Exploring the French productivity puzzle   

Philippe Askenazy (CNRS-PSE, Cepremap and IZA)Christine Erhel (University Paris 1, CES, CEE)

With the contribution of Martin Chevalier (ENS Cachan)

CEPREMAP Productivity project, Paris conference, 23rd of January 2015

 

Introduction

• Labour  productivity  relatively high in  France,  and  quite dynamic until 2007 (OECD data). 

• A reversal in 2007-2009,  followed  by  a  limited  increase  in 2010-12  and  flatness  in  2013-14.  Large  contrast  with previous recessions.

Does  it  result  from  short-term factors or does  it  indicate a change in the French productive model (high job protection, labour costs, high work intensity and productivity)? Changes in  labour market  (e.g.  labour  supply  composition,    reforms altering labour law, pensions)?... 

Introduction

• Structure of the chapter/presentation1. The French productivity puzzle2. A new labour market (A) and supported firms (B)3. Quantitative micro-analyses using establishment/firm data

1-The French productivity puzzle Figure 1: Quarterly GDP, employment and working time indexes.

1990-1997 and 2006-2013. Base 1 = Q1.1992 (t=0) or Q1.2008 (t=0) . INSEE NA

-8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 220.940000000000001

0.960000000000001

0.980000000000001

1

1.02

1.04

1.06

1.08

1.1

Value-Added 1992 recession

Value-Added 2008 recession

Employment 1992 recession

Employment 2008 recession

Working time 1992 recession

Working time 2008 recession

1-The French productivity puzzle• In comparison to 1992-1993: huge differences in GDP trends 

but  similarity  in  (un)employment  trends  and  flatness  in working hours => under-adjustment  in  (un)employment and slowdown  in  productivity  per  head  during  the  2007-2009 crisis

• Facing this puzzle, the most straightforward explanations do not seem significant

The  drop  in  productivity,  that  concerns  only  the  market economy,  cannot be explained by particular industries

  A  slowdown  in  real wage  growth,  but  no  adjustments  that could explain employment (and unemployment) trends

A  stability  of  investment,  and  of  R&D  spending  during  the crisis

2a-A new labour market…

• Changes in the composition of the workforce are large

Increase  in  seniors’ employment rate following  pensions reforms: effect on productivity is unclear

A  continuous decrease  in employment rates for  the middle educated,  whereas  they  increased  for  the  higher levels of education.  An  acceleration  of  the  shift  of  labour  demand from  the  low  or  middle  educated  to  the  most  educated during the recent recession

Could explain up to half of productivity slowdown over the last years. Why?

2a-A new labour marketEmployment for top salaries occupations and top education level, 1000 workers,

1990-2012 (source: INSEE)

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

0.0

1,000.0

2,000.0

3,000.0

4,000.0

5,000.0

6,000.0

Upper tertiaryManagers and professionals

• Top  occupations  or  educated  doubled  their  weight  in  total workforce 

• Employment  growth  for  top  occupations  and  upper-tertiary educated is a-cyclical like long-term investments 

• Generate apparent pro-cyclical productivity

2a-A new labour market  Low and middle educated face still business cycle, while no correlation between the

evolution of high-educated workforce and GDP growth

1990

1992

1994

1996

1998

2000

2002

2004

2006

2008

2010

2012

-4

-3

-2

-1

0

1

2

3

4

5

g secondary or below Gdp growth

1990

1992

1994

1996

1998

2000

2002

2004

2006

2008

2010

2012

-4

-2

0

2

4

6

8

10

g high-educated gdp growth

2a-A new labour market

• A  rise  of  low  productivity  jobs  favoured  by  some  labour market reforms:

Increase of very short term contracts since 2004. Continued after the introduction of an extra social contribution in 2013. Could contribute to productivity slowdown.

New self-employed status in  2009: most  of  them have  low revenues  and  are  less  productive  than  « classic »  self-employment  who  was  sharply  declining.  Rough  estimation shows it could explain one fifth of productivity slowdown.

2a-A new labour market Non-salaried employment, total hours worked by the self-employed, and mixed income of

unincorporated enterprises, 2003-2013. Volume base 1 = 2007

2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 20130.850000000000001

0.900000000000001

0.950000000000001

1

1.05

1.1

Mixed incomeHoursJobs

Source: Authors' computations using National Accounts, base 2010. Mixed income of non-corporate business is deflated by the price index for the total value added.

The vertical line dates the creation of the auto-entrepreneur status.

2b-..and supported firms

• No  clear  evidence  of  credit  rationing  or  change  in  tangible capital allocation in France.

• Financial  situation  of  firms  has  been  sustained  by  various policies.

>>>  a  global  analysis  shows  that  changes in labour force composition and labour market regulation may  be  the  most important drivers of the productivity slowdown. 

But  the  links  micro  determinants  of  productivity  (e.g.  work practices) and  their  evolutions  also  have  to  be  taken  into account, using micro-data.

