• Keine Ergebnisse gefunden

The Offshoring Strategies of US Multinational Corporations Operating in Canada

N/A
N/A
Protected

Academic year: 2022

Aktie "The Offshoring Strategies of US Multinational Corporations Operating in Canada"

Copied!
31
0
0

Wird geladen.... (Jetzt Volltext ansehen)

Volltext

(1)

Munich Personal RePEc Archive

The Offshoring Strategies of US

Multinational Corporations Operating in Canada

Feinberg, Susan and Keane, Michael

2009

Online at https://mpra.ub.uni-muenchen.de/55181/

MPRA Paper No. 55181, posted 10 Apr 2014 18:04 UTC

(2)

The Offshorin g Strategies of

U.S. M ultin ation al Corporation s Operatin g in Can ada

14

Susan Fein berg & M ichael Kean e

R utgers Un iv ersity Arizon a State Un iv ersity an d Un iv ersity of Techn ology , Sy dn ey

A

B S TRACT

SING CONFIDENTIAL DATA on 130 0 U.S.-based m ultinational corporation s (MNCs), we exam ine whether MNCs that m oved jobs out of Canada between 198 3 an d 20 0 3 system atically m oved the jobs elsewhere within the MNC. We also look at whether tren ds in the m ovem ent of jobs within MNCs differ for m an ufacturin g an d service in dustries. We fin d n o eviden ce of system atic offshorin g of Can adian jobs by U.S. MNCs: those that increase (shrin k) em ploym en t in Can ada ten d to exhibit the sam e pattern elsewhere within the firm . We do fin d, however, that the sectors with the fastest job growth by U.S. MNCs in Can ada are those with the lowest m edian real wages. Sim ilarly, we fin d that the relativ e im portance of U.S. MNCs’ Canadian operations seem s to decline over the 20 -year tim e window. U.S. MNC em ploym en t in Canada, relative to em ploym ent in other foreign coun tries, drops from 27.6 percent of total foreign em ploym ent in 198 3– 198 5 to 21.6 percent in 20 0 1– 20 0 3. This change in relative em ploym ent does not appear to be the result of job cuts in Canada but of U.S. MNCs’ choosing to grow

— including high-paying jobs — in countries other than Canada.

U

(3)

I

N TROD U CTION

H E PAST DECADE has witn essed sign ifican t growth in offshoring and outsourcing by firm s operating in Canada. In 20 0 4, a PricewaterhouseCoopers (PWC) study warned that at least 75 0 0 0 Canadian IT jobs would m ove offshore or be repatriated to the United States by 20 10 due to a rising Canadian dollar, higher overall tax rates and the erosion of Canada’s technological advantages (Scott, Ticoll an d Garn er 20 0 4). A particularly in terestin g feature of Can adian offshore outsourcing is the large share of total output in the Canadian econom y, particularly in the m an ufacturin g sector, that is owned by foreign firm s.

This share has risen durin g the past two decades from 44 percent in 198 7 to 52 percent in 1999 (Baldwin and Gu 20 0 5).1 Thus, an investigation of offshoring or outsourcing activities in Canada should take accoun t of the significant foreign ownership in the Canadian econom y. Foreign ownership im plies that offshoring and outsourcing decision s in Can ada m ay often be decision s about m ovem en ts of activities within m ultinational firm s. Sim ilarly, foreign firm s that have self-selected to locate in Canada m ay have very differen t decision processes from “dom estic” Canadian firm s regardin g relocatin g out of Canada.

In this research, we study the offshorin g decisions of U.S. MNCs with operations in Can ada.2 We exam in e the m ovem en t of activities within m ultinational firm s (“offshoring”) rather than activities that are con tracted to outside providers (“offshore outsourcin g”). Our research addresses two broad question s. First, is there eviden ce that activities m oved out of Canada by U.S. MNCs are system atically repatriated within the MNC back to the United States? Second, is there evidence that activities m oved out of Canada by U.S. MNCs are system atically m oved within the MNC to other countries? We exam in e tren ds from 198 3– 20 0 3 in the m ovem en t of activities within approxim ately 130 0 U.S. MNCs with operations in Canada, an d we investigate whether the trends differ for Canadian operation s differ in m an ufacturin g as opposed to service in dustries. We define an activity as a three-digit industry.

We look at changes in Canadian , U.S. and foreign (within-MNC) em ploym ent and wages as indicators of the m ovem ent of activities. This approach has several obvious advan tages an d disadvan tages. A key advan tage of our approach is the ability to detect m ovem ent of activities within firm s and the ability, albeit im perfect, to explore what happen s to activities that are m oved out of Canada. A clear lim itation of our approach is that, if we see that an activity has m oved out of Can ada, we can see on ly whether it has m oved elsewhere within the firm (“offshorin g” as opposed to “outsourcing”). We can n ot m easure the proportion of changes in activities that have been outsourced to other

T

(4)

foreign or dom estic providers. Thus, som e of the observed m ovem en t in activities in Can ada an d elsewhere within the MNC reflects changes due to decision s other than offshoring, for exam ple, in techn ology (e.g.

substitution of capital for labour) or “outsourcin g.”

It is n evertheless particularly in terestin g, for several reason s, to look at m ovem ents of activities within MNCs. First, there are sign ifican t inform ational costs of outsourcing, especially for firm s with n o experien ce operatin g in particular foreign m arkets such as In dia. To the extent that an MNC has existing operations in foreign m arkets, it m ay be easier to m ove activities within the firm . And, as m entioned above, due to the very large share of Can adian output owned by foreign firm s, m uch of the m ovem ent of activities out of Canada will be undertaken by MNCs an d m ay occur within MNCs.

