Restructuring and labour market performance in CEE and CIS transition countries.
The unique setting of transition economies (the sheer scale of structural change, similar initial conditions, different policy approaches and diverging transition paths chosen by the CEE and CIS countries) allows researchers to get a rare insight into the functioning of labour markets, effects and efficiency of alternative LM policies.
In this paper we will summarize the existing literature and analyze available empirical data to answer the following questions:
· Why and to what extent was labour market restructuring necessary in transition countries? What were the starting point and anticipated result of structural change?
· How can we explain labour reallocation from declining to expanding sectors using economic models and institutional knowledge? What was the function of labour market policies and institutions in this process?
· How did labour markets perform across the region? Are economic models successful in explaining real market evolution and the differences in CEE and CIS transition paths?
1. Initial conditions and the importance of labour market restructuring.
Post-soviet transition, aiming to reverse many decades of history under communism and substitute the soviet brand of socialism and planned economy with the Western model of democracy and free market, presented many daunting challenges. Labour market restructuring was one of the most critical tasks for the new governments, due to the bleak initial conditions, shocks resulting from liberalization and stabilization policies, and vital importance of healthy labour markets, not only for economic growth, but for the general living standards of citizens and their support of reform.
Decades under planned economy left a legacy of labour market distortions and inefficiency:
* Labour market was ruled by supply rather than demand – full employment was achieved by means of overstaffed enterprises and low productivity (labour hoarding in transition economies amounted to around 20% of total employment – Mickiewicz 2005);
* Salary was a means of wealth distribution and not related to productivity – resulting in lack of incentive, inefficiency and monetary overhang;
* Social services were provided by the firm rather than specialized state agencies – thus, a brand-new social safety net had to be developed promptly;
* Allocation system was based on political motives rather than economic reasoning and availability of resources
* This led to establishment of an nflated and mostly inefficient industrial sector, exaggerated share of population employed in agriculture, underdeveloped service sector.
Sources: Haltiwanger et al 2003, de la Camara 1997, Aslund 2002
These problems started showing long before the beginning of transition, resulting in sluggish growth, frequent shortages and inability to compete with the West.
As transition started, labour markets experienced massive shocks due to price liberalization, macroeconomic stabilization and the collapse of old trading system. The collapse of Comecon put an end to subsidized supplies and guaranteed markets for industrial product. Price liberalisation rendered many firms and production processes unviable, as raw materials would cost more than the final product. Strict fiscal discipline and hardening budget meant that unproductive firms were not bailed out and unproductive workers had to be released, leading to massive cutbacks and closures. Massive job destruction at the onset of transition caused a sharp decrease in labour market participation, employment, wages and productivity.
Employment and productivity determine future output and growth, while high unemployment has negative social, fiscal and political implications. Thus, initial failure of labour market reform could lead to prolonged economic stagnation, deterioration of the standard of living or fiscal instability, creating opposition to and possible reversal of the reform process.
2. Models of labour market flows and policy tradeoffs in transition
The objective of labour market restructuring was twofold:
* reallocating nearly all labour force from the old (state-owned, mostly large industrial enterprises) to the more productive new sectors (private, small or medium enterprises, previously neglected service sector) or unemployed / out of labour force state;
* finding an optimal balance of economic efficiency (hard budget constraint, productivity), social equity (unemployment rate, wages and non-employment benefits) and fiscal discipline (sustainability of social payments).
Transition is associated with massive job creation, job destruction, and transitional flows between employment, unemployment and inactivity. We will first look at the evolution of these flows in transition countries, and then present theoretical models explaining this evolution, policy tradeoffs involved and reasons behind cross-country variations in labour market performance.
Boeri and Terrell (2002) summarize statistical evidence of transition flows and list probabilities of each flow in some transition countries and the USA (see Annex 1)
There were three possible directions of labour reallocation from the state sector:
* Into private employment via privatization or voluntary transfer – the most common path. Most job vacancies were filled by employed jobseekers rather than the unemployed.
* Out of labour force – very common due to initially generous early retirement, social assistance and disability schemes and “natural” adjustment of participation rate according to preferences (Mickiewicz 2005).
* Into unemployment, via layoffs or voluntary quits – probability of becoming unemployed was not very high by Western standards, due to predominance of other paths (job-job or job-OLF) in CEE (Boeri and Terrell 2002) and due to persisting soft budget constraint, insider control and reluctant labour shedding in the CIS (Mickiewicz 2005).
