Abstract
Selected Issues
Skills Mismatch and Active Labor Market Policy in Lithuania1
A. Labor Market in Lithuania
1. Lithuania’s labor market is characterized by a large flexibility, but persistent structural issues. Contrary to other European economies, wages in Lithuania are very sensitive to unemployment and, therefore, an increase in unemployment quickly leads to a reduction in wage growth. During the crisis, real wage growth dropped to -9 percent in 2009 from the pre-crisis peak of 14 percent in 2007.
Lithuania: Unemployment Rate and NAIRU
(In percent)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources: Statistics Lithuania and IMF staff calaulations.Lithuania: Unemployment Rate and NAIRU
(In percent)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources: Statistics Lithuania and IMF staff calaulations.Lithuania: Unemployment Rate and NAIRU
(In percent)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources: Statistics Lithuania and IMF staff calaulations.2. Wage flexibility is underpinned by one of the lowest densities of trade union and employer organization and the rare occurrence of collective bargaining. Thus, wage setting largely happens at the firm level. Real wages and productivity have been traditionally closely linked and temporary deviations have been self-correcting. However, deviations at the sectoral level can be persistent although in the all-important manufacturing sector, wage growth has remained well below productivity growth (IMF Country Report No. 5/139).
Collective Bargaining Coverage
(In percent of employees with The right to bargain, 2016 or latest available)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Source: OECD.Collective Bargaining Coverage
(In percent of employees with The right to bargain, 2016 or latest available)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Source: OECD.Collective Bargaining Coverage
(In percent of employees with The right to bargain, 2016 or latest available)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Source: OECD.3. In contrast, structural unemployment has been traditionally high, although it appears to be gradually falling. Large structural unemployment can have a significant long-term impact on potential growth and, therefore, on employment (IMF Country Report No. 18/185). The rapid decline in the unemployment rate since the crisis occurred at a time of increasing participation rates and has been below pre-crisis estimates of the non-accelerating inflation rate of unemployment (NAIRU) for a few years without inflationary pressures emerging. This suggests that the NAIRU has been gradually decreasing to around 7 percent, as estimated in a multivariate factor (MVF) model. Regardless of the current level of the NAIRU, it seems clear that the current level of unemployment in Lithuania is largely of a structural nature.
4. In addition, the Beveridge curve shifted outward during the crisis, but the quick recovery appears to have shifted the curve back in, at least partially. It also suggests that the efficiency of the labor market deteriorated during the crisis, but has only partially recovered since.
Lithuania: Beveridge Curve
(Percentage of the labor force)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources; Statistics Lithuania; Eurostat; and Haver Analytics.Lithuania: Beveridge Curve
(Percentage of the labor force)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources; Statistics Lithuania; Eurostat; and Haver Analytics.Lithuania: Beveridge Curve
(Percentage of the labor force)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources; Statistics Lithuania; Eurostat; and Haver Analytics.5. Although several factors explain structural
unemployment, this paper will focus on labor market mismatches and policies to address them. This paper analyzes the evolution of skill mismatches in Lithuania since the crisis and presents international experience on active labor market policies (ALMPs) that have been used to combat them.
B. Skills Mismatch in Lithuania
Lithuania has one of the highest skill mismatches in Europe. Aggregate measures of skill mismatches in Lithuania increased sharply during the crisis, but only partially recovered since. Looking at the individual job level, mismatches in Lithuania are in line with other OECD countries.
Lithuania suffers from relative labor shortage for high-skilled workers and surplus of low- and medium-skilled workers. Thus, there are labor shortages in skill-intensive sectors (e.g., ICT and finance), and excess labor in less-skill-intensive sectors (e.g., construction and trade).
Skill Mismatch Index by Country, 2016
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources: Eurostat and IMF staff calculations.Skill Mismatch Index by Country, 2016
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources: Eurostat and IMF staff calculations.Skill Mismatch Index by Country, 2016
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources: Eurostat and IMF staff calculations.
Skill Mismatch Index, 2006–17
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources: Eurostat; and IMF staff calculations.Skill Mismatch Index, 2006–17
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources: Eurostat; and IMF staff calculations.Skill Mismatch Index, 2006–17
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources: Eurostat; and IMF staff calculations.6. Lithuania has shown a sharp rise in skills mismatch for the country as a whole in the aftermath of the crisis. The skills mismatch index used here follows the framework in Estevao and Tsounta (2011), and measures the gap between labor supply and demand by the level of education attainment for the economy as a whole (Box 1). Under this index, Lithuania has one of the highest skills mismatches in Europe. Furthermore, it increased sharply during the crisis driven by the increases in relative shortage of high-skilled workers and/or relative surplus of low- and medium-skilled workers. While partially recovering since, it has remained significantly above pre-crisis levels in line with Latvia. Estonia on the other hand has recovered to levels comparable to pre-crisis. The impact of the crisis has, therefore, been large and persistent despite the flexibility of the labor market to cyclical fluctuations, highlighting once again the structural nature of current unemployment.
