Gross Job Flows in Europe

1
Gross Job Flows in Europe
  New Directions in LabourMarket Flexibility
ResearchLondon, Wednesday 26 November
2003http//www.dti.gov.uk/er/emar/events.htm    

• Giovanna Vallanti
• London School of Economics
• (joint paper with Ramon Gomez Salvador and Julian
  Messina)

2
Structure of the talk

• Motivation and objectives
• Definitions
• Empirical and theoretical literature
• Description of the data set and sample
• Descriptive statistics on job flows
• Empirical Results
• Conclusions

3
Motivation

• Fast growing literature on job flows
• Davis and Haltiwanger (QJE, 1992)
• Davis and Haltiwanger (Handbook of Labour, 1999)
• Main limitation cross-country comparisons are
  difficult. Few attempts are based on job flows
  calculated from
• different collection methods (administrative vs.
  survey)
• different units of measure (firms vs.
  establishments)
• different sectoral coverage
• different sampling intervals

4
Objectives

• Present comparable estimates of job flows for 12
  EU countries.
• Once controlled for structural factors
• Re-examine the relationship between job flows and
  the business cycle.
• Relate job reallocation flows to labour market
  institutions.

5
Theoretical Considerations

• Bertola (EER, 1990) firing costs reduce both JC
  and JD rates and consequently JR.
• Mortensen and Pissarides (1999) and Pissarides
  (2000) within a search and matching framework
• Unemployment benefits and labour taxes decrease
  JC and increase JD Ambiguous effect on JR
• Job subsidies increase JC (lower cost of
  matching) and increase JD by increasing market
  tightness Higher JR
• Bertola and Rogerson (EER, 1997) wage
  compression, increase JR through an increase in
  both JC and JD.
• Garibaldi (EER, 1998). More stringent EPL make JR
  a-cyclical or pro-cyclical.

6
Empirical Literature. Findings I

• A high number of jobs are simultaneously created
  and destroyed in countries and sectors regardless
  of the cycle phase.
• Job creation and job destruction are negatively
  correlated but not perfectly.
• JR is negatively associated with the cycle in the
  US, Canada and UK (Davis and Haltiwanger,1992
  Baldwin at al, 1994 Konings, 1995).
• For the countries of Continental Europe the
  evidence is mixed and in general job reallocation
  has been found to follow an a-cyclical pattern

7
Empirical Literature. Findings II

• Job reallocation is negatively associated with
  capital intensity. Typically, more jobs are
  created and destroyed in the service sector than
  in the manufacturing sector.
• The intensity of job reallocation empirically
  associates with some firm-specific
  characteristics. It has been found negatively
  correlated with
• firms age
• firms size
• There are important similarities in JR across
  countries, despite cross-country differences in
  institutional flexibility

8
Job flows and Institutions. Findings III

• Garibaldi et al. (1997) OECD (1999) negative
  correlation between JR and EPL.
• Heyman (2001) finds a positive relationship
  between JR and the degree of wage compression.
• Lucifora (IJIO, 1998) and Blanchflower and
  Burgess (1996) lower rate of job turnover in
  unionised sectors.
• Leonard and Van Audenrode (EP, 1993) Industrial
  subsidies have a negative impact on JR
• Salvanes (SJE, 1997), in a multivariate
  framework
• Stricter EPL and the degree of wage
  centralisation have a negative impact on job
  turnover
• Industrial subsidies have a positive impact on JR

9
Definitions

• Gross job creation (JCst) and job destruction
  (JDst) rates
• Gross job reallocation
• Net employment growth

10
Database I

• Amadeus database (Bureau van Dijk) contains
  annual firm-level observations for European
  countries over the last 10 years.
• Information based on the balance sheets of
  private non-financial firms, containing info on
  employment, date of incorporation, location and
  sector of operation.
• Selection criteria a company should satisfy at
  least one of the following size criteria to be
  included in the database
• operating revenue equal to at least 1 (1.5)
  million euro
• total assets equal to at least 2 (3) million euro
• number of employees equal to at least 10 (15)

11
Database II

• Main advantages
• The data collection method is reasonable
  homogeneous across countries
• Information is provided on narrowly defined
  sectors and data on both manufacturing and
  non-manufacturing sectors are reasonably
  representative.
• Main disadvantages
• Entry and exit of firms
• Firm and not establishment data
• Small bias towards larger firms.

