Div

isio

no

f la

bo

ur

Sept. 2010

Tasks, technology and trade

Measuring and interpreting trends in the division of labour in the Netherlands

Ì. Semih AkçomakLex BorghansBas ter Weel

Div

isio

no

f la

bo

ur

Sept. 2010

Observations

� Division of labour seems to have changed …

► Blinder: “>25 percent of all jobs offshorable”

► Grossman and Rossi-Hansberg: “trade in tasks”

► Baldwin: “second unbundling”

► Policy makers: “what is the threat of offshoring?”

Div

isio

no

f la

bo

ur

Sept. 2010

Aim

� Measuring and understanding trends in the division of labour

► How does “(un)bundling” work?

► What are the employment effects?

► Who is affected? Is my job safe?

Div

isio

no

f la

bo

ur

Sept. 2010

Approach

� Develop measures to understand changes in division of labour

� Three dimensions and issues

Dimension IssuesOccupation Separate tasks into different jobs

Industry Outsource tasks to other firms

Space Offshore tasks

Div

isio

no

f la

bo

ur

Sept. 2010

Division of labour and tasks

€Y

€Z€X

1-c

1-c1-c

Task 1 Task 2

Task 3

Occupation

€X=€Y=€Z

c≤1

Initial situation

Price (€)Connectivity (c)

Div

isio

no

f la

bo

ur

Sept. 2010

Division of labour and tasks

€Y

€Z€X

1-c

1-c1-c

Task 1 Task 2

Task 3

Occupation

€X=€Y<>€Z0≤c≤1

Price of tasks changes

Div

isio

no

f la

bo

ur

Sept. 2010

Division of labour and tasks

€Y

€Z€X

1-d

1-d1-c

Task 1 Task 2

Task 3

Occupation

€X=€Y=€Z

d<c

Connectedness changes

Div

isio

no

f la

bo

ur

Sept. 2010

Framework

task 1

occupation

Separate? Trade-off Measures

task 2 task 3 tasks…… task n

Production

• task-occupation

connectivity

• task-occupation

wage differential

Coordination

costs vs.

production

costs

Div

isio

no

f la

bo

ur

Sept. 2010

Framework

task 1

occupation

Separate?

firm

Outsource?

Trade-off Measures

• task-industry

connectivity

task 2 task 3 tasks…… task n

task 1 task 2 task 3 tasks…… task n

occupations

Production

• task-occupation

connectivity

• task-occupation

wage differential

Coordination

costs vs.

production

costs

Make vs. buy

Div

isio

no

f la

bo

ur

Sept. 2010

Framework

task 1

occupation

Separate?

firm

Outsource?

country

Offshore?

Trade-off Measures

• task-industry

connectivity

• Occupation-space

connectivity

• Spatial-occupation

concentration

task 2 task 3 tasks…… task n

task 1 task 2 task 3 tasks…… task n

occupations

task 1 task 2 task 3 tasks…… task n

occupations

Production

• task-occupation

connectivity

• task-occupation

wage differential

Coordination

costs vs.

production

costs

Make vs. buy

Proximity vs.

cost advantages

Div

isio

no

f la

bo

ur

Sept. 2010

Data

� Goal► Data set 1996-2005 with education, wages, task

measures, specialisation and offshoring

� Input► Dutch labour force survey 1996-2005

► Tasks from UK (BSS)

► IO industry classification

► Spatial employment information

� Result► occupation-task-year

► industry-task-year

► occupation-space-year

Div

isio

no

f la

bo

ur

Sept. 2010

Task-occupation connectivity

Individial level task dataimportance importance importance importance

person task 1 task 2 …………….. task n

1 0 3 …………….. 4

2 1 1 …………….. 4

3 3 4 …………….. 3

4 4 4 …………….. 2

5 4 2 …………….. 1

6 4 2 …………….. 2

7 3 1 …………….. 1

8 3 3 …………….. 2

9 2 3 …………….. 3

10 2 4 …………….. 4

.

.

.

.n

Obtain correlation

matrix of tasks (n*n

matrix) c

t1 t2 t3 . . tn

t1

t2

t3

.

.

t n

Use the importance

level of each task for

each occupation as

weights (m)

Calculate c*m for each occupation-task

combination (multiply rows by task

importance and sum by occ

Div

isio

no

f la

bo

ur

Sept. 2010

Data set

occupation task

task

importance

1997

task

importance

2006

task-occupation connectivity

basic jobs strength 3,53 3,63 0,065114111 analysing 2,04 2,16 0,291316711 planning 1,79 2,06 0,233993411 writing 1,86 1,88 0,2662569… … … … …… … … … …technical (low) 8 3,32 3,54 0,065729826 18 2,57 3,02 0,297003226 22 2,20 2,50 0,233195426 30 2,02 2,20 0,2673084… … … … …… … … … …econ, admin etc. 8 1,94 2,07 0,041292751 18 3,03 3,46 0,312353951 22 2,61 2,85 0,251124351 30 2,53 3,06 0,2917716… … … … …… … … … …managers 8 2,32 2,03 0,041196798 18 3,77 4,14 0,315472398 22 4,11 3,94 0,252825598 30 3,13 3,36 0,2943714… …… …