7. Lithuania suffers from labor shortage of high-skilled workers and oversupply of medium- and low-skilled workers. Further, the imbalances across skills level have increased since the crisis. The relative skill mismatch is particularly severe in Lithuania compared to other European economies and seems to reflect the legacy of the pre-crisis years where the boom in low-skills sectors, notably construction, has resulted in an oversupply of low-skilled labor. Shortcomings in the education system also play a role in these dynamics. In particular, there is a gap between educational outcomes and the skills demanded by the labor market. Finding workers with the right skills appears to be a significant constraint for over 40 percent of firms.2 High emigration and certain restrictions on non-EU workers, as well as limited participation in life-long learning, also explain the lack of suitable labor in Lithuania (OECD, 2018a).
Lithuania: Skill Mismatch by Skill Level
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources: Eurostat and IMF staff calculations.Lithuania: Skill Mismatch by Skill Level
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources: Eurostat and IMF staff calculations.Lithuania: Skill Mismatch by Skill Level
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources: Eurostat and IMF staff calculations.8. Vacancy rates and wage growth by sectors also suggest an excess supply of lower-skilled workers and shortage of high-skilled ones. Skill-intensive sectors, such as finance and communication, are high in both wage growth and vacancy rates, suggesting shortages due to the insufficient supply of workers with the right skills. On the contrary, less-skill-intensive sectors, such as transports and construction, are high in wage growth, but have low job vacancy rates, implying there is no shortage of workers in these sectors. Low job vacancy rates even at times of high employment growth in these sectors suggest that there are no labor shortages whereas high wage growth is an indication of the tightness of the labor market operating below NAIRU. Similarly, according to the EU, Lithuania’s high skill needs (“shortage occupations”) lie in ICT professionals, engineers, etc., while street salesperson, childcare workers, teachers’ aides, machine operators and transport clerk are in excess supply, “surplus occupations” (EC, 2017).
Lithuania: Vacancy Rates and Wage Growth in Selected Industries
(in percent)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources: Lithuania Statistics and Haver Analytics.Lithuania: Vacancy Rates and Wage Growth in Selected Industries
(in percent)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources: Lithuania Statistics and Haver Analytics.Lithuania: Vacancy Rates and Wage Growth in Selected Industries
(in percent)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources: Lithuania Statistics and Haver Analytics.9. By focusing on labor mismatches at the individual job level rather than at the aggregate, Lithuania ranks better, but remains a low-skill-oriented economy. The OECD’s Skills for Jobs Database presents information on skills shortage and mismatch in field-of-study and qualification (Box 2). According to these indicators, the field-of-study mismatch is slightly higher, at 35 percent, in Lithuania compared to the EU average of 32 percent. Interestingly, Lithuania turns out to be high in overqualification while low in underqualification relative to other EU countries (OECD, 2017). In addition, Lithuania has relatively low knowledge and ability needs, as technology-related skill needs are relatively small, while low-technology skills are more in demand.
Qualification and Field of Study Mismatch by Country, 2016
(Share of employed age 15–64)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Source: OECD Skills for Jobs Database.Qualification and Field of Study Mismatch by Country, 2016
(Share of employed age 15–64)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Source: OECD Skills for Jobs Database.Qualification and Field of Study Mismatch by Country, 2016
(Share of employed age 15–64)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Source: OECD Skills for Jobs Database.Aggregate Skills Mismatch Index1
• Skills mismatch occurs when skills supplied by a worker differ from the ones demanded by his job. Accordingly, a skills mismatch index can be constructed reflecting the difference between the share of employed (skills demanded) and labor force (skills supplied) at each skill level.
• The skill level is measured by data on education attainment from European Labor Force Survey (EULFS), and categorized as low skilled (less than primary, primary and lower secondary education), medium-skilled (upper secondary and post-secondary non-tertiary education), and high-skilled (tertiary education). To account for the differences in the education quality overtime, tertiary education is adjusted for the quality of education based on data from the World Economic Forum.
• Following the framework presented in Estevao and Tsounta (2011), the aggregate skills mismatch index at time t can be constructed as:
where j is the skill level; Sj,t is the percentage of the population with skill level j at time t (skills supplied) and Mj,t is the percentage of employees with skill level j at time t (skills demanded).
1/ The construction of the aggregate skill mismatch index used in this paper benefits from the IMF Country Report No. 18/242.OECD’s Skills for Jobs Indicators1
• The OECD Skills for Jobs Database presents two sets of indicators: the skill needs indicators, including shortage/surplus of skills and occupations; and mismatch indicators, including qualification mismatch and field-of-study mismatch.
• The qualification mismatch index calculates the share of workers in each economy/occupation that are under- or over-qualified to perform a certain job. This is done by computing the modal educational attainment level for each occupation in each country and point in time and use this as a benchmark to measure whether individual worker’s qualifications match the ‘normal’ educational requirement of the occupation. Thus, over-qualification (under-qualification) depicts a situation for which the highest level of education achieved by an individual worker in an occupation is above (below) the modal level for all workers in that occupation.
• Field-of-study mismatch is distinct from qualification or skills mismatch as a worker may be well matched to his/her job in terms of the key information-processing skills possessed (skills match) or educational attainment (qualification match), but not by the type of education and knowledge received during his/her official training and education.
