q i 0 i G s T qi T Noe ha he ses G s T and G s 0 can differ.. In he UV index (4), he produc quaniy summaions, i G s q i, yield accurae resuls only if he produc-idenifying unis being summed are idenical. However, a segmen comprises producs of very differen qualiy. These producs have dissimilar produc-idenifying unis. Consequenly, hey are unsuiable for he simple quaniy summaions in he UV index (4). For example, in he segmen of men s T-shirs i is inappropriae o add 3 packages of black Hugo Boss T-shirs and 2 packages of whie C&A T-shirs o ge 5 packages of T-shirs. Therefore, he UV formula (3) and he UV index (4) are inappropriae. An amended version of he uni value concep is required. This is accomplished by he QAUV index. The QAUV index includes qualiy adjusmen facors, z i. They are defined o be an appropriae number of common unis per produc-idenifying uni. For example, if consumers considered he package of Hugo Boss T-shirs as hree imes as valuable as he package of C&A T-shirs, one could use z Hugo Boss = 3 and z C&A = 1. Wih hese numbers, he common uni would be equivalen ( 3 ) The following exposiion of he UV index and he QAUV index draws on Auer (2014). EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 77

80 3 Processing scanner daa by an augmened GUV index o one package of C&A T-shirs. Then, one package of Hugo Boss T-shirs would represen hree common unis. Of course, z Hugo Boss = 2 and z C&A = 2/3 would be equally appropriae. Then, he common uni would be wo hirds of a package of C&A T-shirs. Accordingly, he qualiy adjused prices, p i / z i, become moneary unis per common uni of produc i and he qualiy adjused quaniies, q i z i, are he number of common unis ransaced in he form of produc i. By definiion, he common unis of all producs i in se G s are idenical and, for ha reason, reliable resuls are now obained from he quaniy summaions, q i z i. The formula for he qualiy adjused uni value becomes (5) P = i G s (p i /z i )q i z i i Gs q i z i = i Gs p i q i. i Gs q i z i Noe ha he value aggregae, i G s p i q i, remains unaffeced by his ransformaion, since he qualiy adjusmen facors, z i, cancel. Formula (5) replaces formula (3). Formula (5) is he backbone of he upper level aggregaion of he C-approach (Chessa (2016a, pp 56-57)) ( 4 ). Consequenly, he QAUV index is: (6) P T/0 = QAUV P T P 0 = i G s T p it q i T i Gs 0 q i0 z i. i G s 0 pi0 q i 0 i G s T qit z i This index measures he change in he uni value of a common uni in segmen s beween monhs 0 and T ( 5 ). If he daa required for he deerminaion of he qualiy adjusmen facors, z i, is no direcly observable, hen some form of esimaion is required. The simples esimaion procedure is o use an iem s observed prices, p i, as indicaors of he iem s qualiy, z i. For example, if he uni value of a package of Hugo Boss T-shirs in boh monhs is EUR 40, whereas he uni value of a package of C&A T-shirs in boh monhs is only EUR 10, hen z Hugo Boss = 4 and z C&A = 1 are reasonable qualiy adjusmen facors. This approach ress on he assumpion ha he iems prices adequaely reflec he iems differences in qualiy. There is some lieraure ha suggess ha his assumpion is ofen violaed. For example, when firms pursue marke skimming, iems are overpriced a he beginning of he life cycle. The sraegy of marke peneraion leads o he opposie effec. Also Chessa (2016a, p 57) assumes ha an iem s price adequaely reflecs he iem s qualiy. Then he producs uni values can be exploied o quanify he differences in he producs qualiy. The idea was already menioned in de Haan (2002, p 82). He recommends using a ime period in which he producs were being sold in he markeplace and hey were preferably in a sae of equilibrium. Furhermore, Auer (2014) provides a deailed elaboraion and generalizaion of ( 4 ) Chessa (2016a, p 9) denoes he qualiy adjusmen facors, z i, as quaniy weighs and uses he symbol v i. I is cerainly rue ha he qualiy adjusmen facors boil down o quaniy weighs. However, ha erm may hide he rue naure of he qualiy adjusmen facors: hey ransform prices and quaniies relaed o produc-idenifying unis ino prices and quaniies relaed o common unis. Noe also ha in he price index lieraure, he symbol v i is usually reserved for he resul of muliplying price by quaniy: v i = p i q i. Therefore, in he presen paper we denoe he adjusmen facors by z i insead of Chessa s v i. ( 5 ) The QAUV index (6) was originally proposed by Dalén (2001, p 11) and de Haan (2002, pp 81-82). Addiional elucidaions, ogeher wih some empirical applicaions, can be found in de Haan (2004, pp 6-7). A relaed proposal is provided by Silver (2010, p S220). A deailed elaboraion of he QAUV index can be found in Auer (2014). Chessa (2016a, p 57) poins ou ha he Laspeyres index and Paasche index are special cases of he QAUV index formula (5). This was shown already in Auer (2014, pp ). 78 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

81 Processing scanner daa by an augmened GUV index 3 his basic idea. He also poins ou ha he basic idea can be raced back o Lehr (1885, pp 37-39). However, publishing in he German language, Lehr s work wen largely unnoiced. Accordingly, Davies (1924, pp ) some fory years laer came up wih he same basic idea ( 6 ). If he qualiy adjusmen facors, z i, were compued only from daa associaed wih monhs 0 and T, he daa of all oher monhs ( = 1, 2,, T 1) would be redundan and formula (5) would be applied only o monhs 0 and T. However, as oulined below, in he C-approach he qualiy adjusmen facors, z i, are compued from daa of all monhs. This proposal is new o he lieraure. Formula (5) is he basis of he upper sage aggregaion of he C-approach. The formula compues, for each monh ( = 0, 1,, T) he qualiy adjused uni value of segmen s. In Equaion (6), he formula is used for comparing he price levels of monhs = 0 and = T. This is a binary comparison. However, formula (5) can be applied also for mulilaeral comparisons, and he C-approach uses such a mulilaeral procedure ( 7 ). Usually, mulilaeral indices are used for inerregional price comparisons where he price levels of many differen regions mus be compared in a consisen way. However, Balk (1981) realizes ha such mulilaeral indices can be applied also in ineremporal price comparisons where more han wo periods are o be compared. His idea is revived by de Haan and van der Grien (2011) as well as by Ivancic e al (2011). By employing a mulilaeral price index hey avoid a bias ha is known as chain drif. This bias can arise from chained bilaeral price indices. The C-approach does no use chaining. Insead i considers a siuaion wih hireen consecuive monhs, wih he fixed base monh being he December of some year and he final monh being he December of he following year. A well known mulilaeral index formula is he Geary-Khamis index (Geary (1958), Khamis (1972)). This index uses formula (5) and compues he qualiy adjusmen facor, z i, of each produc i in he following way: (7) z i = T φ i =0 p i P, wih (8) φ i = q i T q r r=0 i. Equaion (7) is a weighed average of he uni values, p i, of all T + 1 monhs where he monhly weigh, φ i, is he quaniy share of ha monh, and he uni value, p i, of each monh, is deflaed by P, he qualiy adjused uni value of he whole segmen s during ha monh. The deflaion by P is he reason why formula (5) mus be applied o all monhs simulaneously and no jus o monhs 0 and T. In sum, he Geary-Khamis index uilizes formula (5) o compue a sequence of qualiy adjused uni values: P 0, P 1,, P T. The calculaion of each qualiy adjused uni value, P, is no jus based on he daa of monh bu also on he daa of all oher monhs. This can be seen from formulae (7) and (8). They compue he qualiy adjusmen facors, z i, from he daa of all monhs. ( 6 ) A sysemaic elaboraion of he proposiions of Lehr, Davies and de Haan is provided in Auer (2014). Following Lehr (1885) and Davies (1924), he explains ha formula (6) has a much wider use han enabling he aggregaion of similar producs. Therefore, he labels formula (6) as he family of Generalized Uni Value indices (GUV index) and he qualiy adjusmen facors, z i, as ransformaion raes. ( 7 ) For surveys on he lieraure on mulilaeral price comparisons see, for example, Balk (2008) or Auer (2012). EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 79