For the unemployed, there were several possibilities:
* Into employment – probability of finding a new job was quite low by Western standards, because job creation in the private sector did not instantly catch up with job destruction by public enterprises (Haltiwanger et al 2003), and most vacancies were filled by employed jobseekers (Boeri 1997) => stagnant pools and increasing share of long-term unemployment
* Out of labour force – discouragement in finding a job due to congestion in the matching process, imperfect information and limited mobility (de la Camara 1997), (especially widespread among the disadvantaged groups: youngest and older jobseekers, women, low qualified workers, ethnic minorities, residents of rural areas/areas specializing in heavy industry, etc.); or higher attractiveness of pensions/social assistance compared to unemployment benefits (Bruno 2006).
Table 1. Labour market flows in transition explained
From state E
State enterprise closures (Aghion, Blanchard 1994); hard budget constraints
Getting rid of overemployment =>
Downward pressure on wages => facilitates creation of private E
Payroll taxes to finance UE benefits => ↑labour costs; hindering job creation at high UE;
social cost => opposition to restructuring
Wages vs UE benefits (Boeri, Terrell 2002); chances of finding a job
To private E
stimulating business start-ups
Higher productivity and
Most flows within rather than between sectors;
Early retirement, social assistance & disability schemes (Bruno 2005); ↓childcare provisions
and social pressure
Higher fiscal pressure,
lower long-term growth potential
To private E
Stimulating private job creation; ALMP (Boeri 97)
Reducing fiscal burden and social pressure
The difference between values of pensions and UE benefits (Bruno 2006)
Reducing social pressure and probability of reform reversal
Increasing fiscal burden (dependency ratio),
↓ growth potential
Optimal Speed of Transition models are most widely used to explain labour reallocation from low- to high-productivity sectors. Aghion and Blanchard in their 1994 paper develop a model that explains flows from state to private employment via an unemployment spell, uses labour shedding (job destruction in the state sector) as a policy determinant, and defines the optimal speed of restructuring which maximizes private job creation.
“After initial adjustment, unemployment dynamics depend on the speed of restructuring and private job creation” (this simplified model assumes that job destruction in state sector is transformed into unemployment, ignoring attrition, voluntary leaves and direct flows into private employment). Unemployment affects private job creation through wages (downward pressure, positive effect while unemployment is low) and taxes (upward pressure, negative effect when unemployment is high).
s – job destruction = flow into unemployment =speed of restructuring
H – private job creation
U – unemployment
There is a maximum speed of restructuring/ labour shedding at which private job creation is optimal. If s > H (max), transition fails. If s < H (max), there are two possible equilibria: A (lower, stable) and B (higher, unstable). If initial
U0 > UB, then H < s, leading to a continued increase of unemployment and possible collapse of the private sector (point C). If U0 < UB, then H > s (job creation exceeds job destruction) until unemployment reaches low equilibrium. The model with endogenous restructuring shows that, until unemployment reaches equilibrium level, restructuring slows down because opposition is too high.
While Aghion-Blanchard model succeeds in capturing main policy trade-offs between the speed of transition, short-term unemployment and job creation, it has serious shortcomings that were addressed and improved upon in other literature.
* It only looks at flows between state and private employment via an unemployment spell, which represent just 15% of labour market flows in transition (Boeri 2001). Outflows from employment are assumed to be under government control, overlooking voluntary leaves.
* It does not give much insight into the role of LM policies and institutions, such as non-employment benefits, minimum wage, employment regulations, active labour market policy.
Boeri (2001) introduces dynamics of labour supply into the model, which now captures direct flows between jobs, voluntary leaves and flows out of labour force (75% of gross flows). He uses non-employment benefits as the main determinant of utility value and choice between employment, unemployment and inactivity. In the low benefit scenario (0.25 wage replacement rate) unemploy-ment and inactivity reach stable state similar to developed OECD economies within 8 years. If non-employment benefits are fixed at a high level (0.35 wage replacement rate), after initial increase in job creation the economy follows “low-participation high-labour taxation path” of declining output. This can be remedied by a reform reducing the benefits, with a subsequent increase in unemployment (as “more non-employed people start looking for jobs”). These results are consistent with the experience of Baltic states (low benefit scenario) and Visegrad countries (reform scenario).
