Productivity Levels and International Competitiveness 5 Between Canada and the United States

Transcription

1 Productvty Levels and Internatonal Compettveness 5 Between Canada and the Unted States Frank C. Lee and Janmn Tang 5.1 Introducton T HE PURPOSE OF THIS PAPER s to compare total factor productvty (TFP) levels and nternatonal compettveness between 33 Canadan and U.S. ndustres. To carry out such comparsons, we frst need to construct purchasng power partes (PPPs) for output and nputs by ndustry. We use blateral Canada-U.S. commodty prce data to construct PPPs for output and ntermedate nputs, and estmate PPPs for captal nput based on the relatve prces of nvestment goods, takng nto account the flow of captal servces per unt of captal stock. We then use hourly labour compensaton rates, dsaggregated by dfferent worker types n the two countres, to estmate labour nput PPPs. These PPPs take nto account dfferences n the composton of the output and nputs of the ndustry under consderaton between Canada and the Unted States, thereby allowng nter-country comparsons of both prces and quanttes of output and nputs. Followng Jorgenson and Nshmzu (1978) for comparson between Japan and the Unted States, we use a translog producton functon orgnally ntroduced by Chrstensen, Jorgenson and Lau (1971, 1973) to estmate relatve TFP levels n Canada and the Unted States. Ths framework was used extensvely by Jorgenson and hs assocates, ncludng Jorgenson, uroda and Nshmzu (1987), Jorgenson and uroda (1995), and uroda and Nomura (1999). Followng that tradton, relatve TFP levels can be assumed to reflect dfferences n technology levels snce the qualty of nputs s already taken nto account n ths framework. Based on a common framework usng comparable data sets for Canada and the Unted States, 1 our results show that n 1995, 23 of 33 Canadan ndustres had lower TFP levels than ther U.S. counterparts. 2 Our results also suggest that the relatve TFP level s an mportant element of nternatonal compettveness across ndustres. In fact, Canadan ndustres wth hgher TFP levels than ther U.S. counterparts tend to be more compettve n terms of relatve output prces. Over tme, however, movements n the exchange rate appear to be the most sgnfcant factor behnd nternatonal compettveness.

2 Lee and Tang From 1988 to 1995, the deprecaton of the exchange rate helped 9 ndustres become more compettve than ther U.S. counterparts. In addton, movements n the exchange rate concded wth movements n the relatve output prces of the prvate busness sector n the two countres over the perod. Focusng on a more recent perod, that between 1976 and 1995, Canada s prvate busness sector saw ts compettveness mprove relatve to that of the U.S. busness sector, even as ts TFP performance was not mprovng although a slght rebound has occurred n that respect snce The remanng sectons of the chapter are organzed as follows. In Secton 5.2, we construct PPPs for output and nputs, whle Sectons 5.3 and 5.4 are devoted to a comparson of TFP levels and nternatonal compettveness between Canadan and U.S. ndustres. In Secton 5.5, we dscuss the evoluton of TFP and compettveness n the Canadan and U.S. prvate busness sectors. We conclude our study n Secton Purchasng Power Partes for Output and Inputs IN THIS SECTION, WE DISCUSS the data and methodology used n constructng Canada-U.S. blateral PPPs for output and nputs n 33 ndustres. In ths context, t s useful to keep n mnd that the value of output s defned from the producer s pont of vew and the value of nputs, from the producerpurchaser s standpont. Ths has mplcatons for constructng PPPs, as wll be seen later. Frst, we group the 1992 Canadan and U.S. nput-output tables 3 nto 249 common commodty groups and 33 ndustres. 4 We then match 201 commodty PPPs 5 at purchasers prces wth commodtes n the I-O tables. Among the remanng 48 commodtes n the I-O tables, we frst dentfy 26 that have close substtutes among the 201 commodtes already matched, and then apply to them the PPPs of ther close substtutes. In the case of the remanng 22 commodtes, we use the 1993 market exchange rate. These commodtes are manly prmary goods (such as gran, wheat, copper, steel, and precous metals) that are heavly traded n North Amercan or world markets. The 249 PPPs and the I-O tables are used to develop PPPs for output and nputs other than labour

