can we really explain worker flows in transition?

24
Can we really explain worker flows in transition economies? Can we really explain worker flows in transition economies? Evidence from the Life in Transition Survey Joanna Tyrowicz Lucas van der Velde GRAPE Group for Research in APplied Economics December 2014, Warsaw Economic Seminars

Upload: grape

Post on 17-Jul-2015

86 views

Category:

Economy & Finance


0 download

TRANSCRIPT

Page 1: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Can we really explain worker flows in transitioneconomies?

Evidence from the Life in Transition Survey

Joanna TyrowiczLucas van der Velde

GRAPEGroup for Research in APplied Economics

December 2014,Warsaw Economic Seminars

Page 2: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Table of contents

1 Introduction

2 Stories of reallocation

3 Hypotheses

4 Data and methods

5 Results

6 Conclusions

Page 3: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Introduction

Introduction

Motivation

Analyses so far is highly selective - few countries, few periods

Lack of a solid, complete theoretical basis.

Is transition over? Poland over half million people still in public ormixed ownership companies. In our sample, public size reduced byhalf till 2005.

Our goal: to understand better worker flows in transition economies

Advantage: new, comprehensive retrospective data: Life inTransition Survey (EBRD)

Page 4: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Introduction

Countries under study

Table : Countries analysed by previous literature

Year N 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04

Estonia 2

Russia 2

Ukraine 3

Bulgaria 1

Poland 3

Romania 1

Slovenia 2

Slovak 1

Note: Ticks indicate that the countryperiod was analysed in the literature. Papers were searchedfor in the EconLit database with keywords: ‘reallocation’; ‘transition’ ‘job creation’ ‘jobdestruction’

Page 5: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Stories of reallocation

Aghion & Blanchard (1994) - optimal speed of transition

Inefficient (public sector) jobs collapse

State can subsidize firms (postpone collapse) or redundant workers(with safety nets)

Taxes raised to finance this make creating jobs costly,desynchronizing JD & JC⇒ Simple link between U, JC and pace of privatization, state canalleviate social costs of transition by choosing the adequate pace ofjob destruction in the public sector

Limits

Simplifications concerning the role of sectoral reallocationSimplification of the dynamics of the two sectorsWorkers homogeneous: no demographics or changes in education -inconsistent with empirical evidence ???

Extensions: International migration (Bruno, 2006); heterogeneousworkers (Boeri, 2000; Balla, 2008); job-to-job flows (Tichit, 2006)

Page 6: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Stories of reallocation

Caballero & Hammour (several papers)

Main concept: appropriability

Endogenous job creation and destruction based on capital specificityand incomplete contracts.

⇒ Used to explain reallocation and technological change inadvanced economies.

Limits

No treatment of public sector / taxes / subsidiesWorkers homogeneous: no demographics or changes in education -inconsistent with empirical evidence (???)Sectoral changes increase productivity, which is not always true.(Dimova, 2008; Orazem and Vodopivec, 2009)

Page 7: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Stories of reallocation

Common limitations in applying these theories to the data

Data limitations on controlling for gross (individual level) and net(firm level) flows

The different role of worker flows (reallocations) vs job flows(privatizations)

Privatized vs new (de novo) firms – all private equal?

What if a worker holds more than a one job during the transitionperiod? Which transition do we capture?

Page 8: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Hypotheses

Our statements to be tested

1 Flows are generally not AB or CH

2 Demographic changes (new entries and early exits) explain most ofthe reallocation

3 AB explains unemployment better than CH in transition countries,but they both poorly explain employment

4 Channels of mediation suggested by AB and CH do not seem to bedriving the processes, demographics do

Page 9: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Data and methods

Data sources

Life in transition Survey - 27 transition countries, 18 years

Homogeneous survey compiled by the EBRD in 2006 and 2010.Life history in the 2006 edition. Sample covers years from 1989 to2006.Limitations: missing variables (e.g. wages, firm size), identificationof flows (privatized vs de novo), recall bias.

Other sources

ILO Stat and Fondazione: Wages and EPLEBRD: Transition measures.World Bank: GDP per capita.Penn tables: Labour share in GDP, Employment to population ratio.

Page 10: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Data and methods

LiTS and other data sources

Country YearServices Industry Private Services Industry Private(LFS) (LFS) (SES) (LiTS) (LiTS) (LiTS)

Bulgaria2000 51.8 39.6 57.2 36.0 48.72002 54.9 38.3 55.9 60.0 34.4 53.5

Estonia1997 53.1 33.1 58.4 30.6 52.72002 56.0 32.9 91.8 59.8 30.9 62.2

Latvia1998 47.4 30.1 67.1 23.6 51.22002 49.0 27.7 88.0 67.1 24.4 59.7

Poland2000 46.1 40.1 59.6 34.6 50.02002 51.5 37.8 47.1 59.0 34.3 53.4

Romania1997 48.4 22.8 54.1 39.7 44.22002 58.0 24.7 65.3 58.8 36.1 54.8

Slovakia1998 50.2 29.2 62.6 30.1 39.72002 52.7 27.7 63.0 65.6 28.6 45.9

Note: Own calculation on the basis of data from LiTS, the EU-Labour Force Surveys (LFS) andthe Structure of Earnings Survey (SES).

Page 11: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Data and methods

Unemployment rates

Figure : Fit of the unemployment rates

Page 12: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Data and methods

Overall labour market trends

Figure : Changes in the labour market composition.4

.45

.5.5

5.6

.65

Fem

ale

1990 1995 2000 2005

3035

4045

Age

1990 1995 2000 2005

.1.2

.3.4

.5.6

Hig

h E

d.