82 3 Processing scanner daa by an augmened GUV index Formula (5) reveals ha he qualiy adjused uni values, P, depend on he qualiy adjusmen facors, z i, whereas formula (7) reveals ha he opposie dependency is also rue. Therefore, he values of P ( = 0, 1,, T) and z i (i { G s 0 G s 1 G s T ) mus be compued simulaneously. Compuers allow us o derive he resuls by an ieraive procedure ha is described, for example, in Maddison and Rao (1996, pp 14-17). In each monh, he C-approach calculaes a Geary-Khamis index ha includes all monhs from he fixed base monh (December of he previous year) o he mos recen monh for which daa are available. For example, a he end of January 2015 only he daa of January 2015 and he fixed base monh December 2014 are included in he Geary- Khamis index calculaion (he Geary-Khamis index simplifies o a bilaeral index). This yields he qualiy adjused uni values Dec P (Jan), and P (Jan), where he brackes in he subscrip indicae he las monh for which daa were available, here January The average price change is P Jan/Dec GK(Jan) (Jan) = P Jan P Dec (Jan), Jan/Dec where GK sands for Geary-Khamis. The index value P GK(Jan) is he official measure of he price change of segmen s beween January 2015 and he base monh December In February, he calculaion of he Geary-Khamis index is based on he daa of December, January, and February. This yields new qualiy adjusmen facors, z i, and herefore, new qualiy adjused Dec Jan Jan/Dec Feb/Dec uni values P (Feb), P (Feb), and P (Feb). As a consequence, one ges new price changes P GK(Feb), P GK(Feb), Feb/Jan Feb/Dec and P GK(Feb). The price change P GK(Feb) is he official measure of he price change of segmen s beween February 2015 and he base monh December For March and each subsequen monh unil December he same procedure is execued. Finally, his yields a sequence of welve Jan/Dec Feb/Dec Dec/Dec index numbers: P GK(Jan), P GK(Feb),, P GK(Dec). This sequence Chessa (2016a, p 58) denoes as he real ime index and i is his index ha is published by Saisics Neherlands as he official price index of segmen s. Dec/Dec Only in he December 2015 index compuaion (P GK(Dec) ) he informaion of all 13 monhs is included. In ha monh 12 x 12 index numbers are compued. Among hese index numbers is he sequence of monhly index numbers ha use December 2014 as heir consan base: Jan/Dec Feb/Dec Dec/Dec P GK(Dec), P GK(Dec),, P GK(Dec). Chessa (2016a, p 58) denoes his sequence of index numbers as he benchmark index, because each of hese index numbers uilizes he complee informaion available in Laer in he paper, he benchmark index is compared o he real ime index (Chessa (2016a), pp 61-62). No significan deviaions occur. In 2016 welve new index numbers are compued. However, hey use December 2015 as heir fixed base monh. This complees he descripion of he C-approach. The following secion discusses he srenghs and weaknesses of his procedure. 3. Discussion The C-approach compues he average price change of he iems j belonging o segmen s beween he base monh = 0 and he mos recen monh = T. As described in he previous secion, his is accomplished in a wo sage procedure. A he lower sage of aggregaion, iems are grouped ino producs. For each monh and each produc i, he C-approach compues a uni value, p i, and an aggregaed quaniy, q i. This is done by formulae (1) and (2). A he upper 80 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

83 Processing scanner daa by an augmened GUV index 3 sage of aggregaion, he C-approach aggregaes he uni values, p i, and aggregaed quaniies, q i, of he producs of segmen s ino he T + 1 price levels P ( = 0, 1,, T). For his purpose, (T) he C-approach uilizes he Geary-Khamis index, ha is, formulae (5), (7), and (8). Finally, he raio P T/0 = GK(T) P T (T) P 0 (T) is he official measure of he average price change of segmen s beween monhs = 0 and = T. Two feaures of his C-approach require a more elaborae discussion: (A) The approach applies a wo sage aggregaion insead of a single sage aggregaion and (B) a he upper sage of aggregaion he index compuaion beween some fixed base monh = 0 and he mos recen monh = T uses no only he informaion of hese wo monhs, bu relies on he Geary-Khamis index which uilizes also he informaion provided by all in-beween monhs. The following discussion sars wih he second feaure and hen moves on o he firs feaure Geary-Khamis index The Geary-Khamis index calculaes for each monh ( = 0, 1,, T) a qualiy adjused uni value P. Again, he brackes in he subscrip of he uni value indicae he las monh included in is (T) calculaion. When T = 2, he qualiy adjused uni values P, P, and P are compued. These (2) (2) (2) hree numbers allow for hree price comparisons: P 1/0 = GK(2) P 1, P 2/0 = P 2, and P 2/1 = (2) (2) (2) P 2. P 0 GK(2) P 0 GK(2) P 1 (2) (2) (2) Therefore, he Geary-Khamis index is called a mulilaeral price index. Noe also ha P 2/0 = P 1/0 P 2/0. GK(2) GK(2) GK(2) Therefore, he Geary-Khamis index is classified as a ransiive mulilaeral price index. Transiiviy allows us o compare he qualiy adjused uni values of any pair of monhs in a consisen way. As poined ou before, he compuaion of he Geary-Khamis index is no easy and requires an ieraive compuer procedure. However, oher mulilaeral price indices exis ha, due o heir similariy o he Geary-Khamis index, are likely o generae almos idenical resuls, bu are much simpler o compue. For example, he Gerardi index proposed in Eurosa (1978, p 30) is similar o he Geary-Khamis index bu replaces formulae (7) and (8) by he much simpler formula (9) z i = T Π =0 (p i ) 1/T. This is he geomeric average of all T + 1 uni values, p i, of produc i ( 8 ). In conras o he Geary-Khamis index, he qualiy adjusmen facors, z i, of he Gerardi index are independen from he qualiy adjused uni values P. As a consequence, he simulaneous (T) compuaion of he z i -values and he P -values is no longer necessary. This simplifies he (T) ( 8 ) Of course, oher mulilaeral indices exis ha are also easier o compue han he Geary-Khamis index. An example is he GEKS-index proposed by Gini (1924), Éleő and Köves (1964) and Szulc (1964). However, his index has a compleely differen consrucion principle han he Geary-Khamis index and he Gerardi index (e.g. Balk (2008); Auer (2012)). EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 81

84 3 Processing scanner daa by an augmened GUV index calculaion considerably and i eases he implemenaion of scanner daa procedures in he compuaional rouines of he naional saisical offices. Does he omission of he qualiy adjused uni values, P, inroduce a source of bias in he (T) compuaion of he qualiy adjusmen facors, z i? In inerregional price comparisons he uni values may vary significanly. This is paricularly rue when he regions have differen currencies. However, in he C-approach we compare monhs insead of regions. Usually, he variance of he values of P is much smaller han ha in inerregional price comparisons. Therefore, we are (T) unlikely o injec any significan bias when we compue he z i -values wihou he P -values. (T) Probably, he Gerardi index will give very similar resuls as he Geary-Khamis index. Admiedly, his conjecure remains o be empirically verified wih scanner daa such as hose of Saisics Neherlands ( 9 ). In each consecuive monh, he C-approach compues a new Geary-Khamis index. This yields a sequence of Geary-Khamis indices. Sequence of Geary-Khamis indices means ha in each consecuive monh a new se of qualiy adjused uni values is compued. For example, in monh T 1 he qualiy adjused uni values P 0, P 1,, P T 1 (T 1) (T 1) (T 1) are compued and from hose he indices P 1/0, P 2/0,, P T 1/0. GK(T 1) GK(T 1) GK(T 1) In he following monh he qualiy adjused uni values P 0, P 1,, P T 1, P T (T) (T) (T) (T) are calculaed and from hose he indices P 1/0, P 2/0,, P T 1/0, P T/0. GK(T) GK(T) GK(T) GK(T) Usually one ges P 1/0 P 1/0, P 2/0 P 2/0,. GK(T 1) GK(T) GK(T 1) GK(T) In oher words, he resuls produced by he sequence of Geary-Khamis indices are inconsisen. The C-approach solves his inconsisency problem by simply keeping only he adjused uni 1/0 2/0 T/0 values P GK(1), P GK(2),, P GK(T). This is Chessa s (2016a, p 58) real ime index. To ge a his real ime index, a large se of inconsisen qualiy adjused uni value indices is calculaed and hen, o ensure consisency, mos of hem are discarded. If one used a sequence of Gerardi indices insead of a sequence of Geary-Khamis indices, he inconsisency problem would no go away. I is he sequence of mulilaeral indices ha causes he problem, no he applied index formula. ( 9 ) In privae correspondence Anonio Chessa repored o me ha for clohing he resuls generaed by he Gerardi index show some deviaions o he Geary-Khamis resuls. 82 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