3 Productvty Levels and Internatonal Compettveness Purchasng Power Partes for Output The output PPP s defned as the rato of the amount of Canadan dollars receved by Canadan producers for output sold n Canada, to the amount of U.S. dollars receved by U.S. producers for sellng the same amount of output n the Unted States. Thus output PPPs are at producers prces, mplyng that we frst need to convert commodty PPPs at purchaser s prces, EPPPj, nto commodty PPPs at producers prces, PPP j, by peelng off tax and dstrbuton margns (the ndrect commodty tax margn and the transportaton and trade margns), usng the I-O tables of both countres. 7 We then proceed to construct output PPPs for each ndustry. The output PPP n ndustry s obtaned by aggregatng 249 commodty PPPs n translog form, usng nomnal shares n the commodty mx as weghts for ndustry : Q Q Q (1) ln ) 1/2 [ v, j (Can) + v, j (US)] (PPP, = 249 ln (PPPj ) j= 1 where v Q, j( S) s the value share of commodty j n ndustry n country S, estmated from the make matrces of the I-O tables Purchasng Power Partes for Intermedate Inputs Intermedate nputs nclude energy, materals, and purchased servces. Ther PPPs are computed n the same manner as output PPPs, but they are based on commodty PPPs at purchasers prces, whch nclude tax, transportaton, and trade margns. Wth ths n mnd, the PPP for ntermedate nputs n ndustry s defned as the translog aggregate of the 249 commodty PPPs: 249 M M M (2) ln( ) = 1/ 2[ v j ( Can ) + v, j ( US )] PPP, ln( EPPPj ), j= 1 where v M, j ( S) s the value share of goods (or servces) of type j that are used as ntermedate nputs n ndustry n country S, estmated from the use matrces of the I-O tables. Here, EPPP j, s the PPP at purchasers prces for commodty j as defned earler. 157

4 Lee and Tang Purchasng Power Partes for Captal Input As n Chapter 4, captal nput s broken down here nto four asset types machnery and equpment (M&E), non-resdental structures, nventores, and land. However, the prce data avalable only allow us to construct nvestment PPPs for M&E and structures. Followng Jorgenson and uroda (1995), and uroda and Nomura (1999), we aggregate 249 commodty PPPs to construct nvestment PPPs for new nvestment type k (M&E or structures) n ndustry from the purchasers standpont: (3) I I I ln( k ) 1/ 2[ v, k, j ( Can) + v, k, j ( US )] PPP, = 249, ln( EPPPj) j= 1 where v I ( S) s the value share of nvestment good j of type k n ndustry,, k, j estmated from the nvestment flow matrces of the I-O tables. We then derve a captal nput PPP for each type (M&E and structures) n ndustry by multplyng the rato of each type s rental prce for Canada relatve to the Unted States by ts correspondng nvestment PPP, I (4) P, k( Can)/ P, k( Can) I PPP, k = PPP, I, k P, k ( US )/ P, k( US) where P ( S), k s the captal nput prce of asset type k n country S, whle P I ( S), k s the nvestment prce ndex for that asset type. For each asset type, the rato of the captal nput prce to the nvestment prce ndex s the rental prce of captal nput of ths asset type. As descrbed n prevous chapters, the rental prce of captal nput s estmated by takng account of the rate of return on captal, economc deprecaton rates, and varous tax parameters n each country. Thus, n dervng captal nput PPPs, we mplctly assume that the relatve effcency of new captal goods n a gven ndustry s the same n both countres. However, the declne n the effcency of captal nput for each component s estmated separately for each country. We assume that the captal nput PPP for land s the same as that for structures. Furthermore, we assume that the captal nput PPP for nventores s the same as the weghted average of captal nput PPPs for M&E, structures and land. The total captal nput PPPs n ths paper are then derved by aggregatng ndvdual captal nput PPPs across p types of captal nput (M&E, 158

5 Productvty Levels and Internatonal Compettveness structures, land, and nventores), usng the average compensaton n the two countres for each type of captal nput as weghts: p (5) PPP ) = 1/ 2[ v k ( Can) + v, k ( US) ] ln(, ln( PPP, k ), k= 1 where v, k ( S) s the captal compensaton share of type k captal n ndustry n country S Purchasng Power Partes for Labour Input For each of the 33 ndustres, labour nputs n Canada and the Unted States are matched by sex, employment status, age, and educaton, as shown n Table 5.1. We estmate the labour nput PPP for ndustry by aggregatng the rato of hourly labour compensaton rates between the two countres over q types (112) of labour: L P, l ( Can) ln( ln L, l= 1 P, l ( US) q L L L (6) PPP ) = 1/ 2[ v, l ( Can) + v, l ( US) ] where P L, l ( S) s the average labour compensaton per hour of type l worker n ndustry n country S, and v L, l ( S) s the total labour compensaton share for that worker type. Table 5.1 Classfcaton of the Canadan and U.S. Workforce Worker Characterstcs Number of Categores Type Sex 2 Female; Male Employment Category 2 Pad Employees; Self-employed 1 Age ; ; 25-34; 35-44; 45-54; 55-64; 65+ Educaton Years Grade School; Some or Completed Hgh School; Some or Completed Post- Secondary School; Unversty or Hgher Notes: 1 U.S. self-employed ncludes unpad workers. 2 The age group s for Canada. 159