1990 1995 2000 2005

0.2

.4.6

.8P

rivat

e

1990 1995 2000 2005

Page 13: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Results

Definitions

AB: public ⇒ private sector (within the same industry)

CH: manufacturing ⇒ services (within the same sector)

ABCH: public manufacturing ⇒ private services

NONE: private service ⇒ public manufacturing

SAME: within sector and industry

EXIT: To retirement

ENTRY: Into employment

Page 14: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Results

H1: which flows dominate? How much the models explain?

Figure : Relative importance of different flows

0 .2 .4 .6 .8 1

AZEARMTJK

GEOMKDBIH

MNEKGZALB

MDASRBSVKBLRHRVUZBSVNPOLLTUUKRROMBGRKAZCZEESTRUSHUNLVA

AB CH SAME ABCH

NONE To retirement From school

Page 15: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Results

One speed of transition? + job-to-job flows dominate!

Figure : Evolution of different flows0

24

6

1990 1995 2000 2005

ABCH

ABCHSAME

NONE

Page 16: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Results

H2: which flows explain employment?

Table : Movements to employment

N⇒ E U⇒ E E⇒ E

AB 0.968*** 0.868*** 0.917***CH 0.649*** 0.662*** 0.594***ABCH -0.623*** -0.592*** -0.531***Same industry - Manufacturing 0.102*** 0.348*** 0.479***Same sector - Public 0.883*** 0.882*** 0.916***Same Sector - de novo 0.854*** 0.892***Reincidence of unemployment -0.207*** -0.004***

Personal characteristics Yes Yes YesCountry dummies Yes Yes YesYear dummies Yes Yes Yes

Number of id 15,131 9,968 13,107R2 between 0.825 0.276 0.641R2 within 0.834 0.314 0.641

Notes: Panel linear probability models (RE). Robust standard errors used but not reported.Asterisks denote 1 % confidence levels

Page 17: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Results

H3: which flows explain unemployment (better)?

Table : Link between unemployment rates and flows

AB CH SAME ABCH NONE EXIT ENTRY

flow 2 0.057*** 0.089 0.009 0.220* 0.026 0.006 0.037*flow -0.789*** -0.688 -0.533*** -1.067** -0.595* -0.060 -0.762***N 486 486 486 486 486 486 486R2 0.888 0.885 0.890 0.886 0.886 0.885 0.889

Notes: In all cases the dependent variable was detrended Unemployment Rate. The independentvariables are the total number of flows of each type in each country.

Page 18: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Results

Tehnical note on duration models

Used to model in a time-event framework → event is ”death”

We employ proportional hazard model :

λ(t, x , α, β) = λ0(t, α)φ(xβ)

Cox partial likelihood method:Conditional probability that an observation finishes in t

= λt (t,β,x)∑ni=1 λt (t,β,x)

In a competing risks model, individuals can ”die” in different ways

Page 19: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Results

Examples of survival and hazard functions

0.00

0.25

0.50

0.75

1.00

0 5 10 15 20analysis time

Primary Secondary Tertiary

Kaplan-Meier survival estimates

Page 20: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Results

Examples of survival and hazard functions

.04

.06

.08

.1.1

2

0 5 10 15 20analysis time

Primary Secondary Tertiary

Smoothed hazard estimates

Page 21: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Results

H4: do the channels of mediation work?

Table : Durations models

VARIABLES TOTAL CH AB (de novo)Unemployment rate 8.850*** 9.035*** 8.435***Unemployment rate 2 -12.305*** -18.396*** -16.145***Entry 3.727*** 8.737*** 6.219***Exit 2.950 0.676 7.472**ULC dynamics -0.062 0.878 -0.601Public 0.472*** 0.346** 2.917***De novo 0.049 -0.001 0.486Manufacturing 0.455*** 0.190 0.522***Construction 0.457*** 0.190 0.834***Services 0.527*** -0.283* 0.756***High skill jobs -0.340*** -0.210 -0.143Personal characteristics Yes Yes YesAIC 39362.182 6618.8377 8937.4545BIC 39517.767 6774.4224 9093.0392

Notes: Estimates from a proportional hazard Cox model with country specific baseline hazardratios.

Page 22: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Results

Bonus round: Winners and losers of transition

N⇒ E U⇒ E E⇒ E

Female -0.003*** 0.002 -0.003***(-3.257) (0.510) (-3.395)

Age (in tens) 0.002*** 0.200*** -0.012***(4.351) (30.095) (-5.400)

Age (In thous) -0.004*** -0.294*** 0.008***(-8.930) (-35.521) (3.263)

Urban 0.001 0.000 0.001(0.637) (0.062) (0.718)

Secondary education 0.003*** 0.049*** -0.001(3.344) (10.823) (-0.864)

Tertiary education 0.015*** 0.133*** -0.004**(6.010) (19.091) (-2.281)

Observations 196,743 65,194 133,619Number of id 15,131 9,968 13,107R2 between 0.641 0.641 0.641R2 within 0.641 0.641 0.641

Page 23: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Conclusions

Summarizing

1 The focus on AB and CH movements is not enough in transitioneconomies.

2 Demographic flows were of great significance and should be considered infurther analysis

3 Individual characteristics played an important role: winners and losers oftransition.

4 Unemployment rates have an impact on flows, but the transmissionmechanisms should be reconsidered

What is next?:

Corroborate our findings with national data sources.

Deepen the analysis of the role of institutions on labor market transitions.

Develop an integrated ABCH model of transitions.

Page 24: Can we really explain worker flows in transition?

Can we really explain worker flows in transition economies?

Conclusions

Questions or suggestions?

Thank you for your attention