The data set used in this paper is from the Ben chm ark an d An n ual Surveys of U.S. Direct In vestm en t Abroad from the U.S. Bureau of Econom ic Analysis (BEA). In this research, we use en terprise-level panel data that include detailed inform ation on the population of U.S.- based MNCs an d their foreign affiliates from 198 3– 20 0 3. The detailed m icrodata enable us to exam ine the association between chan ges in em ploym ent and wages in U.S. MNCs’ Canadian operations and chan ges elsewhere within the MNC (in the sam e three-digit sector), such as operations in the United States an d in other foreign (n on - Canadian) locations. As we detail in the data section, we rem ove estim ated data from the population. While this tends to im prove the reliability of the annual em ploym ent data, it also tends to skew the sam ple toward the large MNC paren ts and affiliates that report data.

We find n o evidence that MNCs that m ove jobs out of Canada in particular sectors m ove the jobs elsewhere within the firm . In deed, MNCs that reduce (increase) Canadian jobs in particular sectors typically reduce (in crease) jobs in these sectors in their U.S. an d other foreign operations.

This result holds in both m an ufacturing an d service in dustries an d for U.S. an d (n on -Can adian ) foreign operation s. This fin din g would seem to be “good news” for Canada in the sense that MNCs — firm s that m ay be best positioned to relocate activities out of Canada — do n ot appear to be system atically doing so.

H owever, we do fin d som e poten tially troublin g sign s for the Can adian econ om y durin g the 198 3– 20 0 3 tim e period. First, as shown in Figure 1, although job creation in Canada has outpaced job creation by U.S. MNCs in the Un ited States — m ost n otably in service industries

— the sectors with the greatest job growth in Can ada are those with the lowest m edian real wages (as shown in Figure 2).

(5)

FIGU RE 1

CAN AD IAN TO U .S .EMP LOYMEN T AN D CAN AD IAN/ U .S .DOLLAR P P P EX CH AN GE RATE

6.0%

7.0%

8.0%

9.0%

10.0%

11.0%

12.0%

13.0%

83-85 86-88 89-91 92-94 95-97 98-00 01-03

Cdn/U.S. Employment

0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30 1.35

PPP Exchange Rate

Canadian/U.S. Employment - All Industries

Canadian/U.S. Employment - Services PPP Exchange Rate

Figure 1 shows growth in Can adian -to-U.S.-paren t em ploym en t across all in dustries an d in service in dustries, defin ed as in dustries with SIC codes between 590 an d 90 0 .

(6)

FIGU RE 2

MED IAN WAGE ( EMP LOYEE COMP EN S ATION/ NU MB ER OF

EMP LOYEES) B Y DECILES OF CAN AD IAN AFFILIATE EMP LOYMEN T

CH AN GE ( % AN N U AL)

$22.00

$23.00

$24.00

$25.00

$26.00

$27.00

$28.00

$29.00

$30.00

-30.77% -13.79% -7.56% -3.94% -1.19% 1.77% 4.37% 7.95% 14.40% 30.56%

below 10th

10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100

(000's 1983 US$)

Figure 2 reports m edian levels of Can adian real wages in 198 3 U.S. dollars, grouped by deciles of an n ual percen tage growth in affiliate em ploym en t. The n um bers above the decile ran ges are the m edian s of affiliate em ploym en t growth (% chan ge) in the given decile.

From a policy standpoin t, we see two poten tial areas of con cern.

First, the real wages an d em ploym en t growth of Canadian affiliates are extrem ely sen sitive to Can adian / U.S. exchan ge rate m ovem en ts over the tim e win dow. Relative to the foreign (n on-Can adian ) subsidiaries of U.S. MNCs, the real wages of Can adian subsidiaries decrease from 146 percen t in 198 3– 198 5 to 10 3 percen t in 20 0 1– 20 0 3. Secon d, U.S.

MNCs’ em ploym ent in Canada is growin g m uch m ore slowly than in other coun tries (excludin g the Un ited States). Specifically, U.S. MNC em ploym ent in Can ada relative to em ploym ent in other foreign coun tries drops from 27.6 percent of total foreign em ploym ent in 198 3–

198 5 to 21.6 percent in 20 0 1– 20 0 3. This change in relative em ploym en t does n ot appear to be the result of job cuts in Canada but of U.S. MNCs’

choosing to grow — including high-payin g jobs — in coun tries other than Canada. From a policy standpoin t, this raises con cern s whether

(7)

the faster foreign job growth in dicates a potential strategic reorientation of U.S. MNCs toward region s other than North Am erica.

The rem ain der of this paper is organ ized as follows. In the n ext section , we discuss the BEA data an d the sam ple we constructed in this research. In the following section, we exam in e the association between chan ges in Can adian real wages an d em ploym en t an d sim ilar chan ges in U.S. MNC paren ts. We then in vestigate the relation ship between changes in em ploym ent and wages in U.S. MNCs’ Canadian operations relative to their other foreign affiliates in the sam e in dustry. The last section discusses planned extensions to the current draft of this research an d con cludes the paper.

D

ATA

S MENTIONED ABOVE, the data in this study were obtained from the U.S. Departm en t of Com m erce’s Ben chm ark an d An n ual Surveys of U.S. Direct In vestm en t Abroad, adm in istered by the Bureau of Econom ic Analysis (BEA). These surveys are the m ost com prehensive m icropan el data available on the activities of m ultin ation al firm s.

For this study, we use the BEA data at several different levels of aggregation . First, we use data on U.S. MNC paren ts at the en terprise level. Secon d, we use en terprise-level data on Canadian affiliates, which we aggregate up to the three-digit in dustry3 level for each U.S. parent.4 Third, we use data on other foreign affiliates of the sam e MNC paren t as the focal Can adian affiliate. Sin ce we are in terested in chan ges in sam e- in dustry activities, we aggregate data on other (non -Canadian ) foreign affiliates by paren t-in dustry-year an d keep only the data on com posite

“foreign ” affiliates with the sam e parent in the sam e industry as the Canadian affiliates.