Div

isio

no

f la

bo

ur

Sept. 2010

Workers

task 1

occupation

Separate? Trade-off Measure

task 2 task 3 tasks…… task n

Production

• task-occupation

connectivity

• task-occupation

wage differential

Coordination

vs. production

Task-occupation connectivityPresence of task X given presence task Y

Task-occupation wage differentialAbsolute difference between occupation and task wage

ij i jtowd w w= −

33

1

j

ij j j i j

j

toc c m=

=

= ∑

Div

isio

no

f la

bo

ur

Sept. 2010

Connectivity and wage differential

detailpeople

teach

cspeech

persuad

selling

caring

strengtstamina

hands

tools

product

special

orgworkfaults

cause

solutn analyse

noerrormistake

planme

planothmytime

ahead

read

short

long

write

writeshwritelg

calca

percent

stats

-3-2

-10

1

task-o

ccupation c

onnectivity

-1 0 1 2 3 4

task-occupation wage differential

Div

isio

no

f la

bo

ur

Sept. 2010

Contestability

76

62

5696

51

93

75

53

33

6544

48

71

69

42

6485

66

91

73

49

4686

55

26

31

57

28

29

78

8998

24

11 37

68

0

% c

ha

ng

e in

em

plo

ym

en

t 1

99

6-2

00

5

-2 -1 0 1 2 3

contestability

bandwidth = .8

sdf Elementary and low-skilled service jobs

Div

isio

no

f la

bo

ur

Sept. 2010

(1) (2) (3) (4)

Dependent variable: Change in

employment, 1996-2005

OLS OLS WLS WLS

task-occupation connectivity 0.075** 0.153*** 0.058** 0.075**

(0.030) (0.048) (0.023) (0.032)

task-occupation wage differential 0.003 0.032 0.067** 0.083***

(0.024) (0.026) (0.026) (0.028)

education 1996 -0.084 -0.011

(0.059) (0.044)

log employment 1996 0.029 0.031

(0.027) (0.022)

Constant 0.012 -0.173 0.053** -0.183

(0.029) (0.189) (0.021) (0.176)

Observations 36 36 36 36

R-squared 0.168 0.273 0.329 0.370

Employment changes

Div

isio

no

f la

bo

ur

Sept. 2010

Employment changes

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0.4

0.5

11 26 28 24 29 37 31 44 33 46 48 42 53 51 57 49 55 56 78 68 71 66 64 73 75 69 62 76 98 65 86 91 85 93 96 89

sta

ndard

ized v

alu

es

prediction employment

prediction employment

Div

isio

no

f la

bo

ur

Sept. 2010

Employment changes

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0.4

0.5

11 26 28 24 29 37 31 44 33 46 48 42 53 51 57 49 55 56 78 68 71 66 64 73 75 69 62 76 98 65 86 91 85 93 96 89

sta

ndard

ized v

alu

es

prediction employment

prediction employment

Engineers and scientists

Elementary and low-skilled service jobs

Teachers

Div

isio

no

f la

bo

ur

Sept. 2010

Industry

task 1

firm

Outsource? Trade-off Measure

task 2 task 3 tasks…… task n

Production

• task-industry

connectivityMake vs. buy

Task-industry connectivityPresence of task X given presence task Y

33

1

j

jr jj jr

j

tic c m=

=

= ∑

Div

isio

no

f la

bo

ur

Sept. 2010

Connectivity: occ vs. ind

detailpeople

teachcspeech

persuad

selling

caring

strengtstamina

handstools

product

special

orgworkfaults

cause

solutn analyse

noerrormistakeplanme

planothmytime

ahead

read

shortlong

write

writeshwritelg

calca

percentstats

0.1

.2.3

.4.5

task

-occupati

on

con

nect

ivit

y

0 .1 .2 .3 .4 .5

task-industry connectivity

Tasks are more connected to thefirm than to the occupation

Div

isio

no

f la

bo

ur

Sept. 2010

Employment changes

0.2480.2460.2590.208R-squared

36363636Observations

(0.152)(0.026)(0.163)(0.028)

0.0140.031-0.2120.012constant

(0.019)(0.024)

0.0030.034log employment 1996

(0.053)(0.043)

0.011-0.006education 1996

(0.033)(0.021)(0.032)(0.022)

0.074**0.081***0.102***0.084***task-industry connectivity

WLSWLSOLSOLSDependent variable: Change in

employment, 1996-2005

(4)(3)(2)(1)