Bruno (2006) also looks at the choice between unemployment and inactivity, and uses the spread between rates of pensions and unemployment benefits to explain the shrinking of labour force. Due to aggregate uncertainty optimal speed of transition is lower
Garibaldi and Brixiova (1997) find that “higher unemployment benefits speed up job destruction in the state sector and private job creation at the early stages of the transition, but they increase unemployment in the long run” and that “higher minimum wages can theoretically speed up the reallocation process without affecting the long run equilibrium”.
“Recently, Boeri and Terrell (2002) have compiled more evidence, from all of the CEE and some of the CIS countries, which supports the argument that the different levels of expenditure and structures of non-employment benefit between these two regions explain their different reallocation patterns.”
Boeri (1997) looks into the impact of active labour market policies
* Aggregate uncertainty and reform reversal – more complex (Bruno)
* Private job creation: effect of access to capital and property rights
In the next section we will see how accurate are these models in explaining the differences in labour market performance in CEE and CIS countries.
3. Analyzing empirical data: divergent transition paths of CEE and CIS countries
Differences in labour market performance
U-shaped: Sharp decline at the beginning of transition, quick recovery and return to growth.
L-shaped: deep transitional recession that lasted a decade, first signs of recovery in 2000 (Annex 3).
Job creation/ destruction
Early in transition job destruction excee-ded job creation; later these processes balanced each other out. Most CEE countries achieved job reallocation rates similar to those in mature economies by 1995 (Haltiwanger et al 2002, Faggio and Konings 2001, Jurajda and Terrell 2002).
Job destruction dominated job creation long into transition, first signs of reversal appeared in 2000s, in some countries net employment growth rate is still negative (Haltiwanger et al 2002, Acquisti and Lehmann 2000, Brown and Earle 2002).
Dropped significantly (employment-output elasticity: from 0.59 in the Balkans to 0.80 in Visegrad states). Quantity adjustment Employment decreased only slightly in spite of a huge drop in production (Employment-output elasticity in Russia – only 0.17, Annex 2).
Unemploy-ment Stagnant pool and persistence of long-term unemployment
Higher unemployment turnover rate, lower proportion of long-term UE.
Initially decreased, but stabilized at about 75% of 1989 level within two years, then started growing.
Fell significantly and remained at 50% of 1989 level for nearly a decade (Annex 3). Price adjustment
Rapid structural change and reallocation of jobs.
Slower restructuring because of reluctant labour shedding (Annex 2).
Decreased slightly, but soon returned to growth (GDP growth > employment dynamics - Annex 3). Productivity increased by 30-100% and is gradually converging to OECD levels (Annex 4).
Decreased dramatically (drop in GDP >> decrease in employment). By 2007 only Russia reached 1990 productivity levels, which is only 30% of the US indicator (Slow recovery due to slow restructuring)
CEE transition path seems to be more successful: although it entails higher unemployment rates, it is compensated by faster restructuring, quick economic recovery, higher productivity and better standard of living with similar or lower levels of inequality than in the CIS.
Aghion-Blanchard model focuses on the differences in pace of restructuring (approximating it to privatization) and sees the decline in employment as a function under government control (as if there were no voluntary outflows). It fails to explain why Russia, which initially launched a big bang privatization program, experienced lower unemployment than, for example, gradual reformers Hungary and Slovakia. We need to look at other determinants of outflows from employment, such as minimum wages and unemployment benefits (Garibaldi and Brixiova 1997, Boeri and Terrell 2002, Haltiwanger et al 2003).
Enforcing minimum wage leads to so-called “quantity adjustment”: greater decrease in employment (to E' rather than E*), smaller decrease in wages and high unemployment.
If wages are fully flexible, the anticipated outcome of a shock is price adjustment: greater decline in wages, smaller decline in employment, with only natural unemployment.
Boeri and Terrell (2002) suggest that, while minimum wage was not universally enforced, generous unemployment benefits in CEE at the beginning of transition provided the de facto minimum wage benchmark.
