PRODUCTIVITY TRENDS AND DETERMINANTS IN CANADA

Transcription

1 PRODUCTIVITY TRENDS AND DETERMINANTS IN CANADA 1 Paper presened a he CSLS-Indusry Canada Workshop on Produciviy Issues in a Canadian Conex, Friday Sepember 29, 2000 a he Lord Elgin Hoel, Oawa. By Erwin Diewer, Deparmen of Economics, Universiy of Briish Columbia, Vancouver, Canada, V6T 1Z1. diewer@econ.ubc.ca 1. Inroducion In his opening paper for he Workshop, I will cover only 3 opics: Wha is he evidence on economy wide produciviy growh for he U.S. and Canadian economies over he years ? Wha proporion of he growh in real oupu in Canada for he above years is due o Toal Facor Produciviy (TFP) growh and wha proporion is due o he growh of primary inpus and changes in he erms of rade? Wha are he facors ha explain TFP growh? The above 3 opics will be covered in secions 2,3 and 4 below. I should be menioned ha in secion 2, we will review boh he growh of labour produciviy (oupu per hour worked) in he U.S. and Canada and he growh of TFP. 1 Toal Facor Produciviy is he raio of an index of oupus produced by he economy divided by an index of inpus used by he economy. We regard TFP as being he more accurae produciviy measure because labour produciviy can increase if capial inpu increases dramaically bu a he same ime TFP can fall Trends in Canadian and U.S. Produciviy, In his secion, we will compare Canadian and U.S. labour produciviy and oal facor (or mulifacor) produciviy for he years For he U.S economy, he wo produciviy series are readily available from he Bureau of Labor Saisics websie. For he Canadian economy, Coulombe (2000) has shown ha he official Saisics Canada esimaes of oal facor produciviy are no comparable wih he corresponding U.S. esimaes for hree reasons: U.S. esimaes of labour inpu are based on a deailed demographic model of labour supply whereas Canadian esimaes of aggregae labour inpu are based on an aggregae of indusry inpus of labour; 1 However, we will no acually compare he level of oupu in Canada wih ha of he U.S. This is done in Lee and Tang (2000). 2 In pracice, labour produciviy and Toal Facor Produciviy usually move in he same direcion.

2 Saisics Canada esimaes of mulifacor produciviy do no include he conribuions of invenory and land as inpus in he producion process whereas he U.S. esimaes do include hese conribuions; Saisics Canada depreciaion raes for he componens of reproducible capial are considerably higher han corresponding U.S. depreciaion raes, leading o a slower growh of aggregae Canadian inpu and a faser growh of oal facor produciviy in Canada compared o he U.S. The hird facor is he mos imporan source of difference in mehodology beween he U.S. and Canadian saisical agencies. 3 I is no cerain who is correc (on he magniude of depreciaion raes in he U.S. and Canada) bu i seems likely ha he acual depreciaion raes are no ha differen. The difference in assumed depreciaion raes beween he U.S. and Canada is very large as Coulombe (2000; 11) noes: For he capial concep ha excludes land and invenories, he aggregae implici depreciaion rae in he U.S. averages 4.4 percen beween 1961 and This compares wih he depreciaion rae of 10 percen used o esimae he growh of Canada s business secor capial sock for MFP measuremens. This is a big difference, o say he leas. Such a difference in aggregae depreciaion raes migh be expeced o have a large impac on he growh of capial sock and imporan implicaions for he measuremen of MFP growh. In an aemp o make he Canadian esimaes of oal facor produciviy growh more comparable wih U.S. esimaes, we will assume ha invesmens in nonresidenial srucures in Canada depreciae a a declining balance (or geomeric) depreciaion rae of 3.5% and machinery and equipmen invesmens depreciae a a declining balance depreciaion rae of 12.5%. This will lead o an average depreciaion rae for reproducible capial in Canada ha is somewha higher han he corresponding U.S. raes bu he raes will be much more comparable. Including land and invenories as producive inpus insead of excluding hem will end o reduce he rae of growh of aggregae capial inpu and hus he Saisics Canada esimae of capial growh will end o be larger han he corresponding U.S. esimae and hence he Canadian esimaes of oal facor produciviy growh will end o be smaller han he corresponding U.S. esimaes due o he Canadian exclusion of hese producive inpus. Coulombe (2000; 9-10) esimaes he magniude of his exclusion as follows: 4 By comparison o he U.S. approach, Saisics Canada s mehodology impars an upward bias o he measuremen of capial sock growh and a downward bias o he calculaion of MFP growh. We esimae ha he effec of using a narrower definiion raher han a broader concep of capial sock is o reduce he MFP growh rae by one-enh of 1 percenage poin per year over he period. While his is a small number, MFP annual growh raes are also modes, ypically around 1 percen. Consequenly, he underesimaion amouns o approximaely 10 percen of oal annual MFP growh. 2 3 Coulombe (2000; 11) noes ha by applying BEA depreciaion procedures, he growh of Canada s capial sock since 1980 increases by abou one percen per year. Thus by applying U.S. depreciaion raes, official Canadian mulifacor produciviy growh is reduced by abou.3 o.35 percenage poins per year over he las 20 years or so. 4 Coulombe (2000; 22) noes ha: Diewer and Lawrence (1999), from a compleely differen mehodology and using Canadian daa only, arrive a exacly he same number. They esimae ha he exclusion of land and invenories as inpus decreases mulifacor produciviy growh in Canada by 0.1 percen per year.

