UNEMPLOYMENT IN TRANSITION COUNTRIES (CENTRAL EUROPE): WHY SO HIGH?

Lecture plan

Introduction of the issue Research approaches Matching function approach

Theory Simple statistics Estimation strategy Various empirical problems

Findings and remaining questions Presentation will be available Readings

Basic Ideas

Unemployment unknown under communism But emerged rapidly

Is a major problem in most CE economies

Q: Is unemployment the result of unfinished transition from plan to market => need to complete it macro policies and external shocks => macro policies key economic structures (mismatch) => focus on labor market

institutions, labor mobility and skill formation

Unemployment rates 1991-2005 [%]

0

5

10

15

20

1990 1995 2000 2005

Une

mpl

oym

ent r

ate

[%]

Czech Republic Slovak Republic E. Germany

W. Germany Hungary Poland

Research approach 1: Labor Market Institutions and Unemployment (WB, OECD)

Calculate measures of L mkt institutions Unemployment Insurance (UI): net replacement rates (declining) UI: strictness (flat or increasing) Wage bargaining (high or increasing decentralization) Employment protection (not strong by EU standards) Tax wedge, employer + employee income tax (high and stable)

U not related to institutions in regressions Except possibly for initial UI benefits and tax wedge

Conclude: U not explained by labor market institutions alone If institutions matter, likely in combination with other factors

Heckman’s critique (simplistic indicators, small datasets, noise)

Research approach 2: Job Destruction, Job Creation and Unemployment

L in new sector has not replaced L lost in old sector Q: Is labor reallocation (transition) still at work?

Looks at JC, JD and U as initial conditions and policies vary Amadeus database => construct JC ad JD rates for 10 TEs

Macro-level regression findings: Unemployment has a negative effect on JC in new firms

High U associated with higher UI benefits and taxes => lower JC? Current long-term U depends on history of short term U and

hence JC and JD Firm-level regression results:

Foreign ownership has a positive effect on employment growth

Research approach 3: Initial Human Capital (HC) and Regional L Mkt(Stepan Jurajda)

Transition: High dispersion and lack of convergence in regional unemployment rates (URs) focus on regional differences in HC endowments Idea: Skill and skill-capital complementarities explain high

regional dispersion in unemployment

Findings (BU, CR, HU, UKR): Over one-half of variation in regional URs explained by

concentration of HC Regional variation in HC is wide and rising K and skilled L move to regions with high skill concentration

Research approach 4: Skill Endowments in the CE Countries (Janos Köllo)

Thesis: Presence of many workers with only primary or vocational education => low employment rates Industry and agriculture (simple tasks) declined in CEs Growing tertiary sector demands higher skills (communication) => Employment of low skilled workers fell dramatically

Evidence: IALS, workplace skill requirements, panel data on L and W of unskilled in occupations, firm-level skill share equations (response to tech. change)

=> Policy issues related to education and training

UNEMPLOYMENT AND WORKER-FIRM MATCHING IN CENTRAL EUROPE

Daniel MünichJan Svejnar

Basic Ideas Q: Is unemployment a result of

ongoing transition (restructuring ) macroeconomic policies and external shocks economic structures (mismatch) => focus on L mkt institutions

(as in Western Europe), labor mobility and skill formation

Use district-level panel data on the unemployed U, vacancies V, inflow S into unemployment,

and outflow O from unemployment in CR, HU, PO, SR, and East and West parts of Germany

Examine the three hypotheses in the context of the efficiency of matching of the U and V

Matching of unemployed U and vacant jobs V with frictions (notion like production function) leads to outflow from unemployment O (flow chart)

Using flow identity (inflow=outflow; S=O) equilibrium unemployment rate u* for given inflow rate s, vacancy rate v, and matching function O=(U,V)

wheresoos

su

1

1*

VAUVUOU

VUOo

E

Ss

LF

Uu ),( ,

),( , ,*

MATCHING FUNCTION APPROACH

Structural Model

Probability of a job offer p=p(V/U); probability of a job offer to match = 1 – G(mpr)

Steady-state unemployment rate (UV curve):