85 Processing scanner daa by an augmened GUV index 3 To sum up, for segmens wih sable price levels he Geary-Khamis index appears unnecessarily complex. If he resuls are sufficienly similar, i could be replaced by he much simpler Gerardi index. Neverheless, an inconsisency problem would remain. However, an even simpler procedure avoids he problem of inconsisen price indices. This procedure is described in he recommendaions (Secion 4) ( 10 ) Two sage aggregaion For compuing he average price change of some segmen s, he C-approach relies on a wo sage aggregaion. A he lower sage, he prices and quaniies of iems are aggregaed ino uni values of producs, p i, and aggregae quaniies, q i. This is done for each monh separaely. A he upper sage of aggregaion, he Geary-Khamis index is used. This index aggregaes he uni values, p i, and aggregaed quaniies, q i, of he producs of segmen s ino he price levels T/0 T 0 P ( = 0, 1,, T). Finally, he raio P = P /P is he official measure of he average price (T) GK(T) (T) (T) change of segmen s beween monhs = 0 and = T. Suppose ha he se of iems (GTINs) remained consan over ime: A s0 = A s1 = =A st = A s. In such a scenario he wo sage aggregaion would offer no advanages whasoever over a single sage aggregaion ha reas every iem as an individual produc. Treaing each iem j as a disinc produc eliminaes he firs sage of he wo sage aggregaion and he qualiy adjused uni value formula (5) becomes (10) P = j A s p j q j j As q j z j, where A s is he se of iems j belonging o segmen s. This single sage formula replaces he hree formulae (1), (2), and (5). Coupling formula (10) wih he Geary-Khamis formulae (7) and (8) 1/0 2/0 T/0 or wih he Gerardi formula (9) gives he desired real ime index: P, P,, P. In fac, wih a consan se of iems, he single sage aggregaion is more reliable han he wo sage aggregaion, because i avoids he arificial grouping of iems ino producs ha are considered as homogeneous bu, in realiy, are no. For example, in he example given in Chessa s paper, he produc package of wo whie coon men s T-shirs wih shor sleeves can comprise C&A and Hugo Boss packages and couns hem as equivalen packages. However, for he measuremen of price change i maers wheher he aggregaed quaniy, q i, consiss of C&A or of Hugo Boss T-shirs. For example, suppose ha all T-shir prices remain consan over ime, bu ha he share of Hugo Boss T-shirs increases over ime. If C&A T-shirs and Hugo Boss T-shirs are assigned o he same produc, formula (1) yields an increasing uni-value for ha produc, falsely indicaing a price increase. As a consequence, he average price index compued by (5) is upward biased. In he following, he bias caused by assigning heerogeneous iems o he same produc is denoed as he assignmen bias ( 11 ). This bias can be upward or downward. I can arise only in he firs sage of he C-approach s wo sage aggregaion. In he single sage aggregaion conduced by formula (10) no assignmen bias can arise, because no assignmen is necessary. Of course, he C-approach was explicily designed for siuaions in which he universe of iems changes over ime. Chessa (2016a, p 56, Figure 3) presens an empirical example o show ha ( 10 ) The Geary-Khamis index and he alernaives proposed in he presen paper are no he only approaches for processing scanner daa. Recen surveys include Ausralian Bureau of Saisics (2016), Chessa e al (2017), and Diewer and Fox (2017). ( 11 ) This effec is ofen denoed as uni value bias. However, his erminology is somewha misleading. The problem is no he uni value formula, bu he flawed assignmen of iems o he group of iems for which a uni value is compued. (1) (2) (T) EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 83

86 3 Processing scanner daa by an augmened GUV index in he presence of srong changes in he se of iems he wo sage aggregaion is less prone o bias han a single sage aggregaion. Therefore, we firs discuss under which condiions he enry and exi of iems causes problems for he single sage aggregaion. Aferwards we examine wheher he wo sage aggregaion is able o avoid such problems. If an iem j is replaced by he same iem k, bu he replacemen has a differen GTIN, he replacemen is called a relaunch of iem j. If he analys is able o idenify he relaunch, he wo iems receive he same qualiy adjusmen facors, z j = z k. No problem arises. The single sage aggregaion remains more reliable han he wo sage aggregaion, because no assignmen bias arises. Wha happens in he single sage aggregaion when here is an undeeced relaunch? If he prices of he enering and exiing iem are idenical and sable over ime, here are sill no advanages from he wo sage aggregaion, because in he single sage aggregaion we ge z j = z k. This is rue for he Geary-Khamis formulae (7) and (8) as well as for he Gerardi formula (9). However, when an undeeced relaunch wih deviaing prices occurs or when an iem is replaced by a compleely differen iem, some bias can arise in he single sage aggregaion. Suppose, for example, ha he price of iem j is consan over ime and hen, in monh = T, iem j is replaced by an equivalen relaunch iem k ha has a higher price. Since he relaunch is no deeced, formula (9) would yield z j < z k insead of z j = z k. In he denominaor of formula (10), he sum 0q j0 z j would be oo small relaive o he sum T j A s j A s q jt z j. Therefore, he price level P 0 would be oo large relaive o P T. As a consequence, he measured average price change of he segmen, P T / P 0, would be downward biased. This ype of bias is denoed here as assormen bias. If iem j were replaced by an equivalen relaunch iem k ha has a lower price, he assormen bias would be upward. The C-approach wih is wo sage aggregaion is less prone o assormen bias han he single sage aggregaion. If he analys assigns he new iem k o a produc i ha accuraely reflecs iem s k characerisics, hen iem k receives his produc s qualiy adjusmen facor, z i, and he assormen bias can be significanly reduced. Of course, an appropriae assignmen of he new iem k o he bes fiing produc requires a careful analysis of he new iem. In oher words, in he wo sage aggregaion, he analys mus inves ime and effor ino his assignmen process. In a single sage aggregaion his ime and effor could be spend on deecing relaunches, ha is, on comparing exiing and enering iems. Wheher his ime and effor would sufficienly reduce he assormen bias of he single sage aggregaion is unclear. The wo sage aggregaion has anoher advanage. Someimes iems drop ou of he sample wihou a relaunch. As long as he produc o which his iem was assigned sill conains oher iems, his does no represen a problem for he price measuremen. Similarly, new iems may ener he sample wihou an equivalen predecessor. If some produc exiss o which he new iem can be assigned o, he inclusion of he new iem causes no problems. In a single sage aggregaion he enry and exi of iems raises he usual problems. If he iems are simply ignored, bias could arise. If he iems are no ignored, he missing prices mus be esimaed. In sum, he C-approach s wo sage aggregaion defined by formulae (1), (2), and (5) suffers from assignmen bias, bu, wih careful analyss, may keep he assormen bias in check. The single sage aggregaion defined by formula (10) does no suffer from assignmen bias. However, i is prone o considerable assormen bias. The less sable he se of iems over ime, he more aracive becomes he wo sage aggregaion procedure of he C-approach relaive o a single sage aggregaion procedure. 84 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

87 Processing scanner daa by an augmened GUV index 3 4. Recommendaions The C-approach ackles some imporan problems in price measuremen. The downside o his is he complexiy of he C-approach. Can we ackle he problems by less complex approaches? This final secion oulines a few ideas. Recommendaion 1: augmened GUV index insead of Geary- Khamis index In is upper sage of aggregaion he C-approach compues he average price change of some segmen s beween some fixed base monh = 0 and he mos recen monh = T. For his purpose i uses he Geary-Khamis index, ha is, formulae (5), (7), and (8). The laer wo formulae ensure ha he compuaion of he qualiy adjusmen facors, z i, uses no only he informaion provided by monhs 0 and T, bu also he informaion of all oher monhs (1, 2,, T 1). This is a valuable innovaion of he C-approach. However, here are simpler ways of including he informaion of monhs = 1, 2,, T 1 ino he compuaion of he qualiy adjusmen facors, z i. In Secion 3.1 of his paper i was explained ha here are wo disadvanages associaed wih he Geary-Khamis index. Firs, he compuaion of a Geary-Khamis index is complex, and second, he sequence of Geary-Khamis indices produces a huge number of incoheren qualiy adjused uni values. Very few of hese are necessary o ge he real ime index. The firs disadvanage could be miigaed by using he Gerardi index insead of he Geary-Khamis index. However, he second disadvanage would remain even wih he Gerardi index. Can we compue he qualiy adjusmen facors, z i, by a simpler procedure as he Geary-Khamis or Gerardi index, and sill uilize he informaion provided by all in-beween monhs? A good saring poin for such a procedure is he family of Generalized Uni Value indices (GUV indices). This family of price index formulae is defined and explored in Auer (2014). The Lehr index is an example. Insead of (7) and (8), Lehr (1885, p 39) compues he qualiy adjusmen facors, z i, by he simple uni value formula (11) z i = p i0 q i0 +p it q i T q i0 +q i T. These qualiy adjusmen facors are insered in formula (5) o compue he qualiy adjused uni values P 0 and P T. The raio of hese wo uni values is he Lehr index. Like he Geary-Khamis and he Gerardi index, i measures he price change beween he base monh = 0 and he mos recen monh = T. However, i does no calculae he (irrelevan) price changes beween all oher monhs. Therefore, he inconsisencies discussed in Secion 3.1 of his paper canno arise. Closer inspecion of he Lehr index and he Geary-Khamis index reveals ha beween hese wo indices a close relaionship exiss. Suppose we apply he Geary-Khamis index for a bilaeral price comparison of monhs = 0 and = T. Tha is, only informaion of monhs = 0 and = T is processed: P 1/0 = GK(2) i Gs T p it q i T i Gs 0 p i0 q i 0 i Gs 0 q i0 z i i Gs T q it z i EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 85