3 Thus puing aside he difference in labour inpu measures beween he U.S. and Canada 5, Coulombe esimaes ha Canadian mulifacor produciviy esimaes are around.25 percenage poins per year higher han he corresponding U.S. esimaes over he years due o he differences in he definiion of capial inpu and he differences in he assumed depreciaion raes for he componens of reproducible capial in he wo counries. Coulombe buil up his esimaes of Canadian MFP using esimaes of indusry oupu. However, indusry esimaes of oupu and inermediae inpu are raher fragile in all counries due o he lack of surveys on inermediae inpu flows and in paricular, of service flows beween indusries. Hence, Diewer and Lawrence (1999) buil up esimaes of Canadian mulifacor produciviy growh using esimaes of final demand (adjused for commodiy axes), which hey hough were more reliable. In his secion, we will updae heir MFP esimaes from 1996 o One problem wih he Diewer and Lawrence esimaes is ha hey used Saisics Canada depreciaion raes for he componens of reproducible capial in Canada. As was menioned above, in he presen paper, we will use depreciaion raes ha are closer o he U.S. raes. 6 For a descripion of he daa sources and mehodology ha we are using, see Diewer and Lawrence (1999). For a lising of he major oupu and inpu series, see Appendix 1 below. 7 Table 1 below liss labour produciviy for Canada (LP CAN ) and for he U.S. (LP US ) for he years These series represen esimaes of privae secor gross domesic produc divided by a measure of privae business secor labour inpu. 8 Table 1 also liss esimaes of Toal Facor Produciviy for Canada (TFP CAN ) and for he U.S. (TFP US ) for he years These series represen esimaes of privae secor gross domesic produc divided by a measure of privae business secor labour inpu and capial inpu. The U.S. series are aken from he Bureau of Labor Saisics websie. 3 5 Gu and Ho (2000) consruc a Canadian labour inpu series ha is a counerpar o ha used by he BLS. 6 There are some oher differences in he daa used in his paper compared o Diewer and Lawrence (1999): (i) revised Saisics Canada daa were used; (ii) in he presen paper, daa on invesmen back o 1926 were aken from Leacy (1983) (see series F19, F20, F43 and F44) and used for he years ; (iii) in order o obain saring capial socks for nonresidenial srucures and for machinery and equipmen in 1926, i was assumed ha gross fixed capial formaion in hese componens was growing a a 2 % per year rae in he years prior o 1926 and we assumed ha he declining balance depreciaion rae for nonresidenial srucures was 3.5 % per year and for machinery and equipmen was 12.5% per year. These assumpions gave us saring capial socks ha were roughly equal o he saring socks lised in Leacy (1983) for The oupu series lised in his Appendix were buil up from 34 deailed oupu series on 20 consumpion componens, 1 governmen componen, 5 invesmen componens, 5 expor componens and 4 impor componens covering he years Chain Fisher ideal indexes were used o aggregae hese deailed series ino he usual naional accouns ype aggregaes (bu a producer prices raher han final demand prices). Saisics Canada daa were used hroughou he daa consrucion process. 8 The labour produciviy series have been normalized o equal uniy in The Toal Facor Produciviy series do no have o be normalized because he value of inpu is equal o he value of oupu in each period.