In the tim e period used in this study (198 3– 20 0 3), there are approxim ately 30 0 0 U.S. MNC paren ts with usable Canadian affiliate data.5 These 30 0 0 parents m ap onto approxim ately 40 0 0 0 affiliate- year observation s. As discussed in detail in our other work (Fein berg an d Kean e 20 0 1, 20 0 6), the BEA data con tain both reported an d estim ated observation s. In general, larger paren ts an d affiliates ten d to have a greater proportion of reported-to-total observation s. We use on ly reported data on Can adian affiliates. This screen reduces the n um ber of parents to 1575 an d the num ber of affiliate-year observation s to approxim ately 16 0 0 0 . Our final screen s in volved droppin g affiliates with fewer than 10 em ployees (approxim ately the bottom percentile) and those with m issing com pensation, industry or parent data.

To exam in e chan ges in activity within MNCs, we use reported data on other foreign MNC affiliates (of the sam e U.S. paren t) in the sam e in dustry as the Can adian affiliate. Sin ce the BEA’s defin ition of

“affiliate” correspon ds roughly with the busin ess-un it level, m ost

A

(8)

affiliates report data in on ly on e in dustry. For exam ple, 8 5 percen t of all Canadian affiliates sell at least 75 percen t of their total sales in on e in dustry. We therefore use the prim ary industry of affiliates to reflect

“activities.”

To obtain com parison data for other foreign affiliates within the sam e MNC, we aggregate reported em ploym ent and wage data for all affiliates of U.S. MNC parents and subtract the Canadian data from this sum . Approxim ately 20 percen t of the (reported) total MNC affiliate em ploym ent data is Canadian affiliate data. After elim inating these observation s an d usin g on ly foreign affiliate observations that paired with reported Can adian affiliate data, we were left with roughly 8 50 0 affiliate-year observations on 190 0 affiliates in the Canadian an d “other foreign affiliate” com parison data set.6

Since affiliates in the BEA data set report data at the business-unit level, it is straightforward to obtain sales, com pensation an d wage data for sin gle in dustry en tities. This is som ewhat m ore difficult with regard to U.S. paren ts, sin ce these report consolidated input data. To give som e in sight into U.S. paren ts’ diversification, the m edian parent sells 96 percen t of total sales in a sin gle in dustry. At the 25th percen tile of sin gle-in dustry/ total sales, paren ts sell 73 percent of total sales in a sin gle in dustry.7 In the BEA data, paren ts report ann ual sales in up to eight in dustries, but un fortun ately other data such as em ploym ent an d costs are n ot sim ilarly disaggregated. Early in this project, we exam in ed the feasibility of m atchin g BEA data to Com pustat data on U.S. parents at the busin ess-segm ent level. H owever, as segm en t reportin g is volun tary an d n ot en ough paren ts reported usable data, usin g this data source was not deem ed worthwhile. We therefore used the BEA data on U.S. paren ts an d used the prim ary in dustry of the paren t as a m easure of “activity.” Although there is clearly som e loss of precision , we believed this was the best option . Sim ilar to m atchin g affiliates with each other by in dustry, we use reported paren t em ploym en t an d com pen sation data an d keep the paren ts that are in the sam e prim ary industry as Canadian affiliates with reported data. The Can adian affiliate– U.S. paren t data set has approxim ately 750 0 affiliate-year observations on 118 0 U.S. MNC paren t– Can adian affiliate pairs.

(9)

70.0%

75.0%

80.0%

85.0%

90.0%

95.0%

100.0%

83-85 86-88 89-91 92-94 95-97 98-00 01-03

Cdn-to-U.S. Wage

10.5%

10.7%

10.9%

11.1%

11.3%

11.5%

11.7%

11.9%

12.1%

12.3%

Cdn-to-U.S. Employment

O

FFS H ORIN G IN

N

ORTH

A

MERICA

: W

AGES AN D

E

MP LOYMEN T IN

C

AN AD IAN AFFILIATES AN D

U .S . P

AREN TS

H E PERIOD 198 3– 20 0 3 saw considerable chan ges in the distribution of U.S. MNCs’ jobs and wages in the United States and Canada. As shown in Figure 3, Canadian em ploym ent as a percent of U.S. parent em ploym en t in creases from 10 .8 percen t to 12.2 percen t over the sam ple win dow, but Can adian real wages (defin ed as total em ployee com pensation divided by em ploym ent) as a percent of U.S. parent real wages drop from a high of close to parity to approxim ately 8 0 percen t.

FIGU RE 3

CAN AD IAN-TO-U .S .EMP LOYMEN T AN D WAGES ( EMP LOYEE COMP EN S ATION/ EMP LOYMEN T) ,SAME-IN D U S TRY PAREN TS AN D

AFFILIATES

T

Cdn-to-U.S. Wage Cdn-to-U.S. Employment

(10)

Cdn/U.S. Wage - All Cdn/U.S. Wage - Services PPP Exchange Rate

Not surprisin gly, the large drop in relative real Canadian wages is tied closely to real exchan ge rate m ovem en ts, since the BEA data is reported in U.S. dollars. H owever, this dollar declin e in relative real wages is m ost pron oun ced in service in dustries, as shown in Figure 4.

Although Canadian affiliates’ real wages across all industries in crease by 30 percent from 198 3– 20 0 3 (and U.S. parents’ real wages increase by 32 percen t), the gap between parent an d affiliate wages n arrows in m anufacturing industries and widens in service in dustries. By the 20 0 1– 20 0 3 period, Can adian service wages are only 67 percent of service sector wages for U.S. paren ts while m an ufacturin g wages of Canadian affiliates are approxim ately 8 5 percen t of their U.S. paren ts’

wages. (See Figures A-1 an d A-2.)