1/OECD (2017), Getting Skills Right: Skills for Jobs Indicators.Impacts of Skills Mismatch on Productivity
10. Skill mismatch negatively affects productivity via misallocation of the workforce and insufficient accumulation of firm-specific knowledge. Individual workers are also affected by skills mismatch through a higher risk of unemployment, lower wages, lower job satisfaction and poor career prospects. A high proportion of Lithuanian firms cite an inadequately educated workforce as a significant obstacle to their operations compared with other Central and Eastern European countries in 2013 (OECD, 2016b).
11. There are several channels through which skills mismatch affects productivity. Compared to well-matched workers, over-qualified workers tend to participate less in training, which hinders accumulation of firm-specific knowledge (Verhaest and Omey, 2006). Skill shortages negatively affect technology adoption and investments in tangible and intangible assets (Forth and Mason, 2006). Under-education can be detrimental to firm productivity as shown by Kampelmann and Rycx (2012) using employer-employee panel data for Belgium. A negative impact of over-qualification on productivity happens through a less efficient allocation of resources, while that between under-qualification and productivity through lower allocative efficiency and within-firm productivity (McGowan and Andrews, 2015).
12. Empirical studies found that countries with high skills mismatch tend to have lower productivity. A study with panel data for 26 European countries including Lithuania over 1995–2016, found that skills mismatch index had a negative and statistically significant impact on total factor productivity (TFP) and labor productivity (IMF Country Report No. 18/242). In line with McGowan and Andrews (2015), given that the negative association remains statistically significant even after controlling for the level of human capital, the study also found that the misallocation that generates skills mismatches played an important role in reducing productivity gains from human capital. Lithuania also follows this general pattern with an increase in skill mismatches in the aftermath of the crisis associated with a significant decline in productivity as indicated by the scatter plot chart on skill mismatch index and TFP growth.
Lithuania: Skill Mismatch and TFP Growth
(In percent)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Source; IMF staff calculations.Lithuania: Skill Mismatch and TFP Growth
(In percent)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Source; IMF staff calculations.Lithuania: Skill Mismatch and TFP Growth
(In percent)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Source; IMF staff calculations.C. Active Labor Market Policy in Lithuania
The Economics of Active Labor Market Policies
13. Active labor market policies (ALMPs) encompass government programs to increase the efficiency of the labor market.3 There are different categories of ALMPs:
Job search assistance (JSA): includes job centers and labor exchanges that try to improve labor market matching by disseminating information on job vacancies, providing the unemployed with interview skills, or assisting in writing a curriculum vitae.
Vocational education and training (VET) schemes: encompass classroom training, on-the-job training, vocational education and apprenticeships, and aim to help the unemployed improve their vocational skills and productivity, and hence increase their employability.
Employment subsidies: encompass wage subsidies, hiring subsidies, business start-up subsidies, and in-work benefits to encourage workers’ labor market attachment and firms’ job creation.
Public works programs (PWP): provide the temporarily unemployed with short-term employment.
JSA and VET address skills mismatch more directly, while employment subsidies and PWP are more supportive of employment.
14. A number of empirical studies established positive effects of ALMPs on employment. For example, using panel data for 15 industrial countries in the 1985–2000, Estevao (2003) estimated that a 1 percentage point increase in ALMP spending relative to GDP would increase the employment rate by 1.9 percentage points in the 1990s. More recently, the Bank of Lithuania (2016) simulated the effect of social policy proposals in Lithuania’s New Social Model in 2015 on unemployment using an open economy vector autoregressive (VAR) model.4 They assumed an increase in ALMP expenditure financed by imposing a 15 percent income tax on unemployment benefits and found a reduction of the unemployment rate as a result.
15. The effectiveness of ALMPs can vary according to the type and duration of the program, and the characteristics of the participants. Card et al. (2009, 2015) assessed the relative effectiveness across active labor market programs based on an extensive meta-analysis on 207 different empirical studies covering 857 programs in total. According to their study:
In terms of the type of program: JSA programs have a relatively short-run impact on employment for disadvantaged participants, whereas job training programs tend to produce better outcomes for the long-term unemployed in the medium-run than the short-run.5 Employment subsidies, particularly in the public sector, tend to have negligible or negative impacts at all time horizons.
In terms of the characteristics of the participant: the impact of ALMPs vary across groups, with a larger impact on female workers and the long-term unemployed, while smaller impact on older and young workers.
In terms of the duration of the program: on average, ALMPs have relatively small impact in the short run, while they have larger positive impact over the medium- and long-run.
ALMPs are relatively more effective during periods of slow growth and higher unemployment.
16. However, ALMPs can also have negative unintended consequences. For example, support for unemployed and disabled workers may result in the substitution of existing employment and medium-skilled workers with job seekers and low-skilled workers (“substitution effect”). Low wage subsidies can also disincentivize unskilled workers from gaining further human capital (“skill acquisition effect”). ALMPs can even reduce job opportunities for participants by signaling their low productivity to employers (“stigmatizing”).6
Active Labor Market Programs in Lithuania
ALMP expenditure in Lithuania is low in terms of investment and participation.
ALMPs in Lithuania do not sufficiently reflect the labor market needs or cyclical conditions and rely largely on European funds. Spending is low in downturns and does not reflect increasing needs. Furthermore, training programs tend to focus on oversupplied skills.
The cost-effectiveness of ALMPs could be enhanced by improving program design based on more systemic program evaluation.