4 Table 1: Esimaes of Labour and Mulifacor Produciviy for he Years Year LP CAN LP US TFP CAN TFP US The above series are graphed in Figure 1 below. The op line is U.S. labour produciviy, he nex line below i is Canadian labour produciviy, he nex line is U.S. Toal Facor Produciviy and he boom line is Canadian Toal Facor Produciviy 9. I can be seen ha over he 37 year period, he U.S. produciviy performance has been beer han he Canadian one for boh ypes of produciviy. However, he more imporan TFP gap is no ha large: a 9 In he following secion we indicae more precisely how our esimae of Canadian TFP was consruced.

5 5 he end of he 37 year period, U.S. TFP growh only exceeds Canadian growh by abou 7.5%. U.S. labour produciviy growh exceeds Canadian labour produciviy growh by abou 19.5% over he 37 year period. I is useful o break down he produciviy growh performance of he wo counries ino various subperiods. Our firs subperiod covers he growh over he years 1963 o 1973 (11 years). These years are par of he golden years of produciviy growh in boh counries. The nex period covers he dismal years, 1974 o 1991 (18 years in all). These were he years of he wo energy shocks in 1974 and , high inflaion 10, and a world wide recession around Our final period covers he years (7 years in all) where inflaion subsided and here were no major recessions. Produciviy growh raes were obained from he daa in Table 1 above by aking each year s level and dividing by he level of he previous year. These annual produciviy growh raes were hen averaged over he periods described above. The resuls are repored in Table 2 below. Figure 1: Canadian and U.S. Produciviy Levels; Series1 Series2 Series3 Series Table 2: Average Canadian and U.S. Produciviy Growh Raes Time Period GLP CAN GLP US GTFP CAN GTFP US % 2.08% 0.92% 1.13% % 3.17% 1.83% 2.20% % 1.45% 0.32% 0.45% % 1.99% 1.03% 1.20% 10 Diewer and Fox (1999) argue ha high inflaion will end o reduce produciviy growh for a variey of reasons.

6 6 From Table 2, i can be seen ha over he enire ime period of 37 years, U.S. labour produciviy exceeded ha of Canada by.5% per year on average. For he more imporan Toal Facor Produciviy measure, U.S. TFP growh exceeded ha of Canada by abou 0.2 percenage poins per year. In absolue erms, his does no seem like a large produciviy gap, bu given ha he average TFP growh rae in boh counries is only abou 1 percen per year, his ranslaes ino a 20% relaive gap. I can be seen ha he golden years of produciviy growh were indeed very good in boh counries for he period prior o he firs oil shock near he end of 1973, averaging abou 2% per year in boh counries. However, during he high inflaion period, , his rapid rae of TFP growh fell dramaically in boh counries o.32% per year in Canada and.45% per year in he U.S. Finally, in he new economy 1990 s (he period ), TFP growh has picked up in boh counries, increasing o abou 1% per year in Canada and o 1.2% in he U.S. However, hese growh raes are sill below he TFP growh raes achieved in he pre 1973 period. 11 Noe ha for all ime periods, he U.S. appears o have had faser raes of produciviy growh han Canada. We urn now o an analysis of he relaive conribuion of TFP growh o he growh of real oupu in Canada. 3. The Sources of Real Oupu Growh in Canada Kohli (1990) developed a very illuminaing decomposiion of a counry s nominal GDP growh ino various explanaory facors such as he growh of he counry s domesic prices, of is expor and impor prices and is growh in primary inpus such as labour and capial. 12 We now explain how his mehodology works. Define q D as he quaniy of domesic final demand in period and le p D be he corresponding price. 13 Define q X and q M as he quaniy of expors and impors in period and le p X and p M be he corresponding prices. 14 Then nominal GDP in period is defined as: (1) v p D q D + p X q X p M q M. We lis he quaniies ha appear in equaion (1) in Table 3 below. The q variables are in billions of 1962 dollars bu v is in billions of curren dollars. 11 Griliches (1979) and Diewer and Fox (1998) argue ha curren real oupu is surely higher han is measured by saisical agencies due o he lack of qualiy adjusmen in he measuremen of services. Since he service secor has been growing seadily since he golden years of produciviy growh, i is likely ha curren TFP is higher han is currenly being measured. 12 Kohli s work draws on ha of Diewer and Morrison (1986). See also Fox and Kohli (1998) for a recen applicaion of his mehodology o Ausralia. 13 In our empirical work, q D was defined as a chain Fisher ideal aggregae of 20 separae consumpion series plus one governmen series and 4 invesmen series. See Diewer and Lawrence (1999) for a deailed descripion of hese series. 14 In our empirical work, q X is a Fisher ideal chain aggregae of 5 Canadian expor componens and q M is a fisher chain aggregae of 4 Canadian impor componens.