6 Lee and Tang Summary of Purchasng Power Partes Between Canada and the Unted States, 1993 PPPs for output and three types of nputs n 1993 are reported n Table The output PPPs are generally n lne wth the exchange rate (1.29 n 1993) for most ndustres. However, for coal mnng, tobacco, and electrc utltes, they are on the lower sde. Captal nput PPPs are hghly varable across ndustres. These varatons stem from the varatons n the rental prces of captal nput between the two countres snce captal nvestment prces are generally comparable. For nstance, the rental prce of captal nput n the motor vehcles, rubber and plastcs, and ndustral machnery ndustres s hgher n Canada than n the Unted States, whle the opposte s true n the paper and alled products, petroleum refnng, and other servces ndustres. The hgher rental prce of captal nput n other servces n the Unted States s manly due to a hgher rental prce n prvate educaton and legal servces n that country than n Canada. A close examnaton reveals that the substantal dfferences n the rental prces of captal nput noted between Canada and the Unted States are attrbutable to large dfferences n the captal compensaton fgures from the two countres I-O tables relatve to ther respectve captal stocks. Wth respect to the PPPs for labour nput, we frst observe that varatons across ndustres are very small. In addton, labour nput PPPs are below unty for 17 ndustres, whch s sgnfcantly below the exchange rate. Fnally, ntermedate nput PPPs are farly constant across ndustres and more or less equal to the exchange rate for all ndustres except tobacco. The Canadan tobacco ndustry pays a hgher prce for ntermedate nputs than does ts U.S. counterpart, manly because of the dfference n the taxaton on sem-fnshed tobacco products between the two countres. 160

7 Productvty Levels and Internatonal Compettveness Table 5.2 Purchasng Power Partes by Industry, 1993 (U.S. = 1.00) Industry Output Captal Input Labour Input Intermedate Inputs 1. Agrc., For. and Fsheres Metal Mnng Coal Mnng Crude Pet. and Gas Non-met. Mnng Constructon Food Tobacco Textle Apparel Lumber and Wood Furnture Paper Prntng Chemcals Petroleum Refnng Rubber and Plastcs Leather Stone, Clay and Glass Prmary Metals Fabrcated Metals Industral Machnery Electrcal Machnery Motor Vehcles Other Trans. Equp Msc. Manufacturng Trans. and Warehousng Communcatons Electrc Utltes Gas Utltes Trade Fnance, Ins. and Real Estate Other Servces Prvate Busness Note: 1 For value added from Statstcs Canada s Canada-U.S. GDP purchasng power party. 161

8 Lee and Tang 5.3 Relatve Productvty Levels BASED ON THE PPPS CONSTRUCTED ABOVE, we estmate relatve TFP levels between Canada and the Unted States for 33 ndustres. 9 As Jorgenson and Nshmzu (1978) for the comparson between Japan and the Unted States, our theoretcal framework for ths comparson s based on a translog producton functon orgnally ntroduced by Chrstensen, Jorgenson, and Lau (1971, 1973). Here, output s a translog functon of captal nput, labour nput, and ntermedate nputs, as well as a dummy varable equal to one for Canada and zero for the Unted States, and tme as an ndex of technology for each ndustry. However, as dd Jorgenson and uroda (1995), and uroda and Nomura (1999), we fnd that t s more convenent to work wth the dual prce functon of output to analyse nternatonal compettveness and relatve TFP levels. The dual prce functon s derved from the producton functon under compettve condtons. The prce functon for the th ndustry can be represented as: X' X t D X' XX X ln P = ln P α + αt+ α D + 1/2ln P β ln P + ln P (7) X' XD tt 2 td DD 2 + ln P β D + 1/2 β t + β td + 1/2 β D, where P s the output prce of the th ndustry; lnp X denotes {lnp lnp L lnp M }, a vector of logarthms of captal nput prce (P ), the labour nput prce (P L ), and the ntermedate nput prce (P M ) of the th ndustry; t denotes tme as an ndex of technology; and D s a dummy varable, equal to one for Canada and zero for the Unted States. X' β Xt t In ths presentaton, scalars {α t, α D, β t t, β t D, β DD }, the vectors {α X, β Xt, β XD }, and the matrx {β XX } are constant parameters. However, these parameters dffer among ndustres, reflectng dfferences among technologes. Wthn each ndustry, dfferences n technology among tme perods are represented by tme as an ndex of technology. Dfferences n technology between Canada and the Unted States are assocated wth the dummy varable. Based on the above prce functon, Jorgenson and uroda (1995), and uroda and Nomura (1999) show that dfferences n the logarthms of the TFP levels between Canada and the Unted States, v can be expressed as the negatve D 162