FIGU RE 4

CAN AD IAN-TO-U .S .WAGES AN D P P PEX CH AN GE RATES,SAME- IN D U S TRY U .S .PAREN TS AN D CAN AD IAN AFFILIATES,ALL

IN D U S TRIES AN D SERVICES ON LY

70.0%

75.0%

80.0%

85.0%

90.0%

95.0%

100.0%

83-85 86-88 89-91 92-94 95-97 98-00 01-03

Cdn Wage/U.S. Wage

0.90 1.00 1.10 1.20 1.30 1.40 1.50

PPP Exchange Rate

(11)

Cdn-to-U.S. Wage Cdn-to-U.S. Employment

Although job growth in the Canada service sector outpaced U.S. job growth (Can adian affiliate em ploym en t in creased from 7.4 percen t to 9.1 percen t of U.S. parent em ploym ent in services), Canadian affiliate wages in services declin ed sharply relative to those in the United States (see Figure 5). This can also be seen in Figure A-1. In deed, Can adian affiliate wages in services decreased in real term s at the m edian from the 1992– 1994 to the 20 0 1– 20 0 3 period. This m ay n ot be surprisin g given the alm ost 40 percen t drop in the Can adian PPP exchan ge rate durin g this period. H owever, in real term s, Can adian m anufacturing wages in creased sign ifican tly durin g the sam e tim e period, as can be seen in Figure A-2.

FIGU RE 5

CAN AD IAN TO U .S .WAGE AN D EMP LOYMEN T IN SERVICE

IN D U S TRIES,SAME-IN D U S TRY U .S .PAREN TS AN D CAN AD IAN

AFFILIATES

70.0%

75.0%

80.0%

85.0%

90.0%

95.0%

83-85 86-88 89-91 92-94 95-97 98-00 01-03

Cdn-to-U.S. Wage

6.0%

6.5%

7.0%

7.5%

8.0%

8.5%

9.0%

9.5%

10.0%

Cdn-to-U.S. Employment

(12)

We n ext look at the question whether there is system atic change in job growth, in the sam e or opposite direction, in Can adian affiliates an d their U.S. paren ts. In other words, if we fin d that U.S. paren ts grow while their sam e-in dustry Can adian affiliates shrin k, this would suggest that activities were being m oved out of Canada and repatriated to the United States. We investigate this question by looking at m edian parent (affiliate) job growth by the deciles of affiliate (parent) job growth. We m easure job growth as the an n ual percen tage chan ge in em ploym en t from (t-1) to t for paren ts an d affiliates. All m easures are calculated at the paren t an d affiliate levels usin g the m icro data. We rem oved approxim ately 10 percent of observations at the 5th and 95th percentiles, and we conditioned the deciles on non-zero year-to-year job growth.8 Paren t (affiliate) job growth m easured at the deciles of affiliate (paren t) job growth are shown in Figures 6 an d 7.

FIGU RE 6

PERCEN TAGE EMP LOYMEN T GROW TH IN U .S .MN CPAREN TS B Y

DECILES OF SAME-IN D U S TRY CAN AD IAN AFFILIATE EMP LOYMEN T

GROW TH,19 8 3 – 2 0 0 3

-32.1%

0-10 -14.1%

10-20 -7.5%

20-30 -4.0%

30-40 -1.3%

40-50 1.6%

50-60 4.1%

60-70 7.5%

70-80 14.0%

80-90

29.6%

90-100

-0.02 -0.015 -0.01 -0.005 0 0.005 0.01 0.015 0.02 0.025 0.03

(13)

In Figure 6, the n um bers alon g the X-axis are the decile m edian s for affiliate job growth. The vertical bars give the values of U.S. paren t job growth for each decile of affiliate job growth. Clearly, when affiliates are losin g jobs, their U.S. paren ts are also losin g jobs, an d when affiliates are gain in g jobs, their sam e-in dustry paren ts are also growin g.

If jobs were system atically bein g m oved out of Canada in to the United States, we would expect to see large positive growth in the em ploym en t of U.S. parents whose Canadian affiliates are losin g jobs or at least are in sm aller deciles of job growth. No such pattern is eviden t here. A sim ilar pattern for chan ges in affiliate em ploym en t by deciles of U.S.

parent job growth is shown in Figure 7.

FIGU RE 7

PERCEN TAGE EMP LOYMEN T GROW TH IN CAN AD IAN AFFILIATES B Y

DECILES OF SAME-IN D U S TRY U .S .PAREN T EMP LOYMEN T GROW TH, 19 8 3 – 2 0 0 3

-28.2%

0-10 -11.4%

10-20 -5.4%

20-30 -2.4%

30-40

-0.5%

40-50 1.9%

50-60 4.0%

60-70 7.6%

70-80 13.4%

80-90 29.8%

90-100

-0.04 -0.03 -0.02 -0.01 0 0.01 0.02 0.03 0.04

(14)

These charts look very m uch alike. The scale is larger in Figure 7, reflectin g the closer relation ship between affiliate growth an d paren t growth but in both cases, growing Can adian affiliates have growin g U.S.

paren ts an d vice versa.

In the Appen dix, we provide further detail on the growth of parents an d affiliates in service an d m an ufacturin g in dustries (Figures A-3 to A- 5). In gen eral, the growth patterns, broken down in to service and m an ufacturin g in dustries, resem ble the pattern s shown in Figures 6 an d 7. H owever, affiliate (paren t) job growth in services is n ot con sisten tly the sam e sign as paren t (affiliate) growth in all the deciles.

For exam ple, in on e decile, U.S. parents’ m edian job growth is - 1.7 percent and the corresponding Canadian affiliates’ m edian job growth is 3.2 percen t. The relationships shown in Figures A-3 an d A-4 are som ewhat m ore variable because of sm aller sam ple sizes.9 H owever, the sam e gen eral pattern s we see in Figures 6 an d 7 are eviden t.

It is worth repeatin g here that all the charts in this research show changes and levels within MNCs aggregated by tim e, industry or both in order to describe them graphically. So the growth of paren ts an d affiliates shown in Figures 6 an d 7 is growth within the sam e firm s.

Sin ce paren ts an d affiliates operate in the sam e in dustries, we can in terpret these growth com parison s as chan ges in wages an d em ploym ent by activity within firm s.10 Sim ilarly, the m edian wage, grouped by deciles of Can adian affiliate em ploym en t growth (see Figure 2), also reflects within -firm values. In the next section , we exam ine Canadian affiliate em ploym ent, wages and job growth relative to other foreign (non -Canadian) sam e-industry affiliates in the sam e MNCs.