7 Table 3: Canadian Quaniy Componens of Nominal GDP, Year v q D q X q M Looking a he las 3 columns of Table 3, we see ha boh expors and impors have grown much more rapidly han domesic demand in real erms. However, he growh in impors is much faser han he growh of expors. This is due o increasing impors of high ech equipmen from he U.S. and oher areas, which are falling in price. From Table 4 below, i can be verified ha expor prices are increasing faser han impor prices; i.e., he erms of rade for Canada are improving over he period

8 Table 4: Canadian Price Componens of Nominal GDP, Year p D p X p M Now use he above daa o consruc an implici (chain) Törnqvis index of oupus, wih q D, q X and q M as he 3 quaniies o be aggregaed wih price weighs p D, p X and p M respecively. This aggregae oupu index is o be divided by a Törnqvis index of 5 inpus and his is he TFP index, a say, lised in column 4 of Table 1. The 5 inpus are: labour, nonresidenial srucure services, machinery and equipmen services, invenory services and business and agriculural land services. Denoe he price and quaniy of privae secor labour inpu in period by p L and q L respecively. Denoe he declining balance user coss of he 4 ypes of capial

9 9 inpu in period by u NS, u ME, u IS and u BAL respecively. 15 Denoe he quaniy used of each of hese ypes of capial in period by q NS, q ME, q IS and q BAL. These daa are lised in he Appendix. Kohli (1990) shows ha he nominal GDP in period, v, has he following decomposiion ino explanaory facors if cerain condiions on he counry s echnology hold: 16 (2) v = v 1 a b D b X b M c L c NR c ME c IS c BAL where v 1 is nominal GDP in a base period (period 1), a is he Törnqvis TFP index for period (see column 4 of Table 1), b D, b X and b M are he ranslog price effecs defined in Diewer and Morrison (1986; 666) and c L, c NR, c ME, c IS and c BAL are he ranslog quaniy effecs defined in Diewer and Morrison (1986; 667). Each price effec represens he effec on period nominal GDP due o he change in he price of domesic oupu going from period 1 o period (his is he b D price effec), he price of expors (his is he b X effec) or he price of impors (his is he b M effec). Each quaniy effec represens he effec on period nominal GDP due o he change in he quaniy of each primary inpu going from period 1 o period. The logarihmic change in he nh price effec going from period 1 o is defined empirically as follows: (3) ln (b n /b n 1 ) (1/2)[s n 1 + s n ] ln (p n /p n 1 ) ; n = D, X or M and he period expendiure shares for (ne) oupu n is defined as (4) s D p D q D / v ; s X p X q X / v and s M p M q M / v. The logarihmic change in he nh quaniy effec going from period 1 o is defined empirically as follows: (5) ln (c n /c n 1 ) (1/2)[σ n 1 + σ n ] ln (q n /q n 1 ) ; n = L, NR, ME, IS and BAL where he period expendiure shares for primary inpu n is defined as (6) σ L p L q L / v ; σ NR u NR q NR / v ; σ ME u ME q ME / v ; σ IS u IS q IS / v and σ BAL u BAL q BAL / v. Definiions (4) and (6) along wih he period 1 normalizaions for he b 1 n = 1 and he c 1 n = 1 serve o define he b n and he c n for all periods = 1,2,,37. Since we assume ha he quaniy of business and agriculural land is fixed, he quaniy effec c BAL is always equal o 1 and hence can be ignored in he decomposiion (2). The remaining price and quaniy effecs are lised in Table 5 below. 15 These user coss are explained in Diewer and Lawrence (1999) and in he Appendix. 16 Essenially, he echnology of he counry has o be represenable by a cerain ranslog profi funcion; see Diewer and Morrison (1986) or Kohli (1990) for he deails. The assumpions do no appear o be very resricive.

