accounting for technology, trade and final consumption in employment : an input-output decomposition

Accounting for technology, trade and finalconsumption in employment: an Input-Output

decomposition

Mathilde PAK1 Aurélien POISSONNIER2

1INSEE, Crest, Paris-Dauphine University

2INSEE, Crest, École polytechnique

24th IIOA Conference

Pak, Poissonnier (INSEE) Employment, tech., trade & consumption 7th July 2016 1 / 25

1 Motivation

2 Data

3 Method

4 Main results

5 Conclusion


Plan

1 Motivation

2 Data

3 Method

4 Main results

5 Conclusion


Stylized facts • ◦ ◦

An increasing share of skilled workers in France since 1982


Stylized facts • • ◦

In a context of labour productivity gains, increasing openness...


Stylized facts • • •

... and preferences of consummers for services


Literature

Main determinants of employment changesI Technology: - in the short-run but + in the long-runI External trade: - through imports and + through exportsI Final consumption: +

Distinguished effects by skill level?I Technology: routinization hypothesis (Autor, Levy and Murnane,

2003; Goos and Manning, 2007)I External trade: comparative advantages, jobs offshorability

(Blinder, 2009)I Final consumption: Engel curves (Autor and Dorn, 2013; Goos and

Manning, 2007)


LiteratureMain determinants of employment changes

I Technology: - in the short-run but + in the long-runI External trade: - through imports and + through exportsI Final consumption: +

Distinguished effects by skill level?I Technology: routinization hypothesis (Autor, Levy and Murnane,

2003; Goos and Manning, 2007)I External trade: comparative advantages, jobs offshorability

(Blinder, 2009)I Final consumption: Engel curves (Autor and Dorn, 2013; Goos and

Manning, 2007)

How about accounting for these 3 channels altogether?I IO analysis!I Previous works on UK (Gregory, Zissimos and Greenhalgh, 2001)

and a set of advanced and emerging economies (Los, Timmer andde Vries, 2014)


Plan

1 Motivation

2 Data

3 Method

4 Main results

5 Conclusion


Data

Input-Output data (INSEE national accounts)I Availability: 1980-2010, for 48 products, in current pricesI Work on these data: conversion in previous year prices (using

specific deflator for consumption, investment...) and using ESA2010 concepts from 2010 up to 1980

French labour Force Survey (INSEE)I Employment by skill and sector: from 1982 to 2010 (in headcounts)I Corrections brought to data: treat for changes in occupation and

sector classifications (NAP-NAF-NAF Rev. 1-NAF Rev. 2 & PCS);convert sectors to concept of industry and then product; treat forbreaks in collection process


Data

Input-Output data (INSEE national accounts)I Availability: 1980-2010, for 48 products, in current pricesI Work on these data: conversion in previous year prices (using

specific deflator for consumption, investment...) and using ESA2010 concepts from 2010 up to 1980

French labour Force Survey (INSEE)I Employment by skill and sector: from 1982 to 2010 (in headcounts)I Corrections brought to data: treat for changes in occupation and

sector classifications (NAP-NAF-NAF Rev. 1-NAF Rev. 2 & PCS);convert sectors to concept of industry and then product; treat forbreaks in collection process

⇒ Final database: 1982-2010; 38 products, 9 skill groups


Plan

1 Motivation

2 Data

3 Method

4 Main results

5 Conclusion


Input-Output Analysis FrameworkFrom production required to address final demand...

P + M = IC + FD ⇒{

P = ICd + FDd

M = ICm + FDm

Technological coefficients IC = APDomestic share, ex: FDDd = SFDFD

... to labour content of final demand

P = (I− SICA)−1(SFDFD) = R(SFDFD)

VA = MP

N = TMR(SFDFD)

M = diag((I− tA)1)T: matrix of skill-use coefficients

TMRSFD is the labour content of FDPak, Poissonnier (INSEE) Employment, tech., trade & consumption 7th July 2016 11 / 25

Employment changes decomposition • ◦ ◦

From the labour content: N = TM(A)R(SIC, A)SFDFDPossible to decompose ∆N into

effect of ∆FCeffect of ∆(SXX)

effect of ∆SFC, ∆SGFCF, ∆SIC

effect of ∆Teffect of ∆A (hidden in matrix R and M)effect of ∆GFCF


Employment changes decomposition • ◦ ◦

From the labour content: N = TM(A)R(SIC, A)SFDFDPossible to decompose ∆N into

effect of ∆FC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . final consumptioneffect of ∆(SXX) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . exportseffect of ∆SFC, ∆SGFCF, ∆SIC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . importseffect of ∆T . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . technologyeffect of ∆A (hidden in matrix R and M) . . . . . . . . . . . . . technologyeffect of ∆GFCF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . technology

Is GFCF demand or technology ?Changes in GFCF = changes in future production factorsConsidered as technology


Employment changes decomposition • • ◦

∆Nt,τ = TtMtRtSFCτ ∆FC︸︷︷︸

Changes in Final Domestic Consumption

+TtMtRt(∆SFDDFDDt + ∆(SXX)) + TtMtRt∆SICAtPτ︸︷︷︸Changes in exports and imports

+TtMtRtSICτ ∆APτ + TtMtRtSGFCF

τ ∆GFCF + Tt∆MPτ + ∆TMτPτ︸︷︷︸Changes in technology

NB: ∆N = vector of employment change for each skill level in eachproductionTreatment of the issue of the n! decompositions: average all potentialdecompositions