17. Investment and participation in ALMPs have remained relatively low in Lithuania. Spending on ALMPs is small at 0.3 percent of GDP in 2016 relative to OECD countries. Participation is also low with only 3.7 percent of the unemployed having participated in training programs in 2016.
Active Labor Market Program Expenditures
(In percent of GDP)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Source: OECD (Active Labour Market Policies)Active Labor Market Program Expenditures
(In percent of GDP)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Source: OECD (Active Labour Market Policies)Active Labor Market Program Expenditures
(In percent of GDP)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Source: OECD (Active Labour Market Policies)
Participants in Training Programs
(In percent, (hare of total unemployment)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources: OECD; Euionat; Haver Analytics and IMF staff calculation;Participants in Training Programs
(In percent, (hare of total unemployment)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources: OECD; Euionat; Haver Analytics and IMF staff calculation;Participants in Training Programs
(In percent, (hare of total unemployment)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources: OECD; Euionat; Haver Analytics and IMF staff calculation;18. Spending on ALMPs has not been responsive to cyclical conditions and largely relies on European funds. ALMP spending relative to GDP has been fairly constant in Lithuania even in the aftermath of the global financial crisis when unemployment increased sharply. As a result, participation has been one of the lowest in Europe even after the global financial crisis. This is partly explained by the heavy reliance on external source of funding such as the European Social Fund (ESF)—almost two-thirds in 2015. The heavy dependence on EU funding, targeted at specific groups, has resulted in the unique situation of specifying beneficiary groups in the labor code: older workers, long-term unemployed, youth and the disabled. The new labor code, adopted in 2017, has now included unqualified persons to the list of potential beneficiaries.
19. Lithuania’s ALMP measures consist of support for learning, mobility, assisted recruitment, and job creation.7 The new labor code brought about changes in ALMP measures, which include promoting self-employment and internship, encouraging self-education and non-formal adult education, and providing support for mobility. The code abolished some of the programs, such as public works, job rotation and subsidies for individual activity according to business certificate. Employment subsidies have become the main ALMP measure. Unlike other types of program, which have maintained almost the same level of spending relative to GDP, expenditure on training has fluctuated over time and increased recently after some years at very low levels during 2011–14. In addition, despite the introduction of the training voucher scheme in 2012, training programs are centered in curricula for low-skilled tasks, such as driving, construction, cooking and beauty services, which are already in excess supply of labor.
Lithuania: ALMP Expenditures by Program Type
(In percent of GOP)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources: OECD (Active Labour Market Policies) and Haiver Analytics.Lithuania: ALMP Expenditures by Program Type
(In percent of GOP)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources: OECD (Active Labour Market Policies) and Haiver Analytics.Lithuania: ALMP Expenditures by Program Type
(In percent of GOP)
Citation: IMF Staff Country Reports 2019, 253; 10.5089/9781513509242.002.A001
Sources: OECD (Active Labour Market Policies) and Haiver Analytics.D. Policy Implication
20. Lithuania could take some actions to strengthen ALMPs to address skills mismatch:
Spending on ALMPs should be increased with its funding being stabilized and secured with predictable and reliable public resources. In this regard, the program’s scope and dimension should not be linked to European funds and its level should be responsive to cyclical developments in the labor market.
Given the degree of skills upgrade needed in Lithuania, the recent increase in training spending is welcome and should continue. With 40 percent of the unemployed in 2015 having no professional qualification, the 3 percent of participation rate in training is inadequate. Training and re-training programs will directly address the issue of the over-supply of low-skilled workers for which there is no additional demand. Participation rules could be adapted to allow for long programs and those that require expensive equipments to increase the participation of low-skilled, older workers and workers in rural areas.
The importance of employment subsidies in ALMPs should decrease. Currently, they are the main ALMP program in Lithuania. Given the challenges regarding the large stock of low-skill unemployment, and the lower effectiveness of employment subsidies discussed above, these should be concentrated on the most disadvantaged groups, those unlikely to find a job in their absence, and extra resources should be channeled towards training programs.
The current voucher system to fund training program could be more effective by providing information on available programs for recipients to assess which training fits them best. Currently, there is no systemic rating system based on the labor market outcomes of previous participants.
Comprehensive performance evaluation for ALMPs should be conducted on a regular basis and from a long-term perspective. Currently, programs are assessed quarterly by their impact on integration into the labor market, registration at the labor exchange, and direct benefits.8 However, program evaluation should be strengthened to improve program design from a longer-term and broad-based perspectives.
References
Bank of Lithuania, 2016, “Impact of Labor Market Reforms on Lithuania’s Economy,” Lithuanian Economic Review, June 2016.
Brown, A., and J. Koettl, 2015, “Active Labor Market Programs—Employment Gain or Fiscal Drain?” IZA Journal of Labor Economics, Vol. 4, Issue. 12.
Card, D., J. Klube, and A. Weber, 2015, “What Works? A Meta-Analysis of Recent Active Labor Market Program Evaluations,” Ruhr Economic Papers No. 572 (Essen: Rheinisch-Westfaelisches Institute fuer Writscharftsforschung).
Card, D., J. Klube, and A. Weber, 2009, “Active Labor Market Policy Evaluations: A Meta-Analysis,” CESifo Working Paper, No. 2570 (Munich: Center for Economic Studies and Ifo Institute).