Vacancy supply curve (VS):

s … exogenous inflow

mpr… reservation marginal product from a matchz … income while unemployed

γ0… worker’s costs of search

tightnessLM /0

UV

zmpr

u

v

UV

VS

Literature: Petrongolo B. and C. Pissarides (2001), “Looking into the Black Box: A Survey of the Matching Function,” Journal of Economic Literature 39, June: 392–431. Jackman R., C. Pissarides, and S. Savvouri (1990), “Unemployment Policies and Unemployment in the OECD,” Economic Policy, October: 449–490. Berman E. (1997), “Help Wanted, Job Needed: Estimates of a Matching Function from Employment Service Data,” Journal of Labour Economics 15(1): S251–S292.

rmpGps

su

1*

Beveridge curve dynamics

Beveridge curve during 1970-1990

Beveridge curve (Czech Republic, seasonally adjusted data)

0.4

0.8

1.2

1.6

2.0

2.4

2.0 4.0 6.0 8.0 10.0 12.0Unemployment rate (%)

Vaca

ncy

rate

(%)

I-95

VI-96

VI-99

IX-01XII-04

Beveridge curves for CE countries

REDUCED FORM In equilibrium, O = S and U = U*= const. For given level of (exogenous) inflow S and vacancies V,

equilibrium U* (not necessarily observed!) is defined implicitly by the matching function

Implying

Allows for determination of parameters not observing equilibrium

O A U V S *

US

A V*

Conceptual framework of matching functions

O = M(U,V) Some authors expect the matching function M to display constant

returns to scale Others have identified reasons such as externalities in the search

process, heterogeneity in the unemployed and vacancies and lags between matching and hiring, why increasing returns may prevail

Increasing returns are important because they may constitute a necessary condition for multiple equilibria and provide a rationale for government intervention.

We find that increasing returns appear to be an important phenomenon especially in the later (1997-2003) than the earlier (1993-96) period more pronounced in some of the economies than others

Hypotheses about reasons for high U H1: restructuring still at work -- inflow S (from old jobs)

high => U high due to high turnover

H2: U-V matching “fine”, high U caused by low L demand (macro policies, exchange rate, shocks) => low V relative to S (irrespective of U)

H3: inefficient U-V matching (L mkt institutions or geographical or skill mismatch) => U and V both high but not in the same districts or skill groups

AGGREGATE TIME SERIES OF KEY VARIABLES

Figure 2 – Evolution of U, S, O, and V in west part of Germany (benchmark case)

West Germany an “intermediate” case in 1991-2005 unemployment rate rising in two waves from 5% to 10% inflow rate rising in two waves from 0.9 to 1.6 outflow rate decreasing in two waves from 18.5 to 13.6 Vacancy rate fluctuating (in two waves ) between 0.7 and 1.2

Figure 2

West Germany: Unemployment and vacancy stock-flow rates

0

5

10

15

20

25

1990 1995 2000 2005

Une

mpl

oym

ent a

nd

outf

low

rat

e [%

]

0.00.51.01.52.02.53.0

Inflo

w a

nd v

acan

cy

rate

[%]

Unemployment Rate Outf low rate Inf low rate Vacancy rate

Figure 3 – Evolution of U, S, O, and V in the Czech Republic

CR is somewhat similar (intermediate) U rate rising in two waves from 3% in early-to-mid 1990s to 10% inflow rate has risen outflow rate has declined from a high level vacancy rate rose to 1.9 and then declined to 0.8-1.1

Figure 3

Czech Republic: Unemployment and vacancy stock-flow rates

0

5

10

15

20

25

1990 1995 2000 2005

Un

em

plo

ym

en

t a

nd

o

utf

low

ra

te [

%]

0.00.51.01.52.02.53.0

Infl

ow

an

d v

acan

cy

rate

[%

]


Czech example of seasonality in the data

Figure 4 – U, S, O, and V in East Germany

East Germany – one extreme case unemployment rate rising from 11.5% to 18.6% inflow rate rising dramatically Outflow rate first rising and then stabilizing around 13-14% a vacancy rate rising from 0.4% in 1991 to about 1% in the late

1990s and remaining at or below that level in the 2000s

Figure 4

East Germany: Unemployment and vacancy stock-flow rates

0

5

10

15

20

25

1990 1995 2000 2005

Un

em

plo

ym

en

t a

nd

o

utf

low

ra

te [

%]

0.00.51.01.52.02.53.0

Infl

ow

an

d v

acan

cy

rate

[%

]