9 Productvty Levels and Internatonal Compettveness value of the dfferences between the logarthms of the output prces, less a weghted average of the dfferences between the logarthms of nput prces, (8) ln P(Can) ln P(Can) ln PL(Can) ln PM(Can) vd = v v L v M, P(US) P(US) PL(US) PM (US) =1, the average compensaton share of nput j n Canada and the Unted States for the th ndustry. The prce ratos n the above equaton are the PPPs for output and nputs. j j j where v /2[ v (Can) + v (US)] We frst calculate 1993 relatve TFP levels n Canada and the Unted States for 33 ndustres based on the estmated 1993 PPPs usng Equaton (8). We then use the TFP ndces constructed n the prevous chapter to estmate relatve TFP levels n other years. The estmated relatve TFP levels by ndustry are reported n Table 5.3. In 1995, Canada was less productve than the Unted States n 23 of the 33 ndustres. In partcular, Canada was much less productve n agrculture, forestry and fsheres; crude petroleum and gas; paper; prntng; rubber and plastcs; leather; stone, clay, and glass; fabrcated metals; ndustral machnery; and transportaton and warehousng. On the other hand, n 1995 Canada was sgnfcantly more productve than the Unted States n coal mnng, constructon, tobacco, petroleum refnng, electrc utltes, and gas utltes. To examne the trend n relatve TFP levels n Canadan and U.S. ndustres, we estmated the varance of relatve TFP levels by ndustry for the perod As shown n Fgure 5.1, the varance for all ndustres declned dramatcally n the 1960s. After 1970, however, t remaned farly stable. Ths mples that TFP performance n Canada and the Unted States converged across ndustres durng the 1960s. Indeed, n 19 of the 25 ndustres where Canada lagged behnd the Unted States wth respect to TFP levels n 1961, Canada mproved ts relatve TFP performance from 1961 to 1973; the largest mprovements were n those ndustres where TFP gaps were the wdest (coal mnng and communcatons). At the same tme, Canada lost some of ts relatve TFP advantage n 2 ndustres (tobacco and petroleum refnng) where that advantage was the largest n Between 1973 and 1988, the varance remaned more or less steady. Over ths perod, some lowproductvty Canadan ndustres were catchng up to ther U.S. counterparts, but ther relatve gans were modest. At the same tme, these gans were offset by U.S. ndustres catchng up to hghly productve Canadan ndustres 163

10 Lee and Tang (metal mnng, petroleum refnng, and both machnery ndustres). Over the perod, the varance of the relatve TFP gap between the two countres decreased. Most of the declne can be attrbuted to U.S. ndustres (such as metal mnng, coal mnng, and electrcal machnery) catchng up to, and n some nstances surpassng, the TFP levels of Canadan ndustres. Meanwhle, most Canadan ndustres that were less productve than ther U.S. counterparts ether were unable to catch up to U.S. TFP levels or only made modest gans. To gve another perspectve on ths ssue, we also examned the number of Canadan ndustres that were less productve than ther U.S. counterparts. That number decreased from 20 n 1961 to 17 n 1973, as shown n Table 5.3. However, t rose to 21 n 1988 and 23 n Thus the number of Canadan ndustres that were less productve than ther U.S. counterparts has ncreased snce These numbers provde a snapshot of performance n a gven year, but they do not help to assess the mprovement or deteroraton of Canada s relatve TFP performance over tme. Fgure 5.1 Varance of the Productvty Gap (n Logs) Among Industres 0.08 Varance

11 Productvty Levels and Internatonal Compettveness Table 5.3 TFP Levels n Canada Relatve to the Unted States, (U.S. = 1.00) Industry Agrc., For. and Fsheres Metal Mnng Coal Mnng Crude Pet. and Gas Non-met. Mnng Constructon Food Tobacco Textle Apparel Lumber and Wood Furnture Paper Prntng Chemcals Petroleum Refnng Rubber and Plastcs Leather Stone, Clay and Glass Prmary Metals Fabrcated Metals Industral Machnery Electrcal Machnery Motor Vehcles Other Trans. Equp Msc. Manufacturng Trans. and Warehousng Communcatons Electrc Utltes Gas Utltes Trade Fnance, Ins. and Real Estate Other Servces