O

FFS H ORIN G B EYON D TH E

U

N ITED

S

TATES AN D

C

AN AD A

: W

AGES AN D

E

MP LOYMEN T IN

C

AN AD IAN AFFILIATES AN D

O

TH ER

F

OREIGN

( N

ON

-C

AN AD IAN

) MN C A

FFILIATES

ETWEEN 198 3 AND 20 0 3, U.S. MNCs expanded significantly outside North Am erica. From the standpoint of Canadian MNC affiliates, the foreign expansion of U.S. parents m ean t a declin e in the relative im portance of Canadian operations to the MNC as a whole. The United Kingdom , France and Germ any were the recipien ts of large in flows of U.S. foreign direct in vestm en t (FDI) in the 198 0 s and early 1990 s. In the m id-to-late 1990 s, U.S. MNCs expan ded operation s sign ifican tly in both in dustrialized an d developing coun tries in Asia an d in Eastern Europe.11 Because Canada had enjoyed a dom inant position as an outward location of U.S. FDI for m an y years, relatively less n ew investm ent flowed into Canada an d the Can adian share of em ploym en t in total MNC affiliate em ploym ent dropped from 28 percent in 198 3–

B

(15)

Cdn/MNC Wage Cdn/MNC Employment

198 5 to 22 percent in 20 0 1– 20 0 3 (see Figure 8 ). Although Canadian-to- MNC wages declined con sistently throughout the period, the Canadian em ploym en t share recovered slightly from 22 percen t to 24.5 percen t from 1992– 1994 to 1998 – 20 0 0 and then dropped back to 22 percent in 20 0 1– 20 0 3.

FIGU RE 8

CAN AD IAN AFFILIATE TO NON-CAN AD IAN AFFILIATE ( SAME MN C, SAME IN D U S TRY)EMP LOYMEN T AN D WAGES

90%

100%

110%

120%

130%

140%

150%

83-85 86-88 89-91 92-94 95-97 98-00 01-03

Cdn-to-MNC Wage

20%

21%

22%

23%

24%

25%

26%

27%

28%

29%

30%

Cdn-to-MNC Employment

(16)

Canadian wage MNC wage C$ / US$ PPP

Canadian wage MNC wage C$ / US$ PPP

We would have to kn ow a great deal about the operatin g strategies of MNCs to m ake an y qualitative in feren ces based on the declin in g relative em ploym en t share of Can adian affiliates. After all, although the Canadian econom y is sm all relative to those of the United Kingdom , France, Germ any, Italy and J apan, it n evertheless rem ain s the location of m ore than on e fifth of total foreign jobs of U.S. MNCs. So the fact that these MNCs are expan din g m ore quickly in to n ew m arkets should not pose a threat to Can ada — as lon g as the Can ada continues to host high-quality, high-wage, skill-in ten sive jobs.

The chan ges in the real wages of Canadian an d non-Canadian (sam e-in dustry, sam e-MNC) affiliates of U.S. MNCs in m anufacturing an d service in dustries are shown in Figures 9 and 10 . As is eviden t, the wage gap in services between Canadian and non -Canadian MNC affiliates widen ed sign ifican tly over the sam ple period, and n on - Can adian affiliates’ service sector wages were higher than Can adian wages in all but the first three-year period. This is sim ilar to chan ges in Canadian and U.S. parent service-sector wages shown in Figure A-1.

Despite the steep declin e in the Can adian / U.S. exchan ge rate begin n in g in 198 9– 1991, Can adian affiliates’ real wages in m an ufacturin g rem ain ed above real m an ufacturin g wages in other sam e-in dustry MNC affiliates outside Can ada, although the gap declined.

FIGU RE 9

MED IAN REAL WAGES OF CAN AD IAN AN D NON-CAN AD IAN

AFFILIATES ( SAME MN C,SAME IN D U S TRY) ,SERVICE IN D U S TRIES

16 20 24 28 32 36

83-85 86-88 89-91 92-94 95-97 98-00 01-03

(000's 1983 US $)

0.90 1.00 1.10 1.20 1.30

C$ / US $ PPP

(17)

12 14 16 18 20 22 24 26 28 30

83-85 86-88 89-91 92-94 95-97 98-00 01-03

(000's 1983 US $)

0.90 1.00 1.10 1.20 1.30

CD $ / US $ PPP

Canadian wage MNC wage C$ / US$ PPP

FIGU RE 10

MED IAN REAL WAGES OF CAN AD IAN AN D NON-CAN AD IAN

AFFILIATES ( SAME MN C,SAME IN D U S TRY) ,MAN U FACTU RIN G

IN D U S TRIES

With the n ext set of charts, we look at the question whether there is system atic eviden ce of n on -Can adian affiliate job growth in MNCs with shrinking Canadian affiliate em ploym ent. If we find that non-Canadian affiliates grow while sam e-in dustry Can adian affiliates in the sam e MNC shrin k, this would suggest that activities were bein g m oved out of Canada and into other countries. We in vestigate this question by looking at m edian non-Canadian (Canadian) job growth by the deciles of Canadian (non-Canadian) affiliate job growth. We m easure job growth as the ann ual percentage chan ge in em ploym ent from (t-1) to t for Canadian and non-Canadian affiliates. The results are shown in Figures 11 an d 12, which are an alogous to Figures 6 an d 7 (that com pared Canadian affiliate and U.S. parent job growth).

We fin d pattern s very sim ilar to those that com pared Can adian affiliate an d U.S. paren t job growth. In gen eral, Can adian (n on - Canadian) affiliates grew in sectors in which non-Canadian (Canadian) affiliates grew, an d vice versa. The pattern s are m ore strikin g when we com pare the m edian growth of other sam e-in dustry, sam e-MNC affiliates at (t+1) to deciles of Canadian affiliate growth from (t-1) to t.