Baltic states seem to be an exception from this rule, as they introduced very low minimum wage and unemployment benefits and still experienced labour market performance similar to the CEE states. That might be due to following two factors:
Quality of market institutions
Had twofold influence: first, membership criteria provided ready-to-implement institutional models and reform bench-marks, while membership prospects ensured popular support of reform;
second, access to the Single Market, EU aid and investment facilitated swift GDP recovery and job creation
CIS countries loosely followed IMF guidelines (Washington consensus) which did not garner much public support and did not put emphasis on institutions, governance or rule of law (Dabrowski 2007). Market access, foreign investment and development aid were all inferior to CEE levels.
Distortions inherent to planned economy and shocks resulting from liberalization and stabilization policies led to massive changes on the labour market in transition. Transition was associated with large-scale job destruction, job creation and transitional flows of workers between old state employment, new private employment and out of labour force state towards a new equilibrium.
Optimal speed of transition models tell us that if unemployment is too small, hiring workers is too costly, which hinders job creation. Yet if unemployment is too high, employment taxes levied to finance unemployment benefits increase business costs, which also negatively affects job creation. High unemployment has additional social and political costs, as it causes poverty, increases income inequality and hinders popular support of reform. The state can control outflows from employment directly (by speeding up or slowing down privatisation, state enterprise closures and layoffs) or indirectly, by modifying the rate of unemployment benefits. Optimal rate of unemployment benefits is high enough to “overcome workers' resistance to staff reductions and allow managers to cut down employment”, but not too high, so that the fiscal burden does not hider job creation (Boeri 2001).
At the beginning of transition unemployment increases, as labour hoarding becomes unsustainable and it takes a while for job creation to catch up with job destruction. One way of reducing unemployment at this stage is stimulating flows out of labour force, but if rate of inactivity becomes too high, it generates a heavy fiscal burden on the employed population, reduces the speed of transition and hinders the country's long-term growth prospects. Policy instrument regulating these flows is the spread between rates of pensions and unemployment benefits. The higher it is, the more workers prefer inactivity to unemployment.
As job creation catches up with job destruction, most workers are moving from old to new jobs directly, while the pool of unemployment remains stagnant. Active labour market policies are designed to improve the jobseeker-vacancy matching process and increase unemployment turnover, but they may be too costly. Large scale ALMP are only efficient if unemployment is relatively low and passive LMP are tightened. If unemployment reaches double digits, “selective and narrowly targeted ALMP” are efficient in preventing most affected groups from leaving the labour force. (Boeri 1997).
Non-labour market policies stimulating job creation (building healthy market institutions, improving business climate, infrastructure development, attracting foreign investment, etc.) were also
Looking at the empirical data, we can trace two distinct paths of transition: quantity adjustment (CEE path) and price adjustment (CIS path). The difference cannot be explained by initial conditions and speed of transition alone. One viable explanation is based on the differences in unemployment benefits, which acted as a de facto minimum wage. More generous benefits in CEE led to more labour shedding, faster restructuring and higher productivity. Very low benefits in the CIS encouraged labour hoarding, thus employment declined slightly while wages and productivity plummeted.
Yet the Baltic states with their low benefits managed to follow the path similar to the CEE, but with lower unemployment. They boosted job destruction by quick and definitive privatization, and stimulated job creation by strengthening market institutions, improving business climate and opening up to foreign investors. Meanwhile, CIS countries were plagued by corruption and rent-seeking, maintained insider control of enterprises and had unsafe environment for investors.
The reasons between such different policy choices and quality of institutions lie in the realm of political economy. The most obvious one seems to be stronger desire to “return to Europe” in the CEE and Baltic states, major impact of EU accession conditionality on the quality of institutions.
Annex 1. Transition Probabilities for Six Transition Economies and the U.S.
P(eu): Probability (Employment -> Unemployment)
P(ou): Probability (Out Labour Force -> Unemployment)
P(eo) : Probability (Employment -> Out Labour Force)
P(oe): Probability (Out Labour Force -> Employment)
P(ue): Probability (Unemployment -> Employment)
P(uo): Probability (Unemployment -> Out Labour Force)
Source: Boeri and Terrell 2002
Annex 2. Structural Change, Labour Mobility and Employment-Output Elasticity
(average yearly measures)
Source: Boeri 2001
Structure of employment in 2005: still a long way from convergence with EU-15
Source: LABORSTA, International Labour Organization
Annex 3. Price versus quantity adjustment
Source: Boeri and Terrell 2002