88 3 Processing scanner daa by an augmened GUV index (12) z i = φ i 0 p i 0 P 0 + φ i T p i T P T q i 0 (13) φ i0 = 1 q i0 + q i q i T and φ it = T q i0 + q i. Insering (13) in (12) and seing P 0 = P T = 1, (12) simplifies o (11). In oher words, applying he Geary-Khamis index o a bilaeral price comparison and seing he price levels, P T, in (12) equal o 1, yields he Lehr index. As poined ou before, when = 0 and = T represen monhs insead of regions wih differen currencies, seing in (12) P 0 = P T = 1 is unlikely o cause discernible bias ( 12 ). The Lehr index is simple and he calculaion of he qualiy adjusmen facors by formula (11) is inuiively appealing. However, he Lehr index and all oher GUV indices are bilaeral price indices. They uilize only he informaion of monhs = 0 and = T, disregarding he daa provided by monhs = 1, 2,, T 1. This is a disadvanage. I is usually preferable o make use of all sensible informaion ha is available. The compuaion of qualiy adjusmen facors, z i, is no excepion. Formulae (7) and (8) uilize all informaion, whereas formula (11) does no. Can his disadvanage of he GUV index family be avoided? The obvious soluion is o augmen he applied GUV index, such ha he calculaion of he qualiy adjusmen facors uilize also he informaion provided by monhs = 1, 2,, T 1. For example, we can compue he qualiy adjusmen facors by he formula (14) z i = T =0 T =0 p i q i q i which is a generalizaion of formula (11). The compued values can be insered in formula (5). 0 T This yields he qualiy adjused uni values P and P. The raio of hese wo numbers is he augmened Lehr index ( 13 (T) (T) ). This is sill a bilaeral price index. I gives he average price change beween he wo monhs 0 and T: P T/0 = alehr(t) P T (T). P 0 (T) In each new monh a new augmened Lehr index mus be compued ha compares he mos recen monh, = T, o he fixed base monh, = 0. For example, he price change of segmen s beween December 2014 and February 2015 is measured by P Feb/Dec alehr(feb) (Feb) = P Feb P Dec (Feb). The average price change beween March 2015 and December 2014 is given by ( 12 ) Applying he Gerardi index o a bilaeral price comparison yields he Davies index, anoher member of he family of GUV indices (see Auer (2014, pp ), Auer (2012, p 50)). ( 13 ) The same index has been independenly proposed in Lamboray (2017). 86 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

89 Processing scanner daa by an augmened GUV index 3 P Mar/Dec alehr(mar) (Mar) = P Mar P Dec (Mar). Noe ha in every monh he qualiy adjused uni value of he fixed base monh (here December 2014) is recompued. The oher qualiy adjused uni values are no recompued. Therefore, in a sequence of augmened Lehr indices far fewer compuaions are execued han in a sequence of Geary-Khamis or Gerardi indices. Furhermore, he inconsisencies discussed in Secion 3.1 canno arise. In official price saisics his is an imporan virue. I would be useful o know wheher he augmened Lehr index produces index numbers ha coincide wih hose generaed by a suiable benchmark index. Chessa (2016b, pp 15-16) empirically compares he augmened Lehr index wih he real ime index of he C-approach. He shows ha in segmens wih a coninuous downward rend in prices he augmened Lehr index shows a slighly differen resul han he real ime index. Furhermore, he proposes a bilaeral varian of he Geary-Khamis index (denoed as revisionless QU-GK index ). In he price level compuaion of period = T, he price levels obained from he compuaions of previous monhs (T 1, T 2, ) are no recompued, bu simply insered in Equaion (7). Then, insering (7) in he qualiy adjused uni value formula (5) gives for period = T an equaion wih P T as is only unknown. In addiion, Chessa (2016b, pp 16-17) shows for several segmens ha his bilaeral version of he Geary-Khamis index gives virually he same resuls as he real ime index and he benchmark index generaed by he C-approach. Recommendaion 2: single sage aggregaion insead of wo sage aggregaion for sable ses of iems Chessa (2016a) advocaes a wo sage aggregaion procedure (iems ino producs and producs ino segmens) insead of a single sage aggregaion (iems ino segmens). Obviously, a wo sage aggregaion is more complex han a single sage aggregaion. Therefore, i should be examined wheher he wo sage aggregaion brings more reliabiliy and, if he answer is yes, wheher his gain jusifies he increase in complexiy and inransparency. The wo sage aggregaion was defended on he grounds ha i reduces assormen bias. However, as argued in Secion 3.2 of his paper, i inroduces assignmen bias. In oher words, assormen bias is replaced by assignmen bias, and wihou furher invesigaion i is unclear which of he wo is more damaging. The larger he changes in he se of iems, he sronger he case for he wo sage aggregaion. A single sage aggregaion does no give rise o assignmen bias. Furhermore, also in he conex of a single sage aggregaion, he assormen bias can be reduced. Comparing he characerisics of a newly enering iem k o he characerisics of he exising iems can yield sensible qualiy adjusmen facors z j. However, his requires a lo of ime and effor on he side of he analys. Furhermore, when iems drop ou of he sample wihou replacemen or new iems ener he sample wihou a predecessor, he single sage aggregaion should esimae he missing prices. This requires addiional ime and effor. In segmens suiable for he wo sage aggregaion, he assignmen bias should be avoided as far as possible. I is unclear, wheher he auomaed sensiiviy analysis described in Chessa s paper can achieve his. For example, i is no very convincing o drop he aribue brand on he grounds ha is inclusion did no change he price index of he segmen men s T-shir in a significan way. The aribue brand is highly relevan for characerizing a T-shir, regardless of EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 87

90 3 Processing scanner daa by an augmened GUV index he resuls produced by a sensiiviy analysis. Therefore, his aribue should be included, if i is available. Furhermore, he auomaed sensiiviy analysis is no well rooed in saisical heory. Which alernaive assignmen rule could be implemened? If here are doubs wheher an aribue affecs prices and herefore he delineaion of producs, an auomaed hedonic regression analysis appears aracive. However, if such procedures consume oo much ime and effor, he auomaed sensiiviy analysis of he C-approach is cerainly an opion. In sum, avoiding he assignmen bias of he wo sage aggregaion requires ime and effor on he side of he analys. Wih a single sage aggregaion, his ime and effor could be redireced o he reducion of assormen bias. Wheher his reducion is sufficienly large, is an open quesion ha can be answered only by pracical experience. Puing everyhing ogeher Recommendaion 1 can be implemened even wihou Recommendaion 2. Of course, for segmens wih a sable se of iems a join implemenaion of boh recommendaions appears preferable. Suppose ha he average price change of some segmen s beween he base monh = 0 and some comparison monh = T mus be compued. If he assormen bias can be kep on a low level, he average price change of segmen s can be calculaed in a single sage aggregaion direcly from he individual iems j. This can be accomplished by compuing he following wo qualiy adjused uni values: (15) P 0 = j As 0 p j0 q j 0 j As 0 q j0 z j and P T = j As T p jt q j T j As T q jt z j. In boh uni values he same qualiy adjusmen facors, z i, are o be used. They can be compued by a simple formula like =0 (16) z i = T p i q i T =0 q i. This formula uilizes all available informaion. Finally, he raio (17) P T/0 = alehr(t) P T (T) P 0 (T) gives he average price change of segmen s beween monhs = 0 and = T. Formulae (15) o (17) represen he augmened Lehr index. When he base period ( = 0) is December 2014, hen, a he end of January 2015, he above procedure is execued for T = Jan 2015 and he resuling index number is officially published. A he end of February 2015, he procedure is repeaed for T = Feb 2015, and so on. The las repeiion of he procedure is a he end of December This yields a sequence of 12 index numbers. No revisions are required. These 12 index numbers would replace he real ime index of he C-approach. When segmen s is characerized by high produc urnover and coninuous innovaions (e.g. mobile phones), he wo sage aggregaion of he C-approach becomes aracive. The single sage aggregaion is sill possible, hough. The compuaion of he qualiy adjusmen facors, z i, in formula (16) could use esimaed prices, ^p i, insead of observed prices, p i. The esimaed 88 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

91 Processing scanner daa by an augmened GUV index 3 prices, ^p i, can be obained from hedonic regression analysis or oher reliable procedures. This were a paricularly appealing procedure if he hedonic regression analysis could be parly auomaed. 5. Conclusions The C-approach proposed by Chessa (2016a) is a significan sep owards a more coheren and reliable processing of scanner daa. The approach uses a wo sage aggregaion procedure in which individual iems are bundled ino producs and hen producs ino segmens. On he lower sage, for each produc and each monh, a uni value and an aggregae quaniy is compued. These produc daa are used in he upper sage, where for each monh an average price level is compued. The raios of hese average price levels yield he price index numbers. The compuaions on he upper sage uilize he Geary-Khamis index. I is a major virue of he C-approach ha i uilizes all available informaion. For his purpose i proposes a raher complex wo sage procedure. The presen paper has argued ha in some siuaions he objecives of he C-approach can be achieved by somewha simpler means. Firs, he Geary-Khamis index could be replaced by a simpler procedure. Two opions are he Gerardi index and he augmened Lehr index. A hird promising opion, he revisionless QU-GK mehod, has been proposed by Chessa (2016b) himself. Preliminary empirical invesigaions show ha his index hardly deviaes from he benchmark index of he C-approach. Furhermore, he presen paper recommends o replace he sensiiviy analysis execued a he lower sage of aggregaion, by a mehod ha is deeper rooed in saisical heory (e.g. hedonic regression). Finally, his paper asks wheher he firs sage of he wo sage procedure is really necessary. Possibly, he problem of exiing and enering iems can be ackled also wihin a single sage procedure ha redirecs resources from he firs sage of aggregaion o he deecion and qualiaive evaluaion of enering and exiing iems. Such a move would avoid he C-approach s assignmen bias and, if carefully implemened, may avoid a significan increase in assormen bias. Wheher his is a viable alernaive remains o be seen. Acknowledgemens This paper originaed from a consulancy projec for he European Cenral Bank. The auhor wans o hank he paricipans of a workshop a Eurosa (Luxembourg, December 2016) and in paricular Tony Chessa and Claude Lamboray for heir helpful inpu. Furhermore, he commens of hree referees are graefully acknowledged. References Auer, L. von (2009), The Measuremen of Macroeconomic Price Level Changes, Universiä Trier: Research Papers in Economics, No. 1/09, Universiä Trier. EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 89