10 Table 5: GDP Price and Quaniy Effecs for Canada; Year b D b X b M c L c NR c ME c IS Looking a Table 5, i can be seen ha he smalles effecs on GDP growh are due o he accumulaion of invenories. The larges effecs on nominal GDP growh are due o he changes in domesic prices (i.e., due o inflaion). Comparing enries in Tables 4 and 5, i can be seen ha he domesic price effec series, b D, is virually idenical o he domesic inflaion price series, p D.

11 11 As we menioned above, because he quaniy of business and agriculural land was assumed o be consan in our sudy, he quaniy effec c BAL is idenically equal o uniy. Hence we can rewrie he decomposiion of nominal GDP given by (2) above as follows: (7) (v /v 1 )/b D = a b X b M c L c NR c ME c IS = a b T c L c NR c ME c IS As was menioned above, b D, is essenially equal o he price of domesic oupu, p D. Hence, he lef hand side of (7) is essenially real GDP (normalized o equal 1 in he base period). On he righ hand side, we have a series of facors ha conribue o real growh; namely: TFP growh a, b T b X b M which is he combined effec of changes in expor and impor prices or changes in he erms of rade, labour growh c L, he growh of nonresidenial srucures c NR, he growh in he use of machinery c ME, and he growh in invenory socks c IS. In Table 6 below, we sar wih TFP as a conribuor o real oupu growh and hen in he second column we able he combined effecs of TFP growh and changes in he erms of rade (his is a b T ). In he hird column, we add he effecs of labour inpu growh; in he fourh column, we add he effecs of growh in he sock of nonresidenial srucures; in he fifh column, we add he effecs of growh in machinery and equipmen socks and in he sixh column, we add in he effecs of invenory growh. The sevenh column is (v /v 1 )/b D, normalized deflaed GDP, which is indeed exacly equal o he sixh column. Figure 2 graphs hese columns.

12 Table 6: The Decomposiion of Real GDP ino Growh Facors 12 a *b T *c L *c NR *c ME *c IS (v /v 1 )/b D

13 13 1s Qr 2nd Qr 3rd Qr 4h Qr Eas W es Norh Figure 2: Sources of Real Oupu Growh in Canada Series1 Series2 Series3 Series4 Series5 Series The op line in Figure 2 represens he growh in real GDP in Canada for he years The boom line represens he conribuion of TFP growh. The nex line up represens he addiional conribuion of changes in he erms of rade. I can be seen ha his conribuion is much smaller han he effecs of produciviy growh. The nex line represens he addiional conribuion o real oupu growh of labour inpu growh. I can be seen ha his is he bigges conribuor o real oupu growh of all of he sources of growh. Nex comes he conribuion of increases in he sock of nonresidenial srucures. This conribuion is approximaely equal o he conribuion of increases in TFP. Nex comes he conribuion of increases in he sock of machinery and equipmen. This conribuion is also approximaely equal o he conribuion of TFP growh. The final line adds he conribuion of growh in invenories; his conribuion is raher small. Figure 2 shows a a glance ha he main drivers of real oupu growh in Canada over he pas 37 years are growh in labour and capial inpu. Unforunaely, growh in TFP has no been a very large conribuor o overall Canadian oupu growh. The above analysis does no ell us wha he deerminans of Canadian TFP growh were; i jus ells us ha over he pas 25 years or so, TFP growh does no appear o have been very subsanial. In he following secion, we review a paper by Harris (1999) which aemps o map ou wha facors influence TFP growh.