12
Sample Selection

• 12 countries EU except Italy, Greece and
  Luxemburg.
• Continuing firms (excluding start-ups and
  shout-downs)

13
Final Sample Composition
Note Coverage is defined as the percentage of
employment in the final sample with respect to
total employees (excluding public administration
and financial sectors) as measured trough the EU
Labour Force Survey
14
Employment distribution by sectors
15
Employment by size classes
France
Germany
Spain
UK
16
Employment GrowthOur sample (NET) vs. OECD
17
Job Flows in EU Countries
18
Job Flows by Sector
19
Job Flows by Size Class
20
Job Flows by Age and K Intensity
By Age
By Capital Intensity
21
Multivariate Analysis

• We construct JC, JD and JR statistics for
  narrowly defined sectors according to
• 4 age groups
• 7 sectors of activity
• 4 size groups
• 12 countries
• 10 years
• 4 capital intensity groups
• With capital Without Capital
• potential cells 53,760 17,837
• actual cells 13,440 7,111

22
Basic Methodology

• where
• JFist denotes for JCist, JDist or JRist
• Ds is a dummy for each class as defined above
  (age, sector, size and capital intensity)
• We estimate these equations by OLS, weighting
  each observation by the average employment in
  both consecutive years.
• Standard errors are robust to heteroskedasticity
  and country clustering.
• This methodology allows examination of
  differences in job flows across countries once
  sectoral, size and age effects are account for.

23
Country fixed-effects (reference country UK)
24
Country fixed-effects (after controlling for the
cycle)
25
Job Flows and the Business Cycle

• where BCst is an indicator of the business cycle.
  Two indicators were considered
• Growth of GDP
• Net employment change for the aggregate
• We repeat each of this regressions including and
  excluding time dummies and both with and without
  indicators of capital intensity.

26
JR, JC JD. Cyclical effects
27
Institutional Factors I

• Tax and benefits systems
• duration and generosity of unemployment insurance
  systems. 0,1
• tax wedge difference between the real (monetary)
  labour cost faced by the firms and the
  consumption wage received by the employees.
  Normalized by GDP.
• Wage-setting institutions
• Index on the extent of coordination in the wage
  bargaining process 1,3

28
Institutional Factors II

• Restrictions to hiring and firing
• Overall EPL index 0.5-3.7
• EPL for regular contracts 0.7-4.7
• Share of workers holding temporary contracts (as
  a fraction of total employment)
• Sectoral employment subsidies Eurostat indicator
  of sectoral and ad hoc state aid. Normalized by
  GDP.

29
Institutional determinants of JR, JC and JD OLS
Estimates
Note The regressions include age, industrial
sector, year and firm size dummies. Range values
Union Co-ordination(1-3) Unemployment Benefit
Duration(0-1) Tax Wedge (18.61-53.33) Ratio of
Temporary Contracts (4.33-34.99) Employment
Subsidies (0.23-1.93) Employment Protection
(0.50-3.70) and Employment Protection Regular
Contracts (0.70-4.30). Calculated standard errors
are robust to country clustering. t-statistics in
parenthesis
30
Sensitivity Analysis. The effects of institutions
dropping one country at a time. OLS estimates
Note Min. Minimum value Max Maximum value
pgt.9 Percentage in which the probability of
rejection of the parameter of interest is below
10 per cent. The regressions include age,
industrial sector, year and firm size dummies.
Calculated standard errors are robust to country
clustering
31
Sensitivity Analysis. The effects of institutions
entered one at a time. OLS estimates
Note Min. Minimum value Max Maximum value
pgt.9 Percentage in which the probability of
rejection of the parameter of interest is below
10 per cent. The regressions include age,
industrial sector, year and firm size dummies.
Calculated standard errors are robust to country
clustering. t-statistics are in parenthesis
32
Conclusions

• There are consistent cross-country similarities
  in the patterns of job flows. We find larger
  flows in smaller and younger firms operating in
  the service sector.
• Job reallocation is on average pro-cyclical. EPL
  decreases the cyclical volatility of job
  destruction relative to job creation, making job
  reallocation more pro-cyclical. This may be
  interpreted as evidence of the importance of
  firing costs, which obstruct the destruction of
  employment.
• Evidence points out to firing restrictions, the
  tax burden and the degree of wage setting
  coordination as factors obstructing the
  reallocation of workers. Moreover, evidence
  suggest that firms make use of temporary
  contracts to overcome those institutional
  barriers.

33

• Discussant Felix Ritchie,Office of National
  Statistics
• Then questions from the floor

34
  New Directions in LabourMarket Flexibility
Research London, Wednesday 26 November
2003 http//www.dti.gov.uk/er/emar/events.htm