EC Skills Panorama Survey, 2017, “Lithuania: Mismatch Priority Occupations,” https://skillspanorama.cedefop.europa.eu/en.
Estevao, M., and E. Tsounta, 2013, “Has the Great Recession Raised U.S. Structural Unemployment?”, IMF Working Paper 11/105 (Washington: International Monetary Fund).
Estevao, M., 2003, “Do ALMP Increase Employment?”, IMF Working Paper 03/234 (Washington: International Monetary Fund).
Forth, J., and G. Mason, 2006, “Do ICT Skill Shortages Hamper Firms’ Performance? Evidence from UK Benchmarking Surveys,” National Institute of Economic and Social Research Discussion Papers, No. 281.
IMF, 2014, “Unemployment and Structural Unemployment in the Baltics”, Baltic Cluster Report 2014
Kampelmann, S., and F. Rycx, 2012, “The Impact of Educational Mismatch on Firm Productivity: Evidence from Liked Panel Data,” I, Vol. 31, Issues 6, 918–931.
Martin, J., 2015, “Activation and Active Labor Market Policies in OECD Countries: Stylized Facts and Evidence on Their Effectiveness,” IZA Journal of Labor Policy, Vol. 4, Issue 4.
McGowan, M., and D. Andrews, 2015, “Labor Market Mismatch and Labor Productivity”, OECD Economics Department Working Papers No. 1209 (Paris: Organization of Economic Cooperation and Development).
McKenzie, M., 2017, “How Effective Are Active Labor Market Policies in Developing Countries? A Critical Review of Recent Evidence,” IZA Discussion Paper No.10655.
OECD, 2018a, “OECD Economic Surveys: Lithuania 2018,” (Paris: Organization of Economic Cooperation and Development).
OECD, 2018b, “OECD Reviews of Labor Market and Social Policies: Lithuania,” (Paris: Organization of Economic Cooperation and Development).
OECD, 2017, “Getting Skills Right for Jobs Indicators,” (Paris: Organization of Economic Cooperation and Development).
OECD, 2016a, “Getting Skills Right: Assessing and Anticipating Changing Skill Needs,” (Paris: Organization of Economic Cooperation and Development).
OECD, 2016b, “OECD Economic Surveys: Lithuania 2016,” (Paris: Organization of Economic Cooperation and Development).
Verhaest, D. and E. Omey, 2006, “The Impact of Overeducation and Its Measurement,” Social Indicators Research, Vol. 77, Issue 3, 419–448.
Prepared by Kanghoon Keah.
According to European Investment Bank (EIB Investment Survey, 2015), 40 percent of Lithuanian firms point out “availability of staff with the right skills” as major obstacles to investment, following “business regulations and taxation”.
In contrast, passive labor market policies (PLMP) aim at providing income replacement during periods of unemployment or job search, which include unemployment insurance and early retirement for labor market reasons.
They take into account four macroeconomic variables (real GDP, unemployment, the international trade over GDP, and the real effective exchange rate) and two institutional variables characterizing Lithuania’s labor market (unemployment benefits and expenditure on ALMP over GDP).
They define the ‘short-run’ as less than a year after the end of the program, the ‘medium-run’ as 1–2 years post program, and the ‘long-run’ as 2 + years.
See Brown and Koettl (2015) for more extensive and detailed discussion on side effects of ALMPs.
Support for learning includes vocational training, recruiting under an apprenticeship contract, internship, and the recognition of non-formal and informal learning competences. Supported employment encompasses subsidized employment, supporting the acquisition of work skills, job rotation, and public works. Support for job creation includes subsidizing of job creation, implementation of local employment initiative projects, support for self-employment, and subsidy for individual activity according to business certificate.
The evaluation in 2017 revealed that vocational training and support for the acquisition of work skills have a larger effect in the longer run (6+ month), while subsidized employment had relatively short duration and the impact disappeared over time. The impacts stabilized one year after the program participation.
Annex I. Calibration of Priors for Key Parameters in the Laubach-Williams Model
The model is calibrated for the economy of Lithuania using quarterly data.
1. The output gap equation (IS curve): The parameters are drawn from a constrained regression of output gap on its lagged values and the lagged values of the interest rate gap.
In this regression, the interest rate gap is proxied by using an HP-filtered real interest rate as the natural rate.
The coefficients on two lagged interest rate gaps are constrained to be equal, smoothing the lagged impact of monetary policy stance on output gap.
The coefficients on the lagged output gaps hardly change between the constrained and unconstrained regressions.
2. The Phillips Curve: The parameters are based on the regression coefficients where current inflation is determined by the lagged values of inflation, output gaps, import inflation and oil price inflation.
The coefficients of eight quarters of lagged inflation sum to 1.
The potential output equation: Potential output is assumed to follow a random walk. Trend year-on-year growth is converted to quarterly growth.
3. The natural rate equation: Natural rate is a function of trend growth and other domestic factors. The coefficient of trend growth represents the marginal utility of consumption or the coefficient in CES utility function. As we represent the utility function in the logarithm form, this parameter is assumed to be 1.
4. The dynamics of the other domestic factors: The priors for the autocorrelation coefficients of each factor are drawn from independent estimation. When other factors are included, the contributions to Z are assumed to be equal and add up to one.