92 3 Processing scanner daa by an augmened GUV index Auer, L. von (2012), Räumliche Preisvergleiche: Aggregaionskonzepe und Forschungsperspekiven, ASA Wirschafs- und Sozialsaisisches Archiv, Vol. 6, issue 1, pp Auer, L. von (2014), The Generalized Uni Value Index, Review of Income and Wealh, Vol. 60, issue 4, pp Ausralian Bureau of Saisics (2016), Making Greaer Use of Transacions Daa o Compile he Consumer Price Index, Informaion Paper , Canberra. Balk, B.M. (1981), A Simple Mehod for Consrucing Price Indices for Seasonal Commodiies, Saisical Papers, Vol. 22, pp Balk, B.M. (2008), Price and Quaniy Index Numbers, Cambridge Universiy Press, Cambridge, New York. Chessa, A.G. (2016a), A New Mehodology For Processing Scanner Daa in he Duch CPI, Eurosa Review on Naional Accouns and Macroeconomic Indicaors, 1/2016, pp Chessa, A.G. (2016b), Comparisons of he QU-Mehod Wih Oher Index Mehods for Scanner Daa, paper prepared for a workshop a Eurosa, 7-8 December 2016, Luxembourg. Chessa, A.G., J. Verburg and L. Willenborg (2017), A Comparison of Price Index Mehods for Scanner Daa, paper prepared for he 15h Meeing of he Oawa Group, May 2017, Elville. Dalén, J. (2001), Saisical Targes for Price Indexes in Dynamic Universes, Proceedings of he Sixh Meeing of he Inernaional Working Group on Price Indices, Canberra, Ausralia. Davies, G.R. (1924), The Problem of a Sandard Index Number Formula, Journal of he American Saisical Associaion, Vol. 19, pp de Haan, J. (2002), Generalised Fisher Price Indexes and he Use of Scanner Daa in he Consumer Price Index (CPI), Journal of Official Saisics, Vol. 18, issue 1, pp de Haan, J. (2004), Esimaing Qualiy-Adjused Uni Value Indexes: Evidence from Scanner Daa, Paper presened a he SSHRC Inernaional Conference on Index Theory and he Measuremen of Prices and Produciviy, Vancouver, Canada. de Haan, J. and H.A. van der Grien (2011), Eliminaing Chain Drif in Price Indices Based On Scanner Daa, Journal of Economerics, Vol. 161, issue 1, pp Diewer, E.W. and K. J. Fox (2017), Subsiuion Bias in Mulilaeral Mehods for CPI Consrucion Using Scanner Daa, Discussion Paper 17-02, Vancouver School of Economics, The Universiy of Briish Columbia, Vancouver, Canada. Drobisch, M.W. (1871a), Ueber Mielgrössen und die Anwendbarkei derselben auf die Berechnung des Seigens und Sinkens des Geldwerhs, Beriche der mahemaisch-physicalischen Classe der Königlich Sächsischen Gesellschaf der Wissenschafen, Hef 1. Drobisch, M.W. (1871b), Ueber die Berechnung der Veränderungen der Waarenpreise und des Geldwerhs, Jahrbücher für Naionalökonomie und Saisik, Vol. 16, pp Éleő, Ö. and P. Köves (1964), On an Index Compuaion Problem in Inernaional Comparisons (in Hungarian), Saizikai Szemle, Vol. 42, pp Eurosa (1978), Comparison in Real Values of he Aggregaes of ESA-1975, Saisical Office of he European Communiies: Luxembourg. 90 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

93 Processing scanner daa by an augmened GUV index 3 Fisher, I. (1922), The Making of Index Numbers, Boson: Houghon Miin, 3rd. revised ed. 1927, newly prined 1997 by Pickering and Chao, London. Geary, R.C. (1958), A Noe On he Comparison of Exchange Raes and Purchasing Power Beween Counries, Journal of he Royal Saisical Sociey, Series A, Vol. 121, issue 1, pp Gini, C. (1924), Quelques Considéraions au Suje de la Consrucion des Nombres Indices des Prix e des Quesions Analogues, Meron, Vol. 4, issue 1, pp Ivancic, L., W.E. Diewer and K. J. Fox (2011), Scanner Daa, Time Aggregaion and he Consrucion of Price Indices, Journal of Economerics, Vol. 161, issue 1, pp Khamis, H. (1972), A New Sysem of Index Numbers for Naional and Inernaional Purposes, Journal of he Royal Saisical Sociey, Series A, Vol. 135, issue 1, pp Lamboray, C. (2017), The Geary Khamis Index and he Lehr Index: How Much Do They Differ?, paper prepared for he 15h Meeing of he Oawa Group, May 2017, Elville. Lehr, J. (1885), Beiräge zur Saisik der Preise insbesondere des Geldes und des Holzes, F. D. Sauerländer Verlag, Frankfur a.m. Maddison, A. and D.S.P. Rao (1996), A Generalized Approach o Inernaional Comparison of Agriculural Oupu and Produciviy, Research Memorandum GD-27, Groningen: Groningen Growh and Developmen Cenre. Segniz, E. (1870), Ueber die Berechnung der sogenannen Miel, sowie deren Anwendung in der Saisik und anderen Erfahrungswissenschafen, Jahrbücher für Naionalökonomie und Saisik, Vol. 14, pp Silver, M. (2010), The Wrong and Righs of Uni Value Indices, Review of Income and Wealh, Vol. 56, pp. S206 pp. S223. Szulc, B. (1964), Index Numbers of Mulilaeral Regional Comparisons (in Polish), Przeglad Saysiczny, 3, pp EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 91

94

95 4 Big daa ypes for macroeconomic nowcasing DARIO BUONO ( 1 ), GIAN LUIGI MAZZI ( 2 ), GEORGE KAPETANIOS ( 3 ), MASSIMILIANO MARCELLINO ( 4 ) AND FOTIS PAPAILIAS ( 5 ) Absrac: In his paper we presen a deailed discussion on various ypes of big daa which can be useful in macroeconomic nowcasing. In paricular, we review he big daa sources, availabiliy, specific characerisics and heir use in he lieraure. We conclude his paper idenifying he big daa ypes which could be adoped for real applicaions. JEL codes: C82, E27, E37, E47 Keywords: big daa, nowcasing, daa feaures ( 1 ) Eurosa. ( 2 ) reired from Eurosa. ( 3 ) King's College London. ( 4 ) Universià Bocconi, Ialy. ( 5 ) Queen's Universiy Managemen School, Queen's Universiy Belfas, UK and quanf research.

96 4 Big daa ypes for macroeconomic nowcasing 1. Inroducion The advancemens in compuer echnology during he las decades have allowed he sorage, organisaion, manipulaion and analysis of vas amoun of daa from differen sources and across differen disciplines, and here is nowadays an ever growing ineres in he analysis of hese big daa. In his paper, we summarize he resuls of an invesigaion of he specificiies of various big daa sources relevan for macroeconomic nowcasing and early esimaes, and we discuss he characerisics of big daa sources ha are relevan for heir applicaion in his conex. Perhaps no surprisingly given heir differen origins, he lieraure provides various definiions of big daa. One possibiliy o obain a general classificaion is o adop he 4 Vs classificaion, originaed by he IBM, which relaes o: (i) Volume (Scale of daa), (ii) Velociy (Analysis of sreaming daa), (iii) Variey (Differen forms of daa) and (iv) Veraciy (Uncerainy of daa). However, his classificaion seems oo general o guide empirical nowcasing applicaions. A second opion is o focus on numerical daa only, which can eiher be he original big daa or he resul of a ransformaion of unsrucured daa. Once daa have been ransformed, and following, e.g. Doornik and Hendry (2015), we can disinguish hree main ypes of big daa. Tall daases include no so many variables, N, bu many observaions, T, wih T >> N. This is for example he case wih ick by ick daa on seleced financial ransacions or search queries. Fa daases have insead many variables, bu no so many observaions, N >> T. Large cross-secional daabases fall ino his caegory, which is no so ineresing from an economic nowcasing poin of view, unless eiher T is also large enough or he variables are homogeneous enough o allow proper model esimaion (e.g., by means of panel mehods) and nowcas evaluaion. Finally, Huge daases, wih very large N and T, are he mos ineresing ype of daa in a nowcasing conex even if, unforunaely, hey are no so ofen available for economic variables. However, complexiy can help in expanding he N dimension, for example allowing for dynamics (lead-lag relaions), non lineariies and he micro-srucure of informaion (calendar issues for example). A hird possibiliy o classify big daa is o resor o some sor of official definiion. A paricularly useful axonomy is provided by he saisics division of he Unied Naions Economic Commission for Europe (UNECE), which idenifies big daa according o heir source. A radiional daa source, revamped by he IT developmens, is represened by Business Sysems ha record and monior evens of ineres, such as regisering a cusomer, manufacuring a produc, aking an order, ec. The process-mediaed daa hus colleced by eiher privae businesses (commercial ransacions, banking/sock records, e-commerce, credi cards, ec.) or public insiuions (medical records, social insurance, school records, adminisraive daa, ec.) is highly srucured and includes ransacions, reference ables and relaionships, as well as he meadaa ha ses is conex. Tradiional business daa is he vas majoriy of wha IT managed and processed, in boh operaional and BI sysems, usually afer srucuring and soring i in relaional daabase sysems. A novel daa source is represened by Social Neworks (human-sourced informaion). This informaion is he record of human experiences, by now almos enirely digially sored in personal compuers or social neworks. Daa, ypically, loosely srucured and ofen ungoverned, include hose saved in proper Social Neworks (such as Facebook, Twier, Tumblr ec.), in blogs and commens, in specialized websies for picures (Insagram, Flickr, Picasa ec.) 94 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