Div

isio

no

f la

bo

ur

Sept. 2010

Employment changes

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0.4

0.5

11 26 28 24 29 37 31 44 33 46 48 42 53 51 57 49 55 56 78 68 71 66 64 73 75 69 62 76 98 65 86 91 85 93 96 89

sta

ndard

ized v

alu

es

prediction employment

prediction employment

Div

isio

no

f la

bo

ur

Sept. 2010

Space

task 1

country

Offshore? Trade-off Measure

task 2 task 3 tasks…… task n

Production

Proximity vs.

cost advantages

• Occupation-space

connectivity

• Spatial occupation

concentration

Occupation-space connectivityPresence of occ X given presence occ Y

Spatial occupation concentrationTradability of jobs: difference between the

occupation and population share

36

1

i

i ii i

i

osc c m=

=

=∑

17

1

v

v v v

v

soc occ ps=

=

= −∑

Div

isio

no

f la

bo

ur

Sept. 2010

Concentration and connectivity

11

24

26

2829

31

3337 42

44

46

48

49

51

53

55

56

57

62 64

65

66

68

69

71

73

75

76

78

82

85

86

89

9193

96

98

.05

.1.1

5.2

.25

.3

job c

onnectivity

.2 .3 .4 .5 .6 .7job concentration

Engineers and scientists

LS service and teachers

Div

isio

no

f la

bo

ur

Sept. 2010

Contestability

76

93

91

71

89

75

28

26

46

31

5162

86

98

49

96

78

11

4424

55

57

29

69

37

53

56

64

6633

48

42

7368

6585

0

% c

han

ge

in e

mp

loy

men

t 1

99

6-2

005

-2 -1 0 1 2

spatial contestability

bandwidth = .8

sdf

Engineers and scientists

Teachers

Div

isio

no

f la

bo

ur

Sept. 2010

(1) (2) (3) (4)

Dependent variable:

Changes in employment,

1996-2005

OLS WLS OLS WLS

job concentration -0.142** -0.135**

(0.069) (0.058)

job connectivity 0.039 0.029

(0.034) (0.029)

education 1996 0.102*** 0.133*** 0.118*** 0.154***

(0.037) (0.031) (0.034) (0.034)

log employment 1996 -0.049 -0.017 0.020 0.025

(0.043) (0.027) (0.025) (0.019)

log wage 1996 -0.326*** -0.285*** -0.326*** -0.333***

(0.113) (0.091) (0.102) (0.093)

constant 3.078*** 2.276*** 2.373** 2.390***

(0.966) (0.720) (0.880) (0.717)

Observations 36 36 36 36

R-squared 0.364 0.472 0.270 0.411

Employment changes

Div

isio

no

f la

bo

ur

Sept. 2010

Employment changes

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0.4

0.5

11 26 28 24 29 37 31 44 33 46 48 42 53 51 57 49 55 56 78 68 71 66 64 73 75 69 62 76 98 65 86 91 85 93 96 89sta

ndard

ized v

alu

es

prediction employment

prediction employment

Div

isio

no

f la

bo

ur

Sept. 2010

Offshoring

� Relative amount of all imported intermediate output

70

1

j

i j j i

j

offshoring imp c=

=

= ∑

cji is an element of the Leontieff inverse matrix (I-A)-1

A is a 70×70 matrix with input-output information about

Intermediate purchases

Div

isio

no

f la

bo

ur

Sept. 2010

Offshoring and employment

-2

-1.5

-1

-0.5

0

0.5

1

1.5

2

2.5

3

62 29 56 53 69 49 93 73 57 89 96 42 37 33 76 98 31 91 28 78 55 51 64 71 11 75 85 24 65 48 44 66 86 68 26 46

sta

nd

ard

ize

d v

alu

es

total hours worked employment

total work hour employment

Div

isio

no

f la

bo

ur

Sept. 2010

Offshoring and employment

0.3170.0760.089R-squared

363636Observations

(0.013)

-0.021occupation-space connectivity

(0.023)

0.060**spatial occupation concentration

(0.014)

0.018task-industry connectivity

(0.012)

0.016task-occupation connectivity

(0.001)

0.001task-occupation wage differential

OLSOLSOLSDependent variable: Changes in

offshoring, 1996-2005

(3)(2)(1)

Div

isio

no

f la

bo

ur

Sept. 2010

Will my job be offshored?

� It depends …

► How well connected are the tasks within your job?

► How large is the price differential in your job?

► When tasks are separated from your job, to what extent will they be outsourced by the firm?

► How well connected is your job to other jobs?

► How concentrated is your job in space?

Div

isio

no

f la

bo

ur

Sept. 2010

Will my job be offshored?

� High school teacher – no► Worker – no

► Firm – no

► Space – no

� Low-skilled (service) jobs – yes, but no► Worker – yes

► Firm – yes

► Space – no

� High skilled technical jobs – yes ► Worker – yes

► Firm – ?

► Space – yes

Div

isio

no

f la

bo

ur

Sept. 2010

Conclusion

� Framework

� Empirical measures for division of labour

� Data

� Policy