Employment changes decomposition • • •

Limits related to this kind of accounting decomposition:We capture only first round / short-term effectsWe cannot reveal underlying causal links between employmentand its determinants in the long-run

⇒ Results to be interpreted as short-term effects⇒ We further decompose final consumption to distinguish an incomeeffect (see positive long-term effect of technology)


Plan

1 Motivation

2 Data

3 Method

4 Main results

5 Conclusion


Contributions to employment changes by skill level• ◦ ◦◦

Average contributionTotal

Skill level

HighMiddle

Low Other(in % per year) higher lowerJobs creation 0.6 3.4 1.4 0.4 0.1 -1.7CTB-Final consumption 1.2 1.3 1.3 1.2 1.3 0.9CTB-Trade 0.3 0.3 0.3 0.4 0.3 0.4CTB-Technology -0.9 1.7 -0.3 -1.2 -1.5 -3


Contributions to employment changes by skill level• • ◦◦

Breakdown of final consumption effects to employment changes byskill level


Skill level

HighMiddle

Low Other(in % per year) higher lowerFinal consumption effects 1.2 1.3 1.3 1.2 1.3 0.9Consumption structure -0.1 0.1 0 -0.1 -0.2 -0.3Purchasing power 0.4 0.4 0.4 0.4 0.5 0.6Sociodemographic effects 0.3 0.2 0.2 0.3 0.3 0.4Household saving 0 0 0 0 0 0.1Gov. and NPISHconsumption

0.5 0.6 0.7 0.5 0.6 0.2


CTB to employment changes by skill level • • •◦

Trade effect=TMR(∆(SXX) + ∆SFDDFDD) + TMR(

∆SICA)

P


Skill level

HighMiddle

Low Other(in % per year) higher lowerTrade effects 0.3 0.3 0.3 0.4 0.3 0.4CTB-Exports 0.7 0.7 0.7 0.8 0.7 0.8CTB-Offshore outsourcing -0.3 -0.2 -0.2 -0.3 -0.3 -0.3CTB-Imports (FDD) -0.1 -0.1 -0.1 -0.2 -0.1 -0.1


CTB to employment changes by skill level • • ••

Technologyeffect=TMRSIC∆AP + TMRSGFCF∆GFCF + T∆MP + ∆TMP


Skill level

HighMiddle

Low Other(in % per year) higher lowerTechnology effects -0.9 1.7 -0.3 -1.2 -1.5 -3CTB-Direct labour saving -1.2 1.4 -0.4 -1.5 -1.5 -3.2CTB-IC effects -0.1 -0.1 -0.1 0 -0.2 0CTB-GFCF effects 0.3 0.5 0.3 0.3 0.2 0.2


CTB to employment changes by skill and product


Plan

1 Motivation

2 Data

3 Method

4 Main results

5 Conclusion


Conclusion

What are the effects of trade, technology and final consumption onemployment? How can we explain skill-bias changes in Frenchemployment?

Strutural decomposition analysis using IO dataMain skill-bias determinants: technologyImportant contribution of final consumption explained by thedevelopment of servicesTrade effect: small, but nonetheless positiven regardless of theskill level


Appendix


Product aggregation

Aggregated sector DescriptionAverage contribution

(in % per year)FC Exports Imports Tech.

Manuf. High Tech.

CE - Chemicals 0.4 2.8 -2.1 -3.2CF - Pharmaceuticals 3.3 4.0 -2.5 -4.5C3- Electrical equip. 2.4 8.2 -6.1 -9.8CL - Transport equip. -0.1 2.2 -1.2 -3.3

Manuf. Low Tech.

C1 - Food & drink 0.9 0.7 -0.5 -1.5CB - Textile & leather -0.3 0.7 -4 -2.5CC - Wood & paper 0.8 1.2 -0.8 -2.8C2 - Coke & refined petroleum 0.7 1.0 -1.2 -2.8CG - Rubber & plastic 0.5 1.6 -1.3 -2.2CH - Metals 0 1.4 -1.3 -1.8CM - Other manuf. 0.1 1.4 -0.9 -1.8

Serv. non Tradable

FZ - Construction 0.2 0.1 0.0 -0.2IZ - Accomodation & food serv. 1.2 0.2 -0.1 0.6KZ - Finance 2.2 0.5 -0.1 -2.1LZ - Real estate 2.0 0.2 -0.1 -0.7OQ - Public adm. 7.1 0.1 -0.1 -1.7RU - Other serv. 6.0 0.7 -0.2 2.2

Serv. Tradable

GZ - Trade 1.4 1.0 -0.2 -1.5HZ - Transportation 1.1 1.1 -0.4 -1.2JZ - Info. & comm. 6.4 1.9 -0.8 -2.3MN - Business serv. 3.5 3.7 -1.7 3.8

OtherAZ - Agriculture 0.9 0.9 -0.6 -4.3DE - Energy & utilities 3.3 2.7 -3.7 -4.7


Skill levels based on occupation classification

"high-skill" = Managers (includes liberal professions)"intermediate-skill (higher)" = Intermediate occupations(professionals and technicians)"intermediate-skill (lower)" = Skilled service and sales workers;Skilled machine operators and elementary occupations"low-skill" = Unskilled service and sales workers; Unskilledmachine operators and elementary occupations;"others" = Farmers; Craft and related trades workers and chiefexecutives; Other (Military contingents, unknown)


accounting for technology, trade and final consumption in employment : an input-output decomposition

Economy & Finance