14 4. The Deerminans of Canadian Produciviy Growh 14 Richard Harris (1999; 13-15) idenified 3 main drivers of produciviy growh: Invesmens in machinery and equipmen. Invesmens in educaion, raining and human capial developmen. Openness of he economy o inernaional rade and direc foreign invesmen. The above 3 drivers seem very reasonable. New knowledge is ofen embodied in new machines so old asks can be performed more efficienly. Educaing workers enables hem o accomplish a wide variey of asks more efficienly. An economy wih high ariffs and impor quoas will ofen have many oher disorions ha preven prices from allocaing resources efficienly. In heory, hese efficiency losses induced by ariffs and axes only affec he level of oupu and consumpion and no produciviy growh per se, bu in pracice, a highly disored economy will usually no be an aracive one for underaking research and developmen or for invesing in new plan and equipmen and hence produciviy growh can suffer. Harris (1999; 15-16) also looks a broader facors ha migh influence produciviy growh. Some of he facors ha he liss ha I find very plausible are: Innovaion; i.e., he developmen of new producs and or processes somewhere in he world for he firs ime. Diffusion of innovaions; i.e., he adopion of a new produc or process in he local economy. Economies of scale. Many physical processes are more efficien when hey are operaed a a larger scale. Pu anoher way, commodiies are lumpy or a leas hey are sold in discree lumps. I simply canno buy very iny amouns of mos commodiies. Pu ye anoher way, he economy is filled wih fixed coss. There are fixed coss of developing a new produc, here are fixed coss in selling he commodiy, here are fixed coss in ransporing commodiies, ec. As he scale of he marke becomes larger, hese fixed coss diminish as a proporion of he selling price and economic efficiency improves. 17 Spaial agglomeraion or he growh of ciies. Large ciies allow specialized markes o develop boh on he produc side and on he skill side. In oher words, in rural communiies, he number of goods and services ha can be purchased on local markes is limied 18 and producers may no be able o find specialized workers ha hey require. This poin is relaed o he previous poin: as ciies grow, markes become larger and more specializaion of labour is possible Alfred Marshall (1898; chaper 11) is quie good on his poin: Again, i is rue ha when a hundred ses of furniure or clohing, have o be cu ou on exacly he same paern, i is worh while o spend grea care on so planning he cuing ou of he boards or he cloh, ha only a few small pieces are wased. Marshall (1898; 358). 18 Of course, his siuaion is rapidly changing as far as goods are concerned due o he provision of goods and some services over he inerne. 19 Marshall (1898; 396) described his famous exernal economies of scale as follows: Meanwhile an increase in he aggregae scale of producion of course increases hose economies, which do no direcly depend on he size of individual houses of business. The mos imporan of hese resul from he growh of correlaed branches of indusry which muually assis one anoher, perhaps being concenraed in he same localiies, bu anyhow availing hemselves of he modern faciliies for communicaion offered by seam ranspor, by he elegraph and by he prining press.

15 The provision of public infrasrucure for ransporaion, communicaion and wase removal. This facor is very imporan when i is absen! Managemen pracices. This explanaory facor could perhaps be subsumed under he diffusion of echnology heading bu I hink Harris is righ o give i a separae billing. In paricular, he conribuion of business consulans who bring informaion on global bes pracices o he local economy offer a relaively inexpensive way of increasing produciviy dramaically. 20 High axes (negaive). Unless he revenues raised by high axes are spen incredibly well, here will be deadweigh losses and marginal excess burdens associaed wih high ax regimes. Again, his would seem o be a level effec and no necessarily affec TFP growh. However, in a world where some governmens offer lower ax raes han oher jurisdicions, economic aciviy and foreign invesmen will be araced o he low ax locaions and his will in urn simulae TFP growh given he link beween invesmen and TFP growh. Conversely, fooloose invesmens will avoid high ax jurisdicions and TFP growh will suffer. 21 Small firms (negaive). Small firms canno afford large invesmens in research and developmen, hey may no be able o specialize adequaely and hey may have large fixed coss. In general, very small firms will no be as efficien as large firms. In spie of his, governmens end o favour small firms and penalize large firms in all sors of ways. 22 Labour marke flexibiliy (posiive). This poin fis in wih he second main driver of produciviy growh ha Harris idenified earlier. Recen reforms o he Canadian sysem of unemploymen insurance 23 very modesly penalized repea users of wha is now called employmen insurance. These reforms were necessary in order o move he old unemploymen insurance sysem away from a very hefy subsidy for seasonal workers and owards a sysem ha would provide emporary relief for a worker who (permanenly) los his or her job. However, i is proving o be difficul for governmens o live wih he new regime even hough i improves labour marke flexibiliy. Low inflaion (posiive). I seems difficul o make he case ha his facor would grealy influence produciviy growh. Bu when one looks a he recen economic hisory of OECD counries, one is sruck by he empirical fac ha virually every counry experienced a dramaic drop in TFP growh during he years and a he same ime, a big increase in inflaion occurred. Diewer and Fox (1999) idenify a couple of mechanisms whereby higher inflaion migh ranslae ino lower raes of TFP growh: (i) business income ax sysems were no generally indexed for he effecs of inflaion and so businesses ha used capial inpus wih low depreciaion raes were unfairly penalized and (ii) muliproduc businesses probably did no price heir producs correcly in periods of high inflaion. The debae on his opic is sill open bu we do seem o be seeing a resurgence of TFP growh in recen years as inflaion remains low in mos OECD counries Harris (1999; 19) laer makes he following poin: There is a growing body of evidence ha he growh process is fundamenally driven by he relocaion of resources from low-produciviy growh aciviies o high-produciviy growh aciviies, raher han by limis on he availabiliy of new echnology. I oally agree wih his poin. For some evidence on he vas differences in produciviy ha can resul from firms using essenially he same echnology, see Diewer and Nakamura (1999). 21 Many Briish Columbia privae secor economiss conras he high ax policies of B.C. wih he lower ax policies in Albera and aribue he relaive increase in invesmen in Albera o his facor. The Irish economy is anoher economy ha has experienced a boom due in par o is low raes of business axaion. 22 In Canada, small firms pay a lower rae of business income ax and hey are no subjec o many raher onerous programs ha governmens reserve for large firms. 23 See Nakamura and Diewer (2000).