12 Lee and Tang We now turn to that ssue. When we examne the performance of relatve TFP levels over tme, the pervasveness of the declne n Canada becomes evdent. From 1961 to 1973, only 9 Canadan ndustres experenced a declne n TFP relatve to ther U.S. counterparts. However, that number rose to 16 between 1973 to 1988 and to 17 between 1988 to In summary, the deteroraton of Canada s TFP levels relatve to those of the Unted States has become more wdespread across ndustres snce Compettveness n Canadan and U.S. Industres THIS SECTION ASSESSES DIFFERENCES IN COMP ETITIVENESS between Canadan and U.S. ndustres and lnks these dfferences to ther relatve TFP levels. Followng Jorgenson and uroda (1995), we measure compettveness by relatve output prces, defned as output PPPs dvded by the exchange rate ($CDN per $US). To facltate our analyss, we decompose relatve output prces nto relatve TFP levels and relatve captal, labour, and ntermedate nput prces. We rearrange Equaton (8) and dvde each prce rato by the exchange rate: D L L M M (9) ln RP = v + v ln RP + v ln RP + v ln RP, where RP s the relatve prce of output; v D s the TFP gap between Canada and the Unted States for ndustry ; and RP, RP L, and RP M are the relatve prces of captal, labour, and ntermedate nputs, respectvely. The relatve prces for output, for captal, labour, and ntermedate nputs, and for relatve TFP levels n 1995 are reported n Table 5.4. In 1995, more than half of Canadan ndustres had a lower relatve output prce than ther U.S. counterparts. Wth respect to captal nput, Canada had hgher captal nput prces than the Unted States n 27 ndustres. In partcular, Canadan captal nput prces were substantally hgher than U.S. prces n metal mnng, textles, apparel, furnture, paper, rubber and plastcs, prmary metals, motor vehcles, other transportaton equpment, and mscellaneous manufacturng n However, n some Canadan ndustres such as coal mnng, crude petroleum and natural gas, leather, and other servces captal nput prces were lower than 166

13 Productvty Levels and Internatonal Compettveness n the correspondng U.S. ndustres. As dscussed earler, t s helpful to keep n mnd that dfferences n relatve captal nput prces reflect dfferences not only n captal nvestment prces but also n the rental prce of captal nput. In contrast wth the stuaton regardng captal nput prces, all Canadan ndustres had an advantage over ther U.S. counterparts n terms of labour costs, and the varatons n relatve labour nput prces across ndustres were very small n As a result of ths dfference n labour costs, the ndustral structures of the two countres are also dfferent. Canadan ndustres are generally more labour-ntensve, whle U.S. ndustres tend to be more captalntensve. Ths s evdent when we compare captal ntensty (the rato of captal stock to hours) of the two countres. For nstance, n 1993, captal ntensty n Canada (captal stock PPP-based) was only 79 percent that of the Unted States. 10 Fnally, most Canadan ndustres pad almost the same prce for ther ntermedate nputs as dd ther U.S. counterparts. When examnng the lnks between compettveness, relatve TFP levels, and relatve nput prces, a smple correlaton among these varables s a good startng pont for dscusson. The correlaton coeffcent between relatve output prces and relatve TFP levels s 0.69 based on 1995 data, whle n the case of captal, labour, and ntermedate nputs, the coeffcents stand at 0.47, 0.16, and 0.12, respectvely. These coeffcents ndcate that varatons n relatve output prces across ndustres are strongly related to nter-ndustry dfferences n relatve TFP levels. We summarze the relatonshp between output prces and TFP levels by plottng relatve output prces aganst relatve TFP levels for 1995 across ndustres n Canada and the Unted States, as shown n Fgure 5.2. To better llustrate the relatonshp between comp ettveness and relatve TFP levels, we dvde the fgure nto four quadrants. In quadrants I and II are found those Canadan ndustres whch are less compettve than ther U.S. counterparts, whle quadrants III and IV show Canadan ndustres that are more compettve than ther U.S. equvalents. At the same tme, Canadan ndustres n quadrants II and III are more productve than ther U.S. compettors, whle relatvely less productve ndustres n Canada are located n quadrants I and IV. 167

14 168 Table 5.4 Relatve Prces* and TFP Levels by Industry, 1995 (U.S. = 1.00) Industry Output TFP Captal Input Labour Input Intermedate Inputs 1. Agrc., For. and Fsheres Metal Mnng Coal Mnng Crude Pet. and Gas Non-met. Mnng Constructon Food Tobacco Textle Apparel Lumber and Wood Furnture Paper Prntng Chemcals Petroleum Refnng Rubber and Plastcs Leather Stone, Clay and Glass Prmary Metals Fabrcated Metals Industral Machnery Lee and Tang