97 Big daa ypes for macroeconomic nowcasing 4 or videos (Youube, ec.) or inerne searches (Google, Bing, ec.), bu also ex messages, usergeneraed maps, s, ec. Ye anoher source of big daa, and perhaps he fases expanding one, is he so-called Inerne of Things. Machine-generaed daa are derived from sensors and machines used o measure and record he evens and siuaions in he physical world. I is becoming an increasingly imporan componen of he informaion sored and processed by many businesses. Is well-srucured naure is suiable for compuer processing, bu is size and speed is beyond radiional approaches. Examples include daa from sensors, such as fixed sensors (home auomaion, weaher/polluion sensors, raffic sensors/webcam, ec.) or mobile sensors (mobile phones, conneced cars, saellie images, ec.) bu also daa from compuer sysems (logs, web logs, ec.). The resuling many ypes of big daa have been already exploied in many scienific fields such as climaology, oceanography, biology, medicine, and applied physics. Specific areas of economics have also seen a major ineres in big daa and business analyics, in paricular markeing and finance. Insead, in convenional macroeconomics here have so far been limied applicaions, mosly concenraed in he areas of nowcasing/forecasing, on which we also focus. A firs aim of his paper is o review he main exising macroeconomic applicaions and, in paricular, o provide a discussion of various big daa ypes which could be useful in macroeconomic nowcasing. A survey of he economeric mehods for big daa is insead lef for fuure research. We have idenified en caegories of big daa which are organised as: (i) financial markes daa, (ii) elecronic paymens daa, (iii) mobile phones daa, (iv) sensor daa, (v) saellie images daa, (vi) scanner prices daa, (vii) online prices daa, (viii) online search daa, (ix) exual daa, and (x) social media daa. For each of hese caegories, in Secion 2, we analyse he specific characerisics of he daa, review he relevan macroeconomic papers based on hem, and provide deails regarding daa sources and availabiliy. Nex, in Secion 3, we consider he dual problem: nowcasing a specific macroeconomic variable of ineres, such as gross domesic produc (GDP) growh, inflaion or unemploymen, using big daa. A summary able, repored a he end of he Secion, liss he main sudies, providing deails on he specific ype of big daa adoped. In Secion 4, we propose a furher classificaion, specifically designed for numerical big daa and based on heir relaive cross-secional and emporal dimensions, which has implicaions for he proper daa pre-reamen and required economeric echniques o analyse hem. As menioned above, big daa are now spli ino hree main caegories: Fa, Tall, and Huge. The final goal of he paper is o use he exensive review of exising sudies and daa o provide a descripion of he main feaures of big daa ha make hem useful for nowcasing and flash esimaion, and an indicaion of which big daa caegories seem paricularly promising for nowcasing specific variables and for furher analysis. We deal wih hese issues in Secion 5. Secion 6 summarizes he main findings and proposes a summary classificaion of big daa, based on a se of key feaures ha emerge from he deailed review: source, provider, availabiliy, coninuiy, ype, size and sample, mea daa, feaure, frequency, pre-reamen, link wih arge, previous use, and required economerics. This classificaion is finally used o creae a grid for assessing he need and poenial usefulness of big daa for macroeconomic nowcasing and forecasing, in general and in official insiuions such as Eurosa, according o he oucome of a careful cos-benefi analysis. EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 95

98 4 Big daa ypes for macroeconomic nowcasing 2. Types of big daa for macroeconomic nowcasing This secion is divided ino subsecions, each corresponding o a big daa caegory which could be useful for macroeconomic nowcasing. To faciliae he reading process, and in order o provide a ypology of big daa, we have sandardised he forma of each subsecion as follows. A firs we describe he big daa and discuss is characerisics. Then, we review he relevan lieraure which currenly uses he specific ype of big daa in nowcasing/forecasing or oher applicaions in economics. Finally, we presen a deailed lis of daa sources along wih heir availabiliy Financial markes daa Advances in compuer echnology and daa sorage has allowed for he collecion and analysis of high-frequency financial daa. The mos widely observed forms of financial big daa are he rades and quoes. New York Sock Exchange (NYSE) iniiaed he collecion of his daa in This inraday daa poenially provides deailed informaion which could be used in he analysis of markes efficiency, volailiy, liquidiy as well as price discovery and expecaions. Cenral banks monior aciviy across all financial markes and nowadays high-frequency financial daa includes: Equiies rades and quoes for all ypes of invesors; Fixed-Income rades and quoes for all ypes of invesors; Foreign Exchange rades and quoes for all ypes of invesors; NXL and OTC Derivaives and opion ransacions; and generally all operaions in financial markes. The easies form of he above daa o obain is inraday rades and quoes. However, his daa is available for each securiy individually; herefore if we are ineresed in he consrucion of ime series index based on inraday daa, we would have o consider a large number of securiies. Consequenly, his increases he oal cos for his daa. Furhermore, daa which refers o he monioring of all marke operaions is very sensiive and only cenral banks and oher regulaory bodies have access o, herefore mos sudies in he lieraure rely on anonymised daa obained by various hird-pary providers. Alhough financial big daa is very imporan in he analysis of marke microsrucure, is use in macroeconomic nowcasing and forecasing is mainly on daily or weekly frequency. We consider financial daa as a major daa source, herefore i forms he basis of our explanaory daa used in he empirical applicaion in fuure. Our main aim is o examine wheher one of he oher nine ypes of big daa can improve he macroeconomic nowcasing and resul in improved esimaes APPLICATIONS Financial daa in high-frequency form has been he main elemen in volailiy and marke microsrucure sudies. Moreover, he use of such daa in macroeconomic forecasing, and subsequenly in nowcasing, has been included in many sudies, eiher afer aggregaing he daa o he same frequency as he macroeconomic arges or, more recenly, using mixed 96 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

99 Big daa ypes for macroeconomic nowcasing 4 frequency models. Given ha our focus in his projec is in alernaive big daa ypes, we refer he reader o Sock and Wason (2002a), Sock and Wason (2002b), Giannone, Reichlin and Small (2008), Angelini, Camba-Mendez, Giannone and Reichlin (2011), Banbura, Giannone and Reichlin (2011), Banbura and Runsler (2011), Modugno (2013), Andreou, Ghysels and Kourellos (2015) INDICATIVE DATA SOURCES & EXAMPLES Alhough mos of he hisorical high frequency daa is proprieary, end of day ransacion prices (along wih he daily open, high and low) are publicly available for mos equiies in major sock exchanges. Below, we provide an indicaive lis of daa sources as well as live feeds which could be used o collec inraday daa: Daily equiy prices: Google Finance, Yahoo Finance, S. Louis FRED for major US indexes, ec. are some free online sources; Hisorical Inraday daa: Kibo, PiTrading and Inradaa provide proprieary hisorical inraday daa for asses in major markes. Unsrucured inraday daa could be convered o srucured ime series forma by consrucing daily realised volailiy esimaors which could be included in he nowcasing exercise as risk indicaors; Live feeds: Live-Raes.com, knoema and Oanda provide live feeds for mos of he financial insrumens for free. These feeds could be used as a source for a daa collecion which could hen be used in an aggregaed way in nowcasing CONCLUSIONS Financial ransacions daa is a very imporan source of informaion for he analysis of markes as well as he forecasing of he economy. Given heir high-frequency naure, financial markes illusrae a very fas discoun of news, much faser han lower frequency macroeconomic indicaors. Therefore, financial markes daa is an essenial source of informaion in macroeconomic nowcasing. We will invesigae wheher such daa can be combined wih anoher source of big daa o provide more accurae nowcass and shor-erm forecass. Overall, if he researcher is ineresed in risk senimen indexes or he consrucion of policy uncerainy indexes, realised volailiy esimaors migh prove a more accurae and imely insrumen compared o he sandard measures, say VIX in he US Elecronic paymens daa The erm elecronic paymens is broad and considers all kinds of elecronic funds ransfer. In paricular, forms of elecronic paymens include: (i) cardholder-iniiaed ransacions, (ii) direc deposi paymens iniiaed by he payer, (iii) direc debi paymens iniiaed by businesses which debi he consumer s accoun for he purchase of goods or services, (iv) credi ransfers, (v) elecronic bill paymens in online sores, among ohers. The mos heavily used form of elecronic paymens is he cardholder-iniiaed ransacions, i.e. credi and debi card paymens. According o he Capgemini and BNP Paribas (2016) repor, cards dominae he global non-cash marke, accouning for 65 % of all non-cash ransacions. In Table 1 we repor he credi card usage in 2014 across six regions. Credi ransfers follow wih 17 % share and direc debis wih 12 %. Finally, checks accoun for 6 % globally. China, Hong Kong, India and oher Asian markes rank firs in he cards usage wih 84 % share in he non-cash marke. Second follow he CEMEA markes wih cards usage a 77 % of he non-cash ransacions. Japan, Ausralia, Singapore and EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 97