5. The distributions of the error terms: The priors for shock distributions are based on Pescatori and Turunen (2015).
Prior and Posterior Distributions of Key Parameters
Prior and Posterior Distributions of Key Parameters
| No WEO, without Z determinants | No WEO, with Z determinants | With WEO, without Z determinants | With WEO, with Z determinants | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Description | Prior | Mode | Median | 10th | 90th | Mode | Median | 10th | 90th | Mode | Median | 10th | 90th | Mode | Median | 10th | 90th | |
| c | elasticity of substitution | 1.00 | 1.01 | 0.99 | 0.93 | 1.06 | 1.01 | 0.99 | 0.93 | 1.06 | 0.98 | 1.00 | 0.94 | 1.06 | 1.02 | 1.00 | 0.94 | 1.07 |
| d1 | AR coef. on z | 1.00 | 0.91 | 0.99 | 0.70 | 1.34 | 0.92 | 1.03 | 0.72 | 1.38 | 1.00 | 1.02 | 0.73 | 1.35 | 0.88 | 1.02 | 0.74 | 1.38 |
| rhod | d1+d2 (AR coefs. On z) | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 |
| a1 | AR coef. on output gap | 1.19 | 1.22 | 1.26 | 1.15 | 1.37 | 1.22 | 1.22 | 1.12 | 1.33 | 1.25 | 1.20 | 1.12 | 1.29 | 1.19 | 1.16 | 1.07 | 1.27 |
| rhoA | a1 + a2 Coef.on interest rate gap in IS | 0.91 | 0.97 | 0.95 | 0.90 | 0.98 | 0.97 | 0.95 | 0.91 | 0.97 | 0.94 | 0.92 | 0.88 | 0.95 | 0.93 | 0.92 | 0.87 | 0.96 |
| a _r | curve | 0.06 | 0.01 | 0.03 | 0.01 | 0.07 | 0.01 | 0.09 | 0.02 | 0.19 | 0.01 | 0.03 | 0.01 | 0.07 | 0.05 | 0.07 | 0.02 | 0.20 |
| by | Coef. On output gap in PC | 0.11 | 0.11 | 0.11 | 0.10 | 0.13 | 0.11 | 0.11 | 0.10 | 0.13 | 0.11 | 0.11 | 0.10 | 0.13 | 0.11 | 0.11 | 0.10 | 0.13 |
| b1 | AR coef. On inflation in PC AR coef. On further lags in | 0.47 | 0.32 | 0.34 | 0.26 | 0.43 | 0.32 | 0.33 | 0.25 | 0.42 | 0.30 | 0.36 | 0.27 | 0.44 | 0.39 | 0.35 | 0.27 | 0.43 |
| b2 | inflation in PC weight in z equation for REER | 0.18 | 0.38 | 0.50 | 0.36 | 0.62 | 0.38 | 0.50 | 0.35 | 0.63 | 0.53 | 0.52 | 0.36 | 0.66 | 0.53 | 0.54 | 0.39 | 0.66 |
| dc | gap weight in z equation for | 0.25 | 0.26 | 0.25 | 0.13 | 0.39 | 0.32 | 0.22 | 0.10 | 0.35 | ||||||||
| de | trading partner growth weight in z equation for | 0.25 | 0.04 | 1.63 | -0.14 | 2.81 | 2.01 | 1.45 | 0.13 | 2.67 | ||||||||
| dp | spread | 0.25 | 0.76 | 0.17 | -0.92 | 1.22 | -0.73 | 0.12 | -1.07 | 1.33 | ||||||||
| lambdag | variation in trend growth | 0.90 | 0.27 | 0.39 | 0.22 | 0.89 | 0.27 | 0.37 | 0.21 | 0.70 | 10.60 | 8.86 | 6.90 | 11.44 | 11.86 | 11.07 | 9.76 | 12.18 |
| lambdaz | variation in r* trend | 0.20 | 0.18 | 0.26 | 0.15 | 0.71 | 0.18 | 0.25 | 0.14 | 0.42 | 0.21 | 0.25 | 0.15 | 0.71 | 0.17 | 0.26 | 0.15 | 0.49 |
| eps_n | Shock to potential output eq. | 0.53 | 0.20 | 0.99 | 0.26 | 3.26 | 0.20 | 0.60 | 0.22 | 2.51 | 0.14 | 0.19 | 0.14 | 0.26 | 0.17 | 0.19 | 0.14 | 0.26 |
| eps_IS | Shock to IS curve | 0.70 | 2.69 | 2.47 | 1.23 | 2.82 | 2.69 | 2.45 | 1.60 | 2.81 | 2.62 | 2.55 | 2.32 | 2.84 | 2.61 | 2.51 | 2.26 | 2.80 |
| eps_PC | Shock to PC | 0.81 | 3.98 | 3.83 | 3.45 | 4.29 | 3.97 | 3.81 | 3.46 | 4.24 | 3.49 | 3.79 | 3.43 | 4.20 | 3.81 | 3.78 | 3.43 | 4.22 |
| eps_nu | Shock to output gap noise | 0.50 | 1.70 | 1.70 | 1.55 | 1.87 | 1.70 | 1.70 | 1.52 | 1.88 | 0.28 | 0.28 | 0.25 | 0.32 | 0.28 | 0.28 | 0.25 | 0.32 |
| eps_me | Shock to output gap signal | 0.50 | 0.46 | 0.44 | 0.37 | 0.55 | 0.46 | 0.46 | 0.38 | 0.56 | 0.26 | 0.25 | 0.22 | 0.28 | 0.25 | 0.25 | 0.22 | 0.28 |