15 Table 5.4 (cont d) 169 Industry Output TFP Captal Input Labour Input Intermedate Inputs 23. Electrcal Machnery Motor Vehcles Other Trans. Equp Msc. Manufacturng Trans. and Warehouse Communcatons Electrc Utltes Gas Utltes Trade Fnance, Ins. and Real Estate Other Servces Note: * PPP rates dvded by the exchange rate. Productvty Levels and Internatonal Compettveness

16 Lee and Tang Fgure 5.2 Relatve Output Prces Aganst Relatve TFP Levels, 1995 (U.S. = 1.00) I 13 II Relatve Ouput Prces (5,12,25,10) (27,22) (26,15) IV 3 29 III Relatve TFP Levels Note: The numbers n ths fgure refer to those ndustres lsted n Table 5.2. In 1995, 15 Canadan ndustres were less compettve and less productve than the correspondng U.S. ndustres (quadrant I). In 7 ndustres (food, textles, apparel, paper, prntng, rubber and plastcs, and prmary metals), lower productvty combned wth hgher nput prces (affectng all three types of nputs) to reduce compettveness. Low nput prces n 6 of the remanng ndustres were not strong enough to offset the effects of lower productvty and make these ndustres more compettve. No ndustry was less compettve but more productve than ts U.S. counterpart (quadrant II). An examnaton of quadrant III reveals that 10 Canadan ndustres were more compettve and more productve than the correspondng U.S. ndustres. Seven of these coal mnng; constructon; lumber and wood; petroleum refnng; electrc utltes; fnance, nsurance, and real estate (FIRE); and trade were dentfed as havng relatvely lower nput prces than ther U.S. counterparts. The remanng 3 ndustres tobacco, gas utltes, and motor vehcles had hgher nput prces than ther U.S. compettors, but the dfference was not large enough to make them less compettve than the U.S. ndustres. 170

17 Productvty Levels and Internatonal Compettveness Fnally, quadrant IV shows the ndustres where Canada was more compettve but less productve than the Unted States chemcals; leather; ndustral machnery; electrcal machnery; mscellaneous manufacturng; communcatons; transportaton and warehousng; and other servces. Canada s compettve poston n those cases stemmed from lower nput prces rather than hgher TFP levels. Thus, t appears that the man factor behnd varatons n nternatonal compettveness across ndustres s the gap n relatve TFP levels. Fgure 5.3 Relatve Output Prces Aganst Relatve TFP Levels, 1988 (U.S. = 1.00) Relatve Output Prces I (19,32) 4 (21,1) 14 9 (26,15,18,11,12) 2 25 (10,5,7) 24 (27,28,22) II IV Relatve TFP Levels III Note: The numbers n ths fgure refer to those ndustres lsted n Table 5.2. However, movements n nternatonal compettveness over tme are strongly nfluenced by varatons n the exchange rate through relatve nput prces. For the purpose of llustraton, we compare nternatonal compettveness between 1988 and We plot relatve output prces aganst relatve TFP levels for 1988, as shown n Fgure 5.3, to facltate the dscusson. In 1988, only 8 Canadan ndustres were more compettve than ther U.S. counterparts, compared to 18 ndustres n Ths change s explaned by the fact that the Canadan dollar deprecated by more than 10 percent durng 171

18 Lee and Tang the ntervenng perod. If the exchange rate n 1995 had remaned at ts 1988 level, only 9 Canadan ndustres would have been more compettve than ther U.S. counterparts that year. In addton, several Canadan ndustres lumber and wood, chemcals, leather, ndustral machnery, motor vehcles, mscellaneous manufacturng, transportaton and warehousng, and communcatons would have lost ground and become less compettve than ther U.S. counterparts by Canada-U.S. Dfferences n Productvty and Internatonal Compettveness n the Prvate Busness Sector IN THIS SECTION, we examne the relatve performance of the Canadan and U.S. prvate busness sectors wth respect to TFP levels and compettveness over the perod. 11 We plot relatve TFP levels, relatve output and nput prces, as well as the exchange rate n Fgure 5.4. Fgure 5.4 Relatve Productvty and Compettveness Between Canada and the Unted States n the Prvate Busness Sector 1.8 U.S. = 1.00 Relatve Captal Input Prce 1.6 Exchange Rate 1.4 Relatve Output Prce TFP Relatve Labour Input Prce