100 4 Big daa ypes for macroeconomic nowcasing Souh Korea markes follow hird wih 75 % cards share and Norh America ranks fourh wih 71 % share. Lain America and Europe are las wih 49 % and 47 % share of cards in he non-cash marke. Even in he markes wih he lowes usage, cards share is a leas double of he oher non-cash alernaives. For example, direc debis and credi ransfers accoun for 23 % and 26 % in Europe compared o 47 % of cards. The above saisics highligh ha cards is he main form of non-cash paymens. Card paymens include online as well as offline POS ( 6 ) purchases making hem very useful in he racking of consumer behaviour and reail sales (among ohers). Table 1: Non-cash paymens mix (%) Europe Norh America JASS CHI Lain America CEMEA Global Cards Credi Transfers Direc Debis Checks Source: Capgemini and BNP Paribas (2016). JASS: Japan, Ausralia, Singapore and Souh Korea. CHI: China, Hong Kong, India and oher Asian markes. CEMEA: Poland, Russia, Saudi Arabia, Souh Africa, Turkey, Ukraine, Hungary, Czech Republic, Romania and oher Cenral European and Middle Easern markes. The cards paymens are considered as a caegory of big daa because of high frequency of ransacions. We have housands of ransacions hroughou he day and, wih he huge increase of e-commerce, also during he nigh. One specific characerisic of cards daa is he weekly paern in daily aggregaed daa (or inraday paern in non-aggregaed daa). As indicaed by he lieraure, consumers end o purchase more goods and services owards he end of he week. Cards daa usually is offered aggregaed in order o ensure proecion of personal deails APPLICATIONS The lieraure in economics which uses credi cards daa sared very recenly. Galbraih and Tkacz (2007) is one of he firs papers which published he resuls of cards daa in macroeconomics. In paricular, hey use Canadian debi card ransacions ( 7 ) in order o provide real-ime esimaes of economic aciviy. Their predicive regression analysis provides informaion for consumer behaviour as well as improved nowcas esimaes. A firs hey find ha household ransacions have a weekly paern (on average), peaking every Friday and falling every Sunday. The highfrequency analysis of elecronic ransacions around exreme evens explains expendiure paerns around he Sepember 11 erroris aacks and he Augus 2003 elecrical blackou. Finally, consensus forecas errors for GDP and consumpion (especially non-durable) growh can be parly explained by cards daa. Carlsen and Sorgaard (2010) use Dankor paymens in order o nowcas he reail sales index in Denmark. Dankor is a debi card developed joinly by he Danish banks and inroduced in The Dankor is free of charge o he cusomers, and he card is exensively used by households. This fac makes Dankor an ideal insrumen for racking household aciviy and hus, reail sales. Anoher advanage of using Dankor is he iming of publicaion. Dankor daa is available one week afer he reference monh, whereas he reail sales index is published hree weeks laer. As menioned in he previous sudies, seasonal effecs are also presen which need exra care in order o end up wih a clean daase. The ou-of-sample nowcas exercise is in favour ( 6 ) Poin of Sale (POS). For example, card paymen a a reail sore. ( 7 ) Obained via he Canadian inerbank nework, Inerac. 98 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

101 Big daa ypes for macroeconomic nowcasing 4 of he wo models which use cards daa, however he evaluaion period is again oo narrow: monhly nowcass beween January, 2007 and May, Galbraih and Tkacz (2015) ackle direcly wih he issue of nowcasing Canadian GDP growh using Canadian credi and debi cards ransacions as well as checks. They find ha, among he paymens daa, debi card ransacions seem o produce he mos improved esimaes. The issue of seasonaliy is also presen here. The auhors sugges he use of X-11 mehodology ( 8 ) in order o clean he daa. Their main finding is ha nowcasing using high frequency elecronic paymens improve by 65 % beween he firs and final esimaes presening supporing evidence in he use of elecronic paymens daa. Duare, Rodrigues and Rua (2016) use ATM and POS high frequency daa for nowcasing and forecasing quarerly privae consumpion for Porugal. Their ATM daa is provided by Mulibanco, which is he Poruguese ATM and POS nework. Their mehodology is based on Mixed Daa Sampling (MIDAS) models and builds on he earlier work by Eseves (2009) confirming ha he use of elecronic paymens daa improves nowcasing and forecasing accuracy. Weekly paymen daa produce paricularly good resuls, while daily daa are oo noisy. Barne, Chauve, Leiva-Leon and Su (2016) derive an indicaor-opimized augmened aggregaor funcion over moneary and credi card services using credi card ransacion volumes. This new indicaor, insered in a mulivariae sae space model, produces more accurae nowcasing of he GDP compared o a benchmark model. Finally, Aprigliano, Ardizzi and Monefore (2016) use a mixed frequency dynamic facor model o predic he Ialian GDP growh using sandard business cycle indicaors (such as elecriciy consumpion, indusrial producion, inflaion, sock marke indexes, manufacuring indexes, ec.) as well as paymen sysems daa (cheques, credi ransfers, direc debis, paymen cards). They find ha monhly paymen daa helps in racking he economic cycle in Ialy and improves nowcasing. In a separae screening exercise using he Leas Absolue Shrinkage and Selecion Operaor (LASSO), paymen sysem variables are indicaed as poenial predicors of GDP growh INDICATIVE DATA SOURCES & EXAMPLES Based on he lieraure review and an exensive online search, we discuss poenial daa sources as well as availabiliy for cards daa. A firs we have all credi and debi card financial services corporaions. These companies faciliae elecronic funds ransfers hroughou he world. We briefly menion hem here. -- Visa, Inc. I is no known if VISA provides aggregaed daa o hird-paries, -- American Express. American Express is known o sell anonymised daa o hird-pary markeing companies ( 9 ). American Express also offers anonymised daa o heir business parners as a way o help hem analyse and promoe heir producs ( 10 ), -- Masercard. Masercard, along wih American Express, is known o sell anonymised daa o hird-pary markeing companies ( 11 ); ( 8 ) See Kapeanios, Marcellino and Papailias (2016) for a full discussion. ( 9 ) See online news aricles a hps://goo.gl/co3oc6 and hps://goo.gl/yrj74b. ( 10 ) See online news aricle a hps://goo.gl/ms9j0h. ( 11 ) See online news aricle a hps://goo.gl/co3oc6. EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 99

102 4 Big daa ypes for macroeconomic nowcasing Anoher way o acquire elecronic paymens daa is hrough inerbank neworks. Below we summarise he inerbank neworks for major economies. -- Ausralia: Elecronic Funds Transfer a Poin Of Sale (EFTPOS). Daa is no publicly available. In 2010, ANZ BANK has announced he launch of an online ool ha uses aggregaed daa from merchan EFTPOS ransacions o illusrae esimaed sales paerns, marke share, urnover and o provide insighs ino cusomer behaviour. However, his is available o ANZ business cusomers only, -- Canada: Inerac Associaion (as menioned in he papers above). No publicly available. However, given he exisence of curren lieraure, daa could be poenially purchased, -- China: China Union Pay. No publicly available and unknown if China Union Pay have an ineres in selling anonymised daa, -- France: Groupemen des Cares Bancaires CB. Paymen sysem: EBA Clearing (Euro1) and Trans-European Auomaed Real-ime Gross Selemen Express Transfer Sysem (TARGET2). No publicly available. However, given he exisence of curren lieraure, daa could be poenially purchased or come o an agreemen, -- Germany: Girocard. Euro1 and TARGET2. No publicly available. However, given he exisence of curren lieraure, daa could be poenially purchased or come o an agreemen, -- Ialy: BI-COMP. Euro1 and TARGET2. No publicly available. However, given he exisence of curren lieraure, daa could be poenially purchased or come o an agreemen, -- Japan: Yucho. No publicly available. I is no known if Yucho would be ineresed in selling anonymised daa, -- Porugal: Mulibanco (as menioned in he papers above). No publicly available. However, given he exisence of curren lieraure, daa could be poenially purchased or come o an agreemen; Third-pary daa providers, such as Envesne Yodlee ( 12 ), could also provide aggregaed daa on elecronic ransacions. In mos of hese cases, he daa is proprieary; Finally, he las way o collec cards daa would be via access o cenral bank daabases which are no publicly available. All cenral banks could provide such daa in aggregaed or unsrucured forma. Some of he cenral banks which could provide useful daa regarding his projec are: he European Cenral Bank (ECB), Banque de France, Deusche Bundesbank, Banca d Ialia, De Nederlandsche Bank, Banco de Porugal and Banco de Espana. However, Cenral Banks do no usually provide any daa o hird-paries. Some daa which is publicly available and migh be useful for demonsraion purposes, bu no for an empirical analysis due o he shor sample, include: (1) UK Corporae credi card ransacions in monhly frequency. This is a srucured series based on all ransacions on corporae credi cards in he financial year ; (2) Sample daase of European credi card ransacions during wo days in Sepember 2013 of Dal Pozzolo, Caelen, Johnson, and Bonempi (2015). The daa is publicly available on Kaggle websie ( 13 ). Due o confidenialiy issues, no original feaures are presened apar from he Time of he purchase, he Amoun and a flag which indicaes if he ransacion is genuine ( 12 ) hps://goo.gl/egaiou. ( 13 ) hps://goo.gl/g1sasd. 100 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