Figures 5 and 6 – Evolution of U, S, O, and V in Poland and Slovakia

Poland and Slovakia – also extreme cases unemployment rate rising quickly to the 14%-20% range relatively steady high inflow rates low outflow rates vacancy rates well below 1%

Figure 5

Slovak Republic: Unemployment and vacancy stock-flow rates

0

5

10

15

20

25

1990 1995 2000 2005

Un

em

plo

ym

en

t a

nd

o

utf

low

ra

te [

%]

0.00.51.01.52.02.53.0

Infl

ow

an

d v

acan

cy

rate

[%

]


Figure 6

Poland: Unemployment and vacancy stock-flow rates

0

5

10

15

20

25

1990 1995 2000 2005

Un

em

plo

ym

en

t a

nd

o

utf

low

ra

te [

%]

0.00.51.01.5

2.02.53.0

Infl

ow

an

d v

acan

cy

rate

[%

]


Figure 7 – Evolution of U, S, O, and V in Hungary

Hungary is also special lowest unemployment rate; after reaching 10-11.5% in mid

1990s, lowered to around 8-9% inflow rate as a share of the labor force at 1.2-1.3% outflow rate as a share of the labor force at 1.2-1.4% kept the vacancy rate at 1.0-1.1%

Hungary’s success brought about by keeping the outflow rate relatively high and inflow rate relatively low

Figure 7

Hungary: Unemployment and vacancy stock-flow rates

0

5

10

15

20

25

1990 1995 2000 2005

Un

em

plo

ym

en

t a

nd

o

utf

low

ra

te [

%]

0.00.51.01.5

2.02.53.0

Infl

ow

an

d v

acan

cy

rate

[%

]


Table 2: Persistence of Regional Differentials - Pearson's Rank Correlations

Base Year 1992 Base Year 1999

Year CR SR EG WG HU PL* Year CR SR EG WG HU* PLUnempl. Rate

1996 0.81 0.82 0.54 0.94 n.a. 0.90 2002 0.95 0.95 0.87 0.96 0.95 0.861999 0.73 n.a. 0.32 0.94 n.a. 0.83 2005 0.91 0.92 0.71 0.89 0.92 0.83

Inflow Rate1996 0.79 0.75 0.60 0.89 n.a. 0.90 2002 0.94 0.93 0.87 0.97 0.96 0.941999 0.70 n.a. 0.47 0.87 n.a. 0.87 2005 0.90 0.82 0.51 0.80 0.95 0.92

Outflow Rate1996 0.50 0.57 0.43 0.84 n.a. 0.48 2002 0.86 0.88 0.68 0.87 0.86 0.451999 0.38 n.a. 0.46 0.84 n.a. 0.47 2005 0.83 0.86 0.54 0.78 0.66 0.72

Vacancy Rate1996 0.33 0.25 -0.03 0.51 n.a. 0.55 2002 0.69 0.35 0.66 0.81 0.85 0.81999 0.29 n.a. 0.01 0.51 n.a. 0.46 2005 0.40 0.34 0.00 0.38 0.50 0.6

U/V ratio1996 0.81 0.82 0.54 0.94 n.a. 0.90 2002 0.95 0.95 0.87 0.96 0.95 0.861999 0.73 n.a. 0.32 0.94 n.a. 0.83 2005 0.91 0.92 0.71 0.89 0.92 0.83

_

Figures suggest

West Germany consistent with H1-3; U risen with increasing inflows S (H1), V declined while inflow risen (H2), the U and V rates are relatively high (H3)

CR starts with low U but increasingly conforms to H1 (higher U and S) and H2 (V low relative to S and U)

East Germany conforms to H1 as well as H2 Slovakia and Poland consistent with H1 and H2 throughout the

1990s and 2000s Because of low unemployment, Hungary does not fit clearly into any

H -- has an element of all three Hs: inflow is relatively sizable (H1), the vacancy rate is low relative to inflow (H2), unemployment and vacancies are relatively high (H3)

Literature on matching in TEs

Grown rapidly Produced contradictory results Studies use different methodologies and data Methodologically, they differ especially with respect to the

specification of the matching function and treatment of returns to scale inclusion of other explanatory variables that might affect outflows extent to which they use static or dynamic models

In terms of data, the studies differ in whether they use annual, quarterly or monthly panels of district-level or more

aggregate (regional) data cover short or long time periods

None adjusts the data for the varying size of the (district or region)

Our approach

Unlike other studies, we use a more up-to-date empirical methodology and superior data control for the endogeneity of explanatory variables account for the presence of a spurious scale effect introduced by

the varying size across units of observation (districts) use long panels of comparable monthly data from all districts in

the countries that we analyze

Empirical Specification (simple, but…!)