Prior and Posterior Distributions of Key Parameters
| No WEO, without Z determinants | No WEO, with Z determinants | With WEO, without Z determinants | With WEO, with Z determinants | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Description | Prior | Mode | Median | 10th | 90th | Mode | Median | 10th | 90th | Mode | Median | 10th | 90th | Mode | Median | 10th | 90th | |
| c | elasticity of substitution | 1.00 | 1.01 | 0.99 | 0.93 | 1.06 | 1.01 | 0.99 | 0.93 | 1.06 | 0.98 | 1.00 | 0.94 | 1.06 | 1.02 | 1.00 | 0.94 | 1.07 |
| d1 | AR coef. on z | 1.00 | 0.91 | 0.99 | 0.70 | 1.34 | 0.92 | 1.03 | 0.72 | 1.38 | 1.00 | 1.02 | 0.73 | 1.35 | 0.88 | 1.02 | 0.74 | 1.38 |
| rhod | d1+d2 (AR coefs. On z) | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 | 0.99 |
| a1 | AR coef. on output gap | 1.19 | 1.22 | 1.26 | 1.15 | 1.37 | 1.22 | 1.22 | 1.12 | 1.33 | 1.25 | 1.20 | 1.12 | 1.29 | 1.19 | 1.16 | 1.07 | 1.27 |
| rhoA | a1 + a2 Coef.on interest rate gap in IS | 0.91 | 0.97 | 0.95 | 0.90 | 0.98 | 0.97 | 0.95 | 0.91 | 0.97 | 0.94 | 0.92 | 0.88 | 0.95 | 0.93 | 0.92 | 0.87 | 0.96 |
| a _r | curve | 0.06 | 0.01 | 0.03 | 0.01 | 0.07 | 0.01 | 0.09 | 0.02 | 0.19 | 0.01 | 0.03 | 0.01 | 0.07 | 0.05 | 0.07 | 0.02 | 0.20 |
| by | Coef. On output gap in PC | 0.11 | 0.11 | 0.11 | 0.10 | 0.13 | 0.11 | 0.11 | 0.10 | 0.13 | 0.11 | 0.11 | 0.10 | 0.13 | 0.11 | 0.11 | 0.10 | 0.13 |
| b1 | AR coef. On inflation in PC AR coef. On further lags in | 0.47 | 0.32 | 0.34 | 0.26 | 0.43 | 0.32 | 0.33 | 0.25 | 0.42 | 0.30 | 0.36 | 0.27 | 0.44 | 0.39 | 0.35 | 0.27 | 0.43 |
| b2 | inflation in PC weight in z equation for REER | 0.18 | 0.38 | 0.50 | 0.36 | 0.62 | 0.38 | 0.50 | 0.35 | 0.63 | 0.53 | 0.52 | 0.36 | 0.66 | 0.53 | 0.54 | 0.39 | 0.66 |
| dc | gap weight in z equation for | 0.25 | 0.26 | 0.25 | 0.13 | 0.39 | 0.32 | 0.22 | 0.10 | 0.35 | ||||||||
| de | trading partner growth weight in z equation for | 0.25 | 0.04 | 1.63 | -0.14 | 2.81 | 2.01 | 1.45 | 0.13 | 2.67 | ||||||||
| dp | spread | 0.25 | 0.76 | 0.17 | -0.92 | 1.22 | -0.73 | 0.12 | -1.07 | 1.33 | ||||||||
| lambdag | variation in trend growth | 0.90 | 0.27 | 0.39 | 0.22 | 0.89 | 0.27 | 0.37 | 0.21 | 0.70 | 10.60 | 8.86 | 6.90 | 11.44 | 11.86 | 11.07 | 9.76 | 12.18 |
| lambdaz | variation in r* trend | 0.20 | 0.18 | 0.26 | 0.15 | 0.71 | 0.18 | 0.25 | 0.14 | 0.42 | 0.21 | 0.25 | 0.15 | 0.71 | 0.17 | 0.26 | 0.15 | 0.49 |
| eps_n | Shock to potential output eq. | 0.53 | 0.20 | 0.99 | 0.26 | 3.26 | 0.20 | 0.60 | 0.22 | 2.51 | 0.14 | 0.19 | 0.14 | 0.26 | 0.17 | 0.19 | 0.14 | 0.26 |
| eps_IS | Shock to IS curve | 0.70 | 2.69 | 2.47 | 1.23 | 2.82 | 2.69 | 2.45 | 1.60 | 2.81 | 2.62 | 2.55 | 2.32 | 2.84 | 2.61 | 2.51 | 2.26 | 2.80 |
| eps_PC | Shock to PC | 0.81 | 3.98 | 3.83 | 3.45 | 4.29 | 3.97 | 3.81 | 3.46 | 4.24 | 3.49 | 3.79 | 3.43 | 4.20 | 3.81 | 3.78 | 3.43 | 4.22 |
| eps_nu | Shock to output gap noise | 0.50 | 1.70 | 1.70 | 1.55 | 1.87 | 1.70 | 1.70 | 1.52 | 1.88 | 0.28 | 0.28 | 0.25 | 0.32 | 0.28 | 0.28 | 0.25 | 0.32 |
| eps_me | Shock to output gap signal | 0.50 | 0.46 | 0.44 | 0.37 | 0.55 | 0.46 | 0.46 | 0.38 | 0.56 | 0.26 | 0.25 | 0.22 | 0.28 | 0.25 | 0.25 | 0.22 | 0.28 |
References
Arena, Marco and others, 2019, “Natural Interest Rates in Europe”, forthcoming IMF Working Paper.