16 The above discussion will probably suffice o give he flavour of he work of Richard Harris on produciviy. As can be seen from he above discussion of his poins, I prey much agree wih him. Basically, we have some prey good ideas of wha facors will influence produciviy growh bu firm evidence on mos facors is lacking. 16

17 References 17 Coulombe, S. (2000), The Canada-U.S Produciviy Growh Paradox, Indusry Canada Working Paper Number 32, Oawa, March/ Diewer, W.E. and K.J. Fox (1998), "The Produciviy Paradox and he Mismeasuremen of Economic Aciviy", Insiue for Moneary and Economic Sudies, Discussion Paper 98-E-15, Bank of Japan, C.P.O. Box 203, Tokyo, , Japan, 26 pp., December. Diewer, W.E. and K.J. Fox (1999), Can Measuremen Error Explain he Produciviy Paradox?, Canadian Journal of Economics 32:2, (April), Diewer, W.E. and C.J. Morrison (1986), adjusing Oupu and Produciviy Indexes for Changes in he Terms of Trade, Economic Journal 96, (Sepember), Diewer, W.E. and A.O. Nakamura (1999), Benchmarking and he Measuremen of Bes Pracice Efficiency: an Elecriciy Example, Canadian Journal of Economics 32:2, (April), Fox, K.J. and U. Kohli (1998), GDP Growh, Terms-of-Trade Effecs and Toal Facor Produciviy, The Journal of Inernaional Trade and Economic Developmen7:1, Gu, W. and M.S. Ho (2000), A Comparison of Indusrial Produciviy Growh in Canada and he Unied Saes, Indusry Canada, forhcoming. Harris, R.C. (1999), Deerminans of Canadian Produciviy Growh: Issues and Prospecs, Indusry Canada Discussion Paper Number 8, Oawa, December. Kohli, U. (1990), Growh Accouning in he Open Economy: Parameric and Nonparameric Esimaes, Journal of Economic and social Measuremen 16, Leacy, F.H. (1983), Hisorical Saisics of Canada, Second Ediion, Oawa: Saisics Canada. Lee, F.C. and J. Tang (2000), Produciviy Levels and Inernaional Compeiveness in Canada and he Unied Saes, Indusry Canada, forhcoming. Marshall, A. (1898), Principle of Economics, Fourh Ediion, London: Macmillan. Nakamura, A.O. and W.E. Diewer (2000), Insurance for he Unemployed: Canadian Reforms and heir Relevance for he Unied Saes, pp in Long-Term Unemploymen and Reemploymen Policies, L.J. Bassi and S. A. Woodbury (eds.), Samford, Connecicu: Jai Press.