103 Big daa ypes for macroeconomic nowcasing 4 or fraudulen. In oal, here are 284,808 ransacions. This daase migh be useful in order o illusrae he daily aggregaion process, however i is no sufficien o show he weekly seasonal paern and is subsequen removal and daa cleaning CONCLUSIONS Elecronic paymens include credi and debi card ransacions, credi ransfers, direc debis, cheques, ec. Based on he naure of he daa, i racks well economic aciviy and paricularly household purchases (via credi and debi card ransacions). For his reason i would be very useful in he monioring, nowcasing and forecasing of reail sales, privae consumpion, and oher relaed variables. The lieraure provides empirical evidence in favour of cards daa. However, unsrucured or aggregaed daa is no publicly available and mos of hese sudies are provided wih confidenial daa from inerbank neworks or cenral banks Mobile phones daa Wih he inroducion of mobile phones nearly hiry years ago, scieniss across various fields were posiive ha mobile phones usage and informaion would be an imporan ool for saisics. The daa collecion from basic funcions of a mobile phone, i.e. receiving and making phone calls and shor ex messages, already provides enough informaion abou populaion densiy, locaion, economic developmen of paricular geographic areas and use of public ranspor among ohers. The rapid developmen and growh of mobile phones echnology during he pas weny years allows for even more specific daa collecion as inerne aciviy, mobile banking, GPS racking and oher sensors daa ( 14 ). Overall, mobile phones daa provides deailed informaion of human behaviour and herefore can be useful in social sciences oo. The Deloie (2012) repor which uses daa from he Cisco s VNI Index for 14 counries saes ha a doubling of mobile daa use leads o a 0.5 % increase in he GDP per capia growh rae. Given he heavy use of mobile phones, he colleced daa is characerised as big daa due o he massive volume. News coverage also provides evidence ha mobile daa is promising for he fuure ( 15 ) APPLICATIONS As in he case of card paymens daa, he lieraure which uses mobile phones daa is also very recen. Smih-Clarke, Mashhadi and Capra (2014) employ call daa in order o examine povery in wo developing counries where survey daa is scarce. In paricular, hey use he aggregaed call deail records of mobile phone subscribers and exrac feaures ha are srongly correlaed wih povery indexes derived from census daa. Their daa consiss of calls beween 5 million Orange cusomers from Coe d Ivoire and 928,000 calls beween cusomers of an undisclosed nework in an undisclosed region. The auhors highligh he difficuly of obaining call deails records for oher developed and emerging counries. Deville, Linard, Marin, Gilber, Sevens, Gaughan, Blondel and Taem (2014) use mobile phone daa for populaion mapping. They use more han 1 billion mobile phone calls from Porugal and France and show how spaially and emporarily explici esimaions of populaion densiies can be produced a naional scales. In he same opic also lies he work of Ricciao, Widhalm, Craglia and Panisano (2015) who esimae populaion densiy disribuion from nework-based ( 14 ) In ha sense, mobile phones daa can also be par of sensors daa. ( 15 ) See hps://goo.gl/myb21q, hps://goo.gl/e35iqq and hps://goo.gl/dkdara. EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 101

104 4 Big daa ypes for macroeconomic nowcasing mobile phone daa in Europe. In he same conex, De Meersman, Seynaeve, Debusschere, Lusyne, Dewie, Baeyens, Wirhmann, Demuner, Reis and Reuer (2016) assess he qualiy of mobile phones daa o esimae he acual populaion. They use Belgian mobile phone daa from he major nework operaor, Proximus, and heir resuls are consisen wih he 2011 Census. Mao, Shuai, Ahn and Bollen (2015) also invesigae he case of Coe d Ivoire o analyse he relaions beween is naion-wide communicaions nework and he socio-economic dynamics of is regional economies. They inroduce an indicaor o quanify he relaive imporance of an area on he basis of call records, and show ha a region s raio of in- and ou-going calls can predic is income level. Their resuls demonsrae he poenial of mobile communicaion daa o monior he economic developmen and social dynamics of low-income developing counries in he absence of exensive economeric and social daa INDICATIVE DATA SOURCES & EXAMPLES Below we briefly menion some daa sources as well as examples for demonsraion. Obviously, he nework operaors are he main mobile phone daa source as all informaion is sen and received hrough heir nework. However, he daa is no publicly available and, as menioned in he lieraure, his is due o regulaory limiaions. I is no known if a privae nework operaor company would be ineresed in selling anonymised daa. Third-pary sofware developers, such as HelloSpy ( 16 ), which offer heir cusomers he abiliy o rack heir call hisory. This is proprieary daa CONCLUSIONS Mobile phone daa also seems o be an ineresing source. I could be used in order o provide imely esimaes for populaion and census daa, moreover, as we see in Toole e al (2015), he analysis of phone calls can provide deails abou behaviour which could be used for unemploymen and economic aciviy forecasing. Bu, generally speaking, i seems difficul o find a deailed and long enough daase o be esed in a nowcasing/forecasing applicaion Sensor daa and he inerne of hings Sensor daa mainly refers o any kind of oupu of a device which deecs and responds o inpu sources from he physical environmen. One of he oldes examples is he emperaure monioring via climae sensors. Sensors have been long used in manufacuring of plans, cars, ships, miliary equipmen and sensor daa has been used by operaional echnology engineers for quie many years. Informaion echnology and he rapid developmen of inerne grealy faciliae he collecion and disribuion of sensor daa and gives rise o he so-called Inerne of Things (IoT). A hing includes any iem which can be aached wih sensors. Inerne access allows he hing o auomaically ransmi he daa via neworks and sore i o public clouds and daabases. This, in-urn, provides easy access o sensor daa for mining and analyics purposes. Mobile hings, e.g. smar phones, compuers, ables, waches, home appliances, cars, drones, ec., as well as saionary hings, e.g. wind urbines, emperaure sensors, ec., all come aached wih sensors which can send and receive informaion online. In he his caegory, we include ( 16 ) hps://goo.gl/igvljm. 102 EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors

105 Big daa ypes for macroeconomic nowcasing 4 daases which are colleced from sensors and oher devices which do no rack direcly human aciviy as opposed o mobile phones daa where all informaion is reurned by human acions. According o Ebner (2015) ( 17 ), Garner, a leading informaion echnology research and advisory company, forecass 25 billion conneced hings by The daa generaed by hese sensors could impac almos every major indusry, including healhcare, ransporaion, and energy. Cisco esimaes ha IoT could have an economic impac of $14 rillion by he early 2020s. Apple and Samsung are already developing plaforms which gaher daa from he waches and oher insrumened objecs of he IoT. This daa is hen made available o developers for he creaion of new applicaions and analyses ( 18 ). A recenly sar-up company, ThingSpeak ( 19 ), provides an open-source plaform which allows he user o have his hings ransmi daa o heir servers and hen offers his daa via an API. Mahworks and MATLAB are inegraed in he plaform which can be used direcly for analysis and visualisaion of he daa. Figure 1: The inerne of hings Source: Graham (2016) APPLICATIONS Unforunaely, i seems ha here does no exis a lieraure on economic forecasing or nowcasing based on sensor daa. We believe ha his is due o he unavailabiliy of daa. Mos of he lieraure is based on weaher forecasing, where ample daa exiss, raffic and oher geo-relaed applicaions. Some indicaive sources include Jones (2010), Bazzani, Giovannini, Galoi and Rambaldi (2011), Chan, Khadem, Dillon, Palade, Singh and Cheng (2012), Mass (2012) and Fernandez-Ares, Mora, Arenas, de las Cuevas, Garcia-Sanchez, Romero, Rivas, Casillo and Merelo (2016). Also, Suryadevara, Mukhopadhyay, Wang and Rayudu (2013) forecas human ( 17 ) hps://goo.gl/9h3zf8. ( 18 ) See hps://goo.gl/9h3zf8. ( 19 ) hps://goo.gl/z4y9by. EURONA Eurosa Review on Naional Accouns and Macroeconomic Indicaors 103