Cobb‑Douglas function which may be written in a deterministic form as

(2)

Ui,t‑1 = number of unemployed in district i at the end of period t-1

Vi,t‑1 = number of vacancies in district i at the end of period t-1

Oi,t = outflow to jobs during period t

A captures the efficiency of matching.

AV + U = O 1-ti,1-ti,ti, lnlnlnln

o u v ai t i t i t i i t, , , , 1 1

Empirical Specification

Let lowercase letters stand for logarithms of variables ai be district specific effects

εi,t be an idiosyncratic error term

Can write (2) as

(3) o u v ai t i t i t i i t, , , , 1 1

o u v ai t i t i t i i t, , , , 1 1(3)

• OLS not appropriate if ai are correlated with u and v• Correlation likely to exist due to differences between districts (draw graph)• Specific factor is district size (spurious scale effect)• With panel data, one can use means deviation or first differencing to remove ai

• But RHS u and v are predetermined through previous matching

(endogenous) inconsistent estimates IV needed first differencing preferred

Estimation problems

o o o u u v vt t t t t t t t t 1 1 2 1 2 1 ( ) ( )

First difference transformation contaminates the transformed variables only with recent error terms {εt: t =

T-1, T-2} To see this, rewrite (5) in a first difference form

(6)

U U S O

U U S Ot t t t

t t t t

1 2 1 1

2 3 2 2

o u v

o u vt t t t

t t t t

1 2 2 1

2 3 3 2

Lagged outflows in (4) in turn given by a lagged version of (3)

… and further lags of U (or S), and V can be used as valid instruments.

District mean deviations transformation (fixed-effects): contaminates variables with all error terms.

Disentangle ai

From 1st differences back to levels Problem with poor measurement of vi,t

, , , ,a o u vi t i t i t i t

Newly unemployed

Studies (e.g., Coles and Smith,1994) suggest propensity to match higher at time of entry into unemployment Newly unemployed search through all existing vacancies May have not experienced depreciations of skills

Remaining unemployed match only with the newly posted vacancies

To reflect this, include inflow into unemployment as an additional explanatory variable

Total outflow v. outflow to jobs

Data on outflow to jobs are available only for the Czech Republic, while data on total outflow are available for all the countries

We carry out the estimation for the Czech Republic using both measures and find that the estimates based on total outflow and outflow to jobs are similar

Assume the lack of data on outflow to jobs in other countries should not have a dramatic impact on our results (see also Petrongolo and Pissarides, 2000, for similar evidence from other countries)

Other empirical problems

Measurement error

Continuous vs. discrete process

Segmented labor market ….

Data

Panel of data on 76 Czech, 38(79) Slovak, 21 Hungarian, 34 East German and 140 West German districts. The data cover January 1991- 2005 and contain monthly observations for the following variables:

Oi,t = the number of individuals flowing from unemployment in district i during period t;

Ui, t = the number of unemployed in district i the end of period t; Si,t = the normalized number of individuals flowing into unemployment

(the newly unemployed) in district i during period t;[2] Vi,t = the number of vacancies in district i at the end of period t;

[2] Although the individuals flow into unemployment in the same calendar month, they enter on different days within the month. This means that they face different probabilities of finding vacancies during the calendar month. Assuming, that the inflow is approximately uniform over the month, we multiply the total monthly inflow by .5.

Matching function estimates for West Germany during 1997-2005

Trend Std.Err. β Std.Err. γ Std.Err. δ Std.Err. RTS p-value adjR2

Panel A: Cross-sectional estimators

OLS 0.012 0.001 0.68 0.00 0.15 0.00 - - 0.83 0.00 0.85

OLS (Month Dummies) 0.011 0.001 0.69 0.00 0.13 0.00 - - 0.82 0.00 0.90

OLS (Size Adjusted) 0.010 0.001 0.55 0.03 0.03 0.02 - - 0.58 0.00 0.62

Panel B: Panel data estimators

Random Coefficients 0.010 0.000 0.74 0.01 0.08 0.00 - - 0.81 0.00 0.65

Fixed Effects 0.010 0.000 0.74 0.01 0.07 0.00 - - 0.81 0.00 0.66

1st Differences 0.013 0.003 1.64 0.06 0.07 0.01 - - 1.71 0.00 0.64

Panel C: Panel data estimators (preferred estimation methods)