Barsky, R, A. Justiniano, and L. Melosi, 2014, “The Natural Rate of Interest and its Usefulness for Monetary Policy,” American Economic Review: Papers & Proceedings, Vol. 104, No. 5, pp. 37–43.
Borio, C., M. Drehmann, and K. Tsatsaronis, 2009, “Financial Stress: What It Is, How Can It Be Measured, and Why Does It Matter?” Economic Review Second Quarter (Kansas City, KS: Kansas City Federal Reserve Bank of Kansas City).
Christensen, J.H.E., F. Diebold, and Glen Rudebusch, 2011, “The Affine Arbitrage-Free Class of Nelson-Siegel Term Structure Models,” Journal of Econometrics, Vol 164, No. 1, pp. 4–20.
Del Negro, and others, 2018, “Global Trends in Interest Rates,” Federal Reserve Bank of New York Staff Report No. 866 (New York).
European Department Financial Sector Working Group, 2019, “Financial Conditions in Europe”, forthcoming IMF Working Paper.
Galí, Jordi, 2008, Monetary Policy, Inflation and the Business Cycle: An Introduction to the New Keynesian Framework (New Jersey: Princeton University Press).
Hatzius, Jan, and others, 2010, “Financial Conditions Indexes: A Fresh Look After the Financial Crisis,” NBER Working Paper No. 16150 (Cambridge, MA: National Bureau of Economic Research).
Holston, K., T. Laubach, and J. C. Williams, 2017, “Measuring the Natural Rate of Interest: International Trends and Determinants,” Journal of International Economics, Vol. 108, Supplement 1, pp. S39–S75.
Kiley, Michael T, 2015, “What Can the Data Tell Us About the Equilibrium Real Interest Rate?” FEDS Working Paper, No. 2015–077 (Washington, DC: Federal Reserve Board).
Krippner, Leo, 2015, Zero Lower Bound Term Structure Modeling: A Practitioner’s Guide (New York: Palgrave-Macmillan).
Laubach, Thomas and John C. Williams, 2003, “Measuring the Natural Rate of Interest,” Review of Economics and Statistics, Vol. 85, No. 4, pp. 1063–70.
Lunsford, Kurt G. and Kenneth D. West, 2018, “Some Evidence on Secular Drivers of U.S. Safe Real Rates,” NBER Working Paper No. w25288 (Cambridge, MA: National Bureau of Economic Research).
Pescatori, Andrea and Jarkko Turunen, 2015, “Lower for Longer; Neutral Rates in the United States,” IMF Working Papers 15/135 (Washington: International Monetary Fund).
Rachel, Lukasz and Thomas D. Smith T, 2017, “Secular Drivers of the Global Real Interest Rate,” International Journal of Central Banking, Vol 3. No 13.
Schüler, Yves S., Paul P. Hiebert, and Tuomas A. Peltonen, 2017, “Coherent Financial Cycles for G7 Countries,” ESRB Working Paper Series No. 43 (Frankfurt, Germany: European Systemic Risk Board).
Summers, Lawrence, 2014, “U.S. Economic Prospects: Secular Stagnation, Hysteresis, and the Zero Lower Bound,” Business Economics, Vol 49, Issue 2.
Wicksell, Knut, 1936, Interest and Prices, originally published in 1898, translated by R.F. Kahn (London: Macmillan Publishers).
Woodford, Michael, 2003, Interest and Prices: Foundations of a Theory of Monetary Policy (New Jersey: Princeton University Press).
Wu, Jing Cynthia and Fan Dora Xia, 2016, “Measuring the Macroeconomic Impact of Monetary Policy at the Zero Lower Bound,” Journal of Money, Credit, and Banking, Vol. 48, Issue 2–3, pp. 253–291.
Prepared by Vina Nguyen. This paper draws the methodologies from two forthcoming IMF Working Papers. “Natural Interest Rates in Europe” and “Financial Conditions in Europe.”
Studies done by Bank of England, Canadian Central Bank, and the Fed Chairman’s mentioning of the US interest rate being “a long way” and “just below” the neutral rate.
The case in which Central Banks do not face the trade-off between stabilizing inflation and the output gap.
Rachel and Smith (2017) suggest that because households’ value smooth consumption patterns, every percentage point fall in trend productivity growth could cause equilibrium real rates to fall by up to twice as much.
We use the REER gap, measured as the deviation of REER from its long-term average.
Nominal interest rate minus inflation expectations.
The PLS algorithm cannot produce a result with missing data, and some financial indicators only became available late in the sample. The PLS-Chained Index methodology produces an index with all the available data at that point in time and as new indicators become available, the index is extended to incorporate new information.