19 Productvty Levels and Internatonal Compettveness The results show that Canada s TFP levels were catchng up to U.S. levels, rsng from 76 percent of the U.S. level n 1961 to almost 92 percent n However, the gap between the two countres began to wden after 1985 and stood at 12 percent n Meanwhle, Canada s relatve compettve poston worsened between 1963 and Ths deteroraton would have been much worse wthout the mprovements n relatve TFP levels that occurred n the Canadan busness sector over ths perod. Canada s compettve poston then mproved from 1976 to 1995, not as a result of TFP mprovements but of the Canadan dollar deprecaton through ts mpact on relatve nput prces. Relatve labour prces tend to be n lne wth relatve output prces. Despte the volatlty assocated wth the exchange rate, labour costs were consstently lower n Canada than n the Unted States over the 35-year perod In addton, the trend was farly stable over that perod. In contrast, relatve captal nput prces have been much more volatle. Snce 1975, relatve captal nput prces have declned, n lne wth the deprecaton of the Canadan dollar. In general, however, they have remaned hgher n Canada than n the Unted States, except n 1993 and Summary and Concluson THIS STUDY ILLUSTRATES that t s crtcal to use PPPs rather than the market exchange rate to assess the relatve productvty levels and nternatonal compettveness of two countres. PPPs vary across ndustres and types of output and nputs. Based on a common framework and usng comparable data sets, 23 of 33 Canadan ndustres had lower TFP levels compared to ther U.S. counterparts n Relatve TFP levels are an mportant element n determnng nternatonal compettveness. Our analyss ndcates that Canadan ndustres wth hgh relatve productvty compared to ther U.S. counterparts tend to be more compettve. Over tme, however, movements n the exchange rate appear to be the most sgnfcant factor behnd nternatonal compettveness. From 1988 to 1995, the fallng exchange rate helped 9 Canadan ndustres become more compettve than ther U.S. counterparts. Our analyss of the prvate busness sector renforces our fndngs at the ndustry level showng that movements n the exchange rate concde wth varatons n relatve output prces. Over the perod, durng whch 173

20 Lee and Tang the compettveness of Canada s prvate busness sector mproved relatve to that of the U.S. prvate busness sector, Canada s relatve TFP performance dd not mprove, despte a slght rebound after Ths study s a frst step towards understandng the dfferences n productvty and nternatonal compettveness between Canada and the Unted States. A number of refnements could prove frutful. Frst, t would be useful to collect more data comparng prces between Canada and the Unted States n order to ncrease the relablty of PPP estmates. A second avenue would be to expand captal asset categores for Canada to match Jorgenson s categores for the Unted States or those of the U.S. Bureau of Labor Statstcs. Future research may also beneft from an assessment of the comparablty of the two countres I-O tables, wth a specal focus on captal compensaton data. 174

21 Productvty Levels and Internatonal Compettveness Notes 1 A descrpton of the data s provded n the last chapter. 2 See Chapter 4 for data sources. 3 The I-O tables for both countres nclude make, use, fnal demand, and nvestment flow matrces. 4 The Canadan I-O tables are aggregated from 479 commodtes and 170 ndustres; the U.S. tables are aggregated from 541 commodtes and 541 ndustres. 5 These are 1993 PPPs, aggregated on the bass of data pertanng to more than 2,000 commodtes obtaned from Statstcs Canada. Statstcs Canada uses the data to estmate a blateral GDP PPP between Canada and the Unted States. 6 Although these 249 commodtes cover all commodtes n the I-O tables, some of them may not be used as nputs. In that case, they are not entered nto the calculaton of nput PPPs. 7 Hooper and Vrankovch (1995) adjust commodty PPPs for nternatonal trade n constructng output PPPs. Our analyss shows that ncorporatng ths methodology does not sgnfcantly change the results snce t s based on two restrctve assumptons: both export and mport prces equal world prces; and world prces equal the average of the prces n the two countres, weghted by ther expendtures. Snce we are unable to justfy these two assumptons, we use output PPPs wthout nternatonal trade adjustments. 8 The output PPP for the prvate busness sector s approxmated by the blateral value-added PPP for the total economy, as calculated by Statstcs Canada. 9 An assessment of the mplcatons of qualty adjustments to captal and labour nputs for estmatng relatve TFP levels s found n the Annex of ths chapter. 10 Canada s captal ntensty s based on an alternatve set of captal stock estmates produced by the Investment and Captal Stock Dvson of Statstcs Canada. These alternatve captal stock estmates are based on the same declnng-balance rates as those used n the Unted States. Captal ntensty for Canada would be much lower f we used captal stock data from Statstcs Canada s LEMS database (see detals n Appendx G). 175