1st Differences + IV 0.014 0.002 1.31 0.04 0.14 0.03 - - 1.45 0.00 0.63

1st Differences + IV 0.012 0.002 1.27 0.04 0.16 0.03 0.12 0.01 1.56 0.00 0.64

1st Differences + IV* 0.009 0.002 1.28 0.04 0.13 0.03 0.15 0.01 1.55 0.00 0.63

*Estimated coefficient on lagged outflow added: Á=.200 (.033)

Number of observations = 14734

_

_

Table 3Matching function estimates

Panel A: 1994-1996

Country TrendStd.Err. β

Std.Err. γ

Std.Err. δ

Std.Err. adjR2 RTS

p-value Nobs

CR -0.112 0.027 0.75 0.16 0.23 0.11 0.26 0.03 0.65 1.24 0.31 2661

SR 0.045 0.058 0.95 0.58 0.17 0.16 0.17 0.05 0.31 1.29 0.60 1292

EG 0.045 0.026 0.91 0.45 -0.08 0.10 0.26 0.06 0.48 1.10 0.85 1211

WG -0.103 0.005 1.27 0.07 0.22 0.04 0.20 0.02 0.67 1.69 0.00 5004

HU n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a.

PL 0.285 0.097 2.60 0.77 0.16 0.12 0.19 0.05 0.71 2.948 0.01 637

Panel B: 1997-2005

Country TrendStd.Err. β

Std.Err. γ

Std.Err. δ

Std.Err. adjR2 RTS

p-value Nobs

CR -0.039 0.008 1.16 0.07 0.51 0.06 0.19 0.02 0.74 1.86 0.00 7770

SR 0.004 0.010 1.51 0.14 0.24 0.05 0.07 0.01 0.49 1.82 0.00 6682

EG -0.021 0.005 1.49 0.11 0.34 0.04 0.31 0.02 0.68 2.14 0.00 3602

WG 0.012 0.002 1.27 0.04 0.16 0.03 0.12 0.01 0.64 1.56 0.00 14734

HU 0.016 0.016 1.55 0.26 0.51 0.11 0.34 0.06 0.28 2.40 0.00 1920

PL 0.022 0.011 0.76 0.11 0.08 0.06 0.19 0.03 0.67 1.03 0.83 1072

Conclusions

West Germany -- rising unemployment and inflow, declining vacancies and relatively efficient matching (high returns to scale) -- outcome most consistent with H1 and H2

Czech Republic similar -- rising unemployment, inflow and outflow, and a declining vacancy rate and high returns to matching, it increasingly gives support to H1 and H2; since CR has increasingly pursued low interest rates and fiscal deficits, the support for H2 implies the presence of negative exogenous demand shocks

East Germany also in line with H1 and H2 -- relatively high unemployment and inflows, a low vacancy rate and very efficient matching (training)

Slovakia -- low returns to scale in matching, and high unemployment, rising inflow rates and a low vacancy rate; loose monetary and fiscal policies and a floating exchange rate. Its outcome is hence consistent with a combination of H1, H2 and H3

Hungary has lowered its unemployment rate to around 8% and it has the highest estimated returns to matching. Given its low vacancy rate relative to inflows, the existing unemployment seems to be consistent with H1 and H2

Conclusions (2)

Overall, our findings suggest that the transition economies contain two broad groups of countries

First group = CR, Hungary and (possibly) East Germany resembles West Germany -- efficient matching and unemployment

appears to be driven by restructuring and low demand for labor The East German case is complex -- major active labor market

policies => in some sense it resembles more the second group, exemplified by Slovakia and Poland

These countries, in addition to restructuring and low demand for labor, appear to suffer from a structural mismatch (i.e., display less efficient matching)

UNEMPLOYMENT IN TRANSITION COUNTRIES (CENTRAL EUROPE): WHY SO HIGH?

Documents

Transcript of UNEMPLOYMENT IN TRANSITION COUNTRIES (CENTRAL EUROPE): WHY SO HIGH?