2a-A new labour market and supported firms Corporate loan stocks and loans drawn by SMEs

Millions of current Euros. 2006-2014

2006

Jan

2006

May

2006

Sep

2007

Jan

2007

May

2007

Sep

2008

Jan

2008

May

2008

Sep

2009

Jan

2009

May

2009

Sep

2010

Jan

2010

May

2010

Sep

2011

Jan

2011

May

2011

Sep

2012

Jan

2012

May

2012

Sep

2013

Jan

2013

May

2013

Sep

2014

Jan0

20000

40000

60000

80000

100000

120000

140000

160000

180000

SMEs more than three years old

SMEs less than three years old

Source: Bank lending survey. Banque de France.

3-Micro-analyses using REPONSE

• Databases and methodology A  merged  dataset:  REPONSE  (2005  and  2011,  panel), 

composition  by  occupations,  job  creations/destrubctions (DMMO-EMMO), firm accounts

Analysis is based on total factor productivity estimationsLn(labour productivity)  =  α.ln  (Capital intensity)  +  λ.Workforce composition + ξ.HR-Practices + µ.controls + ε  Cross-section, panel, IV

A  nested  logit  deals  with  the more  specific  issue  of  skilled labour hoarding in  firms  where  employment  has  been declining (using REPONSE 2011)

3-Micro-analyses using REPONSE• Results (1): labour force composition and productivity No significant effect of seniors’ share on  productivity  in  2011,  and  no 

statistical difference in coefficients between before and after reforms

According  to REPONSE 2011,  skilled occupations have been maintained in  a majority  of  establishments  experiencing  a  decline  in  employment: some  skilled labour hoarding during  the  crisis.  Such  labour  hoarding appears related to some firms’ strategic goals (especially innovation)

Some  indications  that higher short-term contracts churning rate would no  longer be associated with a higher productivity, suggesting a change in the use of CDD that may hamper productivity

>>>these results are consistent with our general macro-estimations

3-Micro-analyses using REPONSE

Dependent variable: ln(value added per employee)

(1) (3) (5) (6)

2005 2011 2011 Declining

2011 Non-declining

Ln(Total assets per employee) 0.32***(0.03)

0.30***(0.03)

0.32***(0.06)

0.29***(0.02)

Share of employees aged 55- -0.41*(0.23)

-0.24(0.16)

0.08(0.20)

-0.32(0.21)

Share of high-skilled occupationsRef.= share of medium-skilled

0.27*(0.15)

0.26**(0.11)

0.27(0.27)

0.36***(0.11)

Controls: Establishment age, pct. of women, pct. of low-skilled, 2-digit industry, firm size

Yes Yes Yes Yes

R² 0.60 0.62 0.67 0.65

N 1591 1938 595 1341

Aged workers or skilled occupations and apparent labour productivity

3-Micro-analyses using REPONSE• Results (2): work practices and productivity High involvement dimensions include employee shareholding and 

organised  voice.  High performance practices:  quality management, job rotation, autonomous workteams.

No changes in the correlation between performance practices and productivity between 2005 and 2011; 

In  2005,  organised  voice  and  employee  shareholding  were associated  with  higher  productivity.  In  2011,  no  significant relationships for these variables

>>>  a break in the impact of high involvement practices on productivity after the 2008 shock?

E.g.  consistent with  the  drop  of  stock markets  (still  30%  below  the pre-recession level)

Work practices and productivity

3-Micro-analyses using REPONSE

Dependent variable: ln(value added per employee)

(1) (2) (3) (4) (5)

2011 2011Mono-est.

2011 2005-2011Panel

2011Relative productivity

Ln(Asset/employee) 0.302***(0.028)

0.284***(0.028)

0.361***(0.044)

0.340***(0.056)

Organized voice 0.004(0.026)

0.004(0.040)

-0.003(0.025)

-0.022(0.025)

(+) ns

Empl. shareholding -0.033(0.023)

0.038(0.028)

-0.040(0.026)

-0.009(0.049)

(-) ns

Quality management -0.020(0.011)

0.000(0.038)

0.006(0.024)

(+) ns

Autonomous team -0.041**(0.020)

-0.077***(0.028)

0.000(0.000)

(+) ns

Job rotation -0.010(0.019)

-0.043(0.041)

0.008(0.017)

(-) ns

Organized voice in 2005 0.004(0.029)

Employee shareholding in 2005 0.080(0.055)

Ln(Productivity per employee in 2005)

0.550***(0.059)

Ln(Asset per employee in 2005) -0.264***(0.047)

2-digit industry Yes Yes Yes Yes YesOther controls Yes Yes Yes No YesN 1857 717 1426 530 2569R² 0.63 0.61 0.71 0.60 0.03

Conclusion

• Main factors explaining France productivity slowdown according to our analysis:

-development  of  low  productive  jobs,  in  relationship  with labour market reforms;-increase  in  high-skilled  labour  force  resulting  in  less employment reactivity facing a recession;-lower performance of high involvement practices.

• A revival is possible…the  two  last  factors  might  be transitory…  if  the European macroeconomic environment moves.