22 Lee and Tang 11 The aggregate prce functon gves the value-added prce as a functon of captal and labour nput prces, so that the ntermedate nput prce s excluded. Smlar to Equaton (8), the dfference n the logarthms of the TFP levels between the Canadan and U.S. prvate busness sectors can be expressed as the negatve value of the dfference between the logarthm of the value-added prce and the weghted average of the dfference between the logarthms of captal and labour nput prces. 176

23 Productvty Levels and Internatonal Compettveness Annex: Qualty of Captal and Labour Inputs and Relatve TFP Levels IN THIS ANNEX, WE FIRST COMPARE relatve levels of captal and labour nput qualty n Canada and the Unted States and assess ther mplcatons for relatve TFP levels. Followng Dougherty (1992), we estmate relatve captal nput levels (PPP-adjusted) for Canada and the Unted States, wth each country s asset type (M&E, structures, land, and nventores) weghted by the average compensaton share n the two countres: (A-1) ln [ ( Can) / ( US) ] [ v, k( Can) + v,k( US) ] ln [ A, k( Can) /A, k( US) ] = 4 k= 1 1/2. Here, (S) denotes captal nput n ndustry n country S, v, k (S) s the captal compensaton share of type k captal asset n total captal compensaton n ndustry n country S, and A,k (S) s the net stock of type k captal asset n ndustry n country S. We then use the followng expresson to estmate relatve captal qualty levels for Ca nada and the Unted States: k k (A-2) ln [ q ( Can) / q ( US )] ln [ ( Can) / ( US )] ln [ A ( Can) / A ( US )] =, where A ( S) A ( S denotes the total captal stock n ndustry n country S. = 4 k= 1, k ) Lkewse for captal nput, relatve labour nput levels n Canada and the Unted States for ndustry can be expressed as: (A-3) L L ln [ L ( Can) / L ( US) ] [ v,j( Can) + v,j( US) ] ln [ H, j( Can) / H, j( US) ] = 112 j= 1 1/2, L where v, j (S) denotes the labour compensaton shares of type j workers n ndustry n country S, and H,j (S) denotes the hours worked by workers of type j n ndustry n country S. As wth captal qualty, relatve labour qualty levels are estmated by the followng expresson: (A-4) ln [ q L ( Can) / q L ( US )] ln [ L ( Can) / L ( US )] ln [ H ( Can) / H ( US )] where = 112 j= 1 ( S), j =, H ( S) H s the total number of hours worked by all types of workers n ndustry n country S. 177

24 Lee and Tang We then use the relatve qualty levels of captal and labour nputs to estmate relatve raw TFP levels (commonly referred to as relatve Solow resduals). The relatonshp between the relatve raw TFP levels and our estmates of relatve TFP levels s gven below: (A-5) ϕ D = v D + v L q (Can) L + q ln v (Can) ln, L q (US) q (US) D D L where ϕ s the raw TFP, v s the TFP, and v and v are the average captal and labour compensaton shares of the two countres n ndustry, as dscussed n Secton 5.3. In Table 5.A1, we report relatve qualty levels of captal and labour nputs and assess ther mplcatons for relatve TFP levels. Generally speakng, there are some varatons n the relatve levels of captal qualty across ndustres between Canada and the Unted States. On the other hand, labour qualty n Canada s slghtly lower than n the Unted States n vrtually all ndustres. In most cases, the effect of captal qualty s offset by labour qualty, resultng n a slght dfference between relatve raw TFP levels and the estmated TFP levels that ncorporate captal and labour nput qualty dfferences. 178

25 Productvty Levels and Internatonal Compettveness Table 5.A1 Relatve Captal and Labour Qualty Levels and TFP Levels, 1995 (U.S. = 1.00) In dustry Captal Qualty Labour Qualty TFP Raw TFP 1. Agrc., For. and Fsheres Metal Mnng Coal Mnng Crude Pet. and Gas Non-met. Mnng Constructon Food Tobacco Textle Apparel Lumber and Wood Furnture Paper Prntng Chemcals Petroleum Refnng Rubber and Plastcs Leather Stone, Clay and Glass Prmary Metals Fabrcated Metals Industral Machnery Electrcal Machnery Motor Vehcles Other Trans. Equp Msc. Manufacturng Trans. and Warehousng Communcatons Electrc Utltes Gas Utltes Trade Fnance, Ins. and Real Estate Other Servces Prvate Busness Sector