(18)

FIGU RE 11

PERCEN TAGE EMP LOYMEN T GROW TH IN CAN AD IAN AFFILIATES B Y

DECILES OF NON-CAN AD IAN AFFILIATE ( SAME IN D U S TRY,SAME

MN C)EMP LOYMEN T GROW TH,19 8 3 – 2 0 0 3

-64.3%

0-10

-27.7%

10-20 -12.5%

20-30

-5.9%

30-40 -1.6%

40-50 2.0%

50-60 6.4%

60-70 12.6%

70-80 25.5%

80-90

62.1%

90-100

-0.025 -0.02 -0.015 -0.01 -0.005 0 0.005 0.01 0.015 0.02 0.025

As is illustrated in Figure 11, sectors where n on -Can adian (sam e in dustry, sam e MNC) affiliates grew, Can adian affiliates also ten d to grow, an d where n on -Can adian affiliates’ em ploym en t declin es, Can adian affiliates’ em ploym en t ten ds to decline as well. Again , sim ilar to the com parison s with U.S. paren ts, MNCs seem to grow or con tract sim ilar activities in m ultiple m arkets at the sam e tim e. If activities were being system atically offshored from Canada to non -Can adian MNC affiliates, we would expect to see declining Canadian affiliate em ploym ent corresponding to increased em ploym en t in other sam e- in dustry affiliates. We see n o eviden ce of such pattern s.

The growth in non-Canadian affiliates at (t+1) grouped by deciles of Canadian affiliate growth is shown in Figure 12. Sim ilar to Figure 11, we see n o eviden ce that jobs are system atically m oving out of Canadian affiliates into n on-Canadian affiliates within MNCs.

(19)

FIGU RE 12

PERCEN TAGE EMP LOYMEN T GROW TH IN NON-CAN AD IAN AFFILIATES AT (T + 1) B Y DECILES OF CAN AD IAN AFFILIATE ( SAME IN D U S TRY, SAME MN C) EMP LOYMEN T GROW TH,19 8 3 – 2 0 0 3

-30.7%

0-10 -13.9%

10-20 -7.5%

20-30 -3.9%

30-40 -1.2%

40-50 1.8%

50-60

4.4%

60-70 8.0%

70-80 14.4%

80-90 30.6%

90-100

-0.03 -0.025 -0.02 -0.015 -0.01 -0.005 0 0.005 0.01 0.015 0.02

Fin ally, we exam in ed em ploym en t growth in Can adian and n on - Canadian affiliates over tim e in m an ufacturin g an d service industries.

We in clude these figures in the Appen dix (Figures A-6 to A-8 ). Not surprisin gly, given the fin dings in Figures 11 an d 12, there is a close correspondence between Canadian and non -Can adian affiliate growth over tim e. Across all in dustries, the em ploym ent growth of Canadian an d n on -Can adian affiliates has the sam e sign in six of the seven three- year periods. In terestin gly, there is a closer correspon den ce between Canadian and n on -Can adian affiliate growth in services than in m an ufacturing. In three of the seven tim e periods, m anufacturing em ploym ent growth of Canadian and non-Canadian affiliates has opposite sign s. In con trast, growth in services has the sam e sign in every tim e period in Canadian and non-Canadian affiliates.

(20)

D

IS CU S S ION

, E

X TEN S ION S AN D

C

ON CLU S ION S

N TH IS STUDY, we find that in U.S. MNCs with growing (shrinking) Canadian affiliate em ploym ent, the sam e pattern s of em ploym en t growth (shrin kage) appear in other division s — both the U.S. paren t an d other foreign (non -Canadian) affiliates. Thus, we find n o evidence of system atic offshoring by U.S. MNCs with Canadian operations.

The eviden ce shown in this paper needs to be interpreted with som e caution . First, sin ce we excluded estim ated data, the MNCs left in the sam ples used here are prim arily large firm s. So we do n ot capture the extent to which offshorin g is occurring in sm aller MNCs or sm all, n ew division s of existin g MNCs.

Second, offshoring within MNCs m ight occur in differen t in dustry sectors. For exam ple, Gen eral Motors Canada’s m anufacturing operations (industry code 371) m ight m ove m any adm inistrative or clerical fun ction s offshore. If these offshore fun ction s were classified under an in dustry code for business services, we would n ot capture them in the an alyses shown here. H owever, we believed we would lose even m ore precision by aban don in g the “activity within-firm ” approach used here. For exam ple, if we dropped the “activity” and looked sim ply at changes in em ploym ent in Canada an d elsewhere within the MNC, we would have n o evidence at all that the Canadian activity m ight have been m oved within the firm .

Third, as m entioned in the introduction, this is a sim ple descriptive analysis. If we find, for exam ple, that em ploym ent in a particular activity within an MNC shrinks in both Canada and the Un ited States, m any things other than offshoring could be driving the em ploym ent change. The MNC could be substituting capital for labour, dem and for the MNC’s products m ight be declining, etc. Sim ilarly, our data are not appropriate for exam in in g outsourced activities. If an activity shrinks in Canada because it is m oved out of the MNC — either within Can ada or offshore — we can n ot exam in e what happens to that activity. On ce an activity is m oved outside the MNC, it is no longer recorded in our data.

Thus, even with the m icro, within-firm -level pan el data used here, on e can really provide on ly suggestive eviden ce on offshorin g within MNCs. Clearly, future work on this topic will n eed to use a variety of different data sources to better understand offshoring and outsourcing.

To investigate these activities at a m icro, within-firm level, evidence based on case studies by researchers workin g within firm s would be an excellen t com plem en t to panel data analyses.

From a policy standpoin t, our results poin t to several poten tial areas for further in vestigation . First, the m ore rapid growth of low- wage service jobs in Can ada by U.S. MNCs is a potentially alarm ing trend. It would be useful to investigate whether the declin e in real wages in the service sector shown here is sim ply an artifact of the lower real

I

(21)

exchan ge rate, or whether the lower service wages poin t to a potential Canadian com petitiveness problem in service in dustries. There is som e eviden ce of the latter problem . For exam ple, our results show that, despite the lower real exchan ge rate, m edian Can adian m anufacturin g wages rem ain higher than those in other foreign affiliates of U.S. MNCs.

H owever, we fin d the opposite pattern for Can adian an d other foreign affiliate wages in services. In deed, in the 198 6– 198 8 tim e period (before the real exchan ge rate began its steepest declin e), the m edian wage in services in non-Canadian foreign affiliates exceeded the Can adian wage for the first tim e. As the exchan ge rate fell in subsequen t periods, the gap between Canadian and other foreign affiliate wages in services sim ply got wider. Again, we cann ot show whether this is due to the type of jobs created in Can adian and other foreign affiliates or whether this is just driven by exchan ge rates. But sin ce there is a clear differen ce in the chan ges over tim e in relative m an ufacturin g an d service wages in Can ada an d the rest of the MNC, there is som e reason for con cern about service sector wages.

On the positive side, service jobs are bein g created by U.S. MNCs in Canada at a faster pace than they are bein g created in the Un ited States.

The policy challen ge is to create an environm ent in which Canada is the preferred location for U.S. MNCs’ high -wage, high-quality service jobs.

An other poten tial area for con cern is the declining im portan ce of Can adian operation s (as m easured by the share of total MNC em ploym ent in Canada) to the MNC as a whole. As we indicated before, on e would have to un dertake a rigorous study of MNC strategy to understand the im plications of this change (for exam ple, the im plication for the creation of executive and m anagerial jobs in Can adian affiliates). Our results suggest that such a study m ight produce valuable in sights in to the future role of Canadian operations in U.S. m ultin ation al firm s.

(22)

E

N D N OTES

1 A study by the Public Policy Forum / In form ation Techn ology Association of Can ada in dicates that, in 20 0 4, Can ada was a n et recipien t of Am erican outsourcin g. It is thu s possible that som e of the percen tage in crease in the foreign -own ed share of Can adian in dustry m ay be due to “offshorin g” from the Un ited States (Public Policy Forum an d ITAC Roun dtable 20 0 4).

2 The statistical an alysis of firm -level data on U.S. m ultin ation al corporation s reported in this study was con ducted at the In tern ation al In vestm en t Division , Bureau of Econ om ic An alysis, U.S. Departm en t of Com m erce, under arran gem en ts that m ain tain ed legal con fiden tiality requirem en ts. Views expressed are those of the authors an d do n ot n ecessarily reflect those of the Departm en t of Com m erce.

3 In dustries are in BEA ISI codes, which are very sim ilar to SIC codes at the sam e level of aggregation .

4 Sin ce som e U.S. MNC paren ts have several Can adian affiliates in the sam e in dustry in a given year, an d we are exam in in g chan ges in wages an d em ploym en t at the activity (in dustry) level within firm s, we do n ot n eed to preserve m ore than on e observation at the level of paren t-affiliate in dustry for a given year. Ign orin g affiliates with estim ated data, this step created a population of 6650 paren t-affiliate-in dustry observation s from a poten tial population of 68 59 affiliates. Thus, very few observation s were aggregated in this step.

5 Usable im plies that em ploym en t data is n ot m issin g.

6 Note that the size of the com parison data set depen ds on the variables bein g com pared.

There are m ore usable em ploym en t data than wage data, so the com parison (sam e in dustry, sam e MNC) Can adian an d n on -Can adian affiliate em ploym en t data set con tain s approxim ately 11 0 0 0 affiliate year observation s while the com parison affiliate data set for wages contain s 630 0 affiliate-year observations.

7 Note that large, diversified paren ts are overrepresen ted in the affiliate-level data, sin ce these paren ts ten d to have several affiliates (represen tin g differen t busin ess un its) in a given coun try each year.

8 Further detail on the con struction of these charts is available from the authors.

9 Con structin g these charts required at least two con secutive observation s per affiliate-paren t pair, an d 198 3 data were used on ly to con struct em ploym en t at (t-1).

10 The three-digit SIC code level adm ittedly is a very rough m easure of “activity” in the sen se of lookin g for offshorin g of certain types of jobs, such as clerical jobs. H owever, we kn ow of n o large-sam ple pan el data on m ultin ation al firm s that con tain m ore disaggregated an n ual data on job type within the firm . In its Ben chm ark Surveys, the BEA collects m ore detailed data on job type, but these surveys are adm in istered on ly every five years.

11 Despite this expan sion , the m edian paren t had foreign operation s in on ly two coun tries in 20 0 3. Again , these sm all paren ts are un derrepresen ted in the affiliate-level data.

(23)

B

IB LIOGRAP H Y

Baldwin , J ohn R. an d Wulong Gu. 20 0 5. “Global Links: Multin ation als, Foreign Ownership and Productivity Growth in Can adian Manufacturing.” The Canadian Econ om y in Transition Research Paper Series, 11-622-MIE, Statistics Canada.

Feinberg, S., and M. Keane. 20 0 1. “U.S.-Canada Trade Liberalization and MNC Production Location.” R ev iew of Econ om ics an d Statistics 8 3(1): 118 -132.

———. 20 0 6. “Accoun tin g for the Growth of MNC-based Trade Usin g a Structural Model of U.S. MNCs.” Am erican Econ om ic R ev iew 96(5): 1515-1558 .

Public Policy Forum and ITAC Roundtable. 20 0 4. “IT Offshore Outsourcing Practices In Canada.” Public Policy Forum Workin g Paper.

Scott, R., D. Ticoll and T. Garner. 20 0 4. A Fin e Balan ce: The Im pact of Offshore IT Serv ices on Can ada’s IT Lan dscape. Toron to:

PricewaterhouseCoopers.

(24)

Parent industry wage Canadian wage PPP Exchange Rate

A

P P EN D IX

FIGU RE A-1

MED IAN REAL EMP LOYEE COMP EN S ATION/ EMP LOYMEN T, SAME-IN D U S TRY CAN AD IAN AFFILIATES AN D U .S .PAREN TS, SERVICE IN D U S TRIES

20 24 28 32 36 40 44

83-85 86-88 89-91 92-94 95-97 98-00 01-03

(000's 1983 US $)

0.90 1.00 1.10 1.20 1.30 1.40 1.50

PPP Exchange Rate

(25)

Parent industry wage Canadian wage PPP Exchange Rate

FIGU RE A-2

MED IAN REAL EMP LOYEE COMP EN S ATION/ EMP LOYMEN T SAME- IN D U S TRY CAN AD IAN AFFILIATES AN D U .S .PAREN TS,

MAN U FACTU RIN G IN D U S TRIES

20 22 24 26 28 30 32 34

83-85 86-88 89-91 92-94 95-97 98-00 01-03

(000's 1983 US $)

0.90 1.00 1.10 1.20 1.30 1.40 1.50

PPP Exchange Rate

(26)

FIGU RE A-3

PERCEN TAGE EMP LOYMEN T GROW TH IN CAN AD IAN AFFILIATES B Y

DECILES OF SAME-IN D U S TRY U .S .PAREN T EMP LOYMEN T GROW TH, 19 8 3 – 2 0 0 3 ,SERVICE IN D U S TRIES

-27.0%

0-10 -10.6%

10-20 -4.7%

20-30

-1.8%

30-40 1.0%

40-50 3.9%

50-60 5.2%

60-70

9.4%

70-80 16.6%

80-90 31.8%

90-100

-0.03 -0.02 -0.01 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07

(27)

FIGU RE A-4

PERCEN TAGE EMP LOYMEN T GROW TH IN U .SMN CPAREN TS B Y

DECILES OF SAME-IN D U S TRY CAN AD IAN AFFILIATE EMP LOYMEN T

GROW TH,19 8 3 – 2 0 0 3 ,SERVICE IN D U S TRIES

-32.0%

0-10 -14.6%

10-20 -7.5%

20-30 -3.6%

30-40 0.6%

40-50

3.2%

50-60 6.1%

60-70 10.0%

70-80

17.6%

80-90 36.8%

90-100

-0.02 -0.01 0 0.01 0.02 0.03 0.04 0.05 0.06

(28)

FIGU RE A-5

PERCEN TAGE EMP LOYMEN T GROW TH IN CAN AD IAN AFFILIATES B Y

DECILES OF SAME-IN D U S TRY U .S .PAREN T EMP LOYMEN T GROW TH, 19 8 3 – 2 0 0 3 ,MAN U FACTU RIN G

-26.2%

0-10 -10.8%

10-20 -5.6%

20-30 -2.5%

30-40 -1.0%

40-50 0.7%

50-60

2.9%

60-70 5.8%

70-80 10.7%

80-90 25.8%

90-100

-0.05 -0.04 -0.03 -0.02 -0.01 0 0.01 0.02

(29)

Change in Canadian employment Change in non-Canadian employment

FIGU RE A-6

MED IAN PERCEN TAGE AN N U AL CH AN GE IN EMP LOYMEN T,CAN AD IAN AN D NON-CAN AD IAN ( SAME IN D U S TRY,SAME-MN C) AFFILIATES, 19 8 3 – 2 0 0 3 ,ALL IN D U S TRIES

83-85

86-88

89-91

92-94

95-97

98-00

01-03

-5.0%

-4.0%

-3.0%

-2.0%

-1.0%

0.0%

1.0%

2.0%

3.0%

4.0%

(30)

Change in Canadian employment Change in non-Canadian employment

FIGU RE A-7

MED IAN PERCEN TAGE AN N U AL CH AN GE IN EMP LOYMEN T,CAN AD IAN AN D NON-CAN AD IAN ( SAME IN D U S TRY,SAME-MN C) AFFILIATES, 19 8 3 – 2 0 0 3 ,MAN U FACTU RIN G IN D U S TRIES

83-85

86-88

89-91 92-94

95-97

98-00

01-03

-6.0%

-5.0%

-4.0%

-3.0%

-2.0%

-1.0%

0.0%

1.0%

2.0%

3.0%

4.0%

(31)

Change in Canadian employment Change in non-Canadian employment

FIGU RE A-8

MED IAN PERCEN TAGE AN N U AL CH AN GE IN EMP LOYMEN T,CAN AD IAN AN D NON-CAN AD IAN ( SAME IN D U S TRY,SAME-MN C) AFFILIATES, 19 8 3 – 2 0 0 3 ,SERVICE IN D U S TRIES

83-85

86-88

89-91

92-94

95-97

98-00

01-03 -3.0%

-1.0%

1.0%

3.0%

5.0%

7.0%

9.0%

Referenzen

ÄHNLICHE DOKUMENTE

2 In particular we do not allow that all voters cast abstain/negative votes for all candidates. With this requirement we avoid stating that all candidates must be elected in case

Since the objective of the study is to analyze the impact of the real exchange rate on the non-oil exports in the framework of cointegration and a possible asymmetric adjustment

( for the real exchange rate is selected by the AIC.).. For example,  −1  means the expected future sum of short-term real interest differentials,

Real effective exchange rate, Real oil price, Relative productivity, Azerbaijani manat, Dutch Disease, Oil-exporting Countries, Johansen Co-integration Approach, Error

The MR-LSTAR model allows for both discrete and smooth mean reversion like the BP and SGE transition functions, but it has the same drawbacks when compared with using the SGE

In this study, strong empirical support is provided indicating a bi-directional relationship between the ex ante real interest rate yield on S&P high grade long term

basic, APL, fortran, and SPL, experience with digital electr. Need: mass memory such as floppy disk and a hard-copy device. Designing automated Measurement system

Three main stylized facts emerge from the experiences of Brazil, Chile and Colombia: first, temporary shocks, including (but not necessarily limited to) the type of policy