## 1ltuea2019003

## Source details

**Canonical URL:** [1ltuea2019003](https://www.imf.org/-/media/files/publications/cr/2019/1ltuea2019003.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2019/1ltuea2019003.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2019/1ltuea2019003.pdf.json)

---

### Labor market in Lithuania: characteristics and trends
- Labor market characterized by "large flexibility, but persistent structural issues."
- Wages are very sensitive to unemployment: real wage growth fell to "-9 percent in 2009" from a pre-crisis peak of "14 percent in 2007."
- Low density of trade union and employer organization; wage setting largely at firm level.
- Structural unemployment historically high but "appears to be gradually falling."
- NAIRU estimated to have been "gradually decreasing to around 7 percent" per a multivariate factor (MVF) model.
- Beveridge curve: shifted outward during the crisis and partially shifted back in during recovery, indicating deterioration in labor market efficiency during the crisis with only partial recovery.
- Paper focus: evolution of skill mismatches since the crisis and international experience on active labor market policies (ALMPs).

### Skills mismatch: measures, patterns, and sectoral evidence
- Aggregate measures indicate Lithuania "has one of the highest skill mismatches in Europe."
- Skills mismatch index (Estevao and Tsounta 2011 framework) increased sharply during the crisis with only partial recovery since.
- Relative shortages and surpluses:
  - Relative labor shortage for high-skilled workers; surplus of low- and medium-skilled workers.
  - Skill-intensive sectors with shortages: ICT and finance.
  - Less-skill-intensive sectors with excess labor: construction and trade.
- Causes cited: legacy pre-crisis boom in low-skills sectors (notably construction), shortcomings in the education system, high emigration, restrictions on non-EU workers, limited participation in life-long learning (OECD, 2018a).
- Firm constraints: finding workers with the right skills a significant constraint for "over 40 percent of firms." (EIB Investment Survey, 2015)
- Vacancy rates and wage growth patterns:
  - Skill-intensive sectors (finance, communication): high wage growth and high vacancy rates → shortages.
  - Less-skill-intensive sectors (transports, construction): high wage growth but low job vacancy rates → no labor shortages; high wage growth signals labor market tightness operating below NAIRU.
- EU occupation classification:
  - Shortage occupations: ICT professionals, engineers, etc. (EC, 2017).
  - Surplus occupations: street salesperson, childcare workers, teachers’ aides, machine operators, transport clerk.
- OECD indicators (Box 2):
  - Field-of-study mismatch in Lithuania: "35 percent" (EU average "32 percent").
  - Lithuania high in overqualification and low in underqualification relative to other EU countries (OECD, 2017).
  - Demand composition: relatively low knowledge and ability needs; technology-related skill needs relatively small; low-technology skills more in demand.

### Aggregate Skills Mismatch Index (Box 1 summary)
- Data and categories:
  - Uses EULFS data with education categories: Low skilled, Medium-skilled, High-skilled (tertiary adjusted for quality using World Economic Forum data).
- Formula:
  - Aggregate Skills Mismatch_t = sum_{j=1}^3 (S_{j,t} - M_{j,t})^2
    - S_{j,t} = percentage of population with skill level j at time t (skills supplied)
    - M_{j,t} = percentage of employees with skill level j at time t (skills demanded)

### Impacts of skills mismatch on productivity
- Channels of negative impact:
  - Misallocation of the workforce.
  - Insufficient accumulation of firm-specific knowledge.
  - Individual consequences: higher unemployment risk, lower wages, lower job satisfaction, poorer career prospects.
- Evidence and mechanisms cited:
  - Over-qualified workers participate less in training → hinders firm-specific knowledge (Verhaest and Omey, 2006).
  - Skill shortages negatively affect technology adoption and investment (Forth and Mason, 2006).
  - Under-education detrimental to firm productivity (Kampelmann and Rycx, 2012).
  - Over-qualification reduces productivity via less efficient resource allocation; under-qualification reduces allocative efficiency and within-firm productivity (McGowan and Andrews, 2015).
- Empirical finding: panel data for 26 European countries (1995–2016) found skills mismatch index had a negative and statistically significant impact on total factor productivity (TFP) and labor productivity (IMF Country Report No. 18/242).
- Lithuania-specific: increase in skill mismatches after the crisis associated with a significant decline in productivity.

### Active Labor Market Policy (ALMP): definitions, evidence, and Lithuanian practice
- ALMP categories:
  - Job search assistance (JSA): job centers, labor exchanges, CV/interview help.
  - Vocational education and training (VET) schemes: classroom training, on-the-job training, apprenticeships.
  - Employment subsidies: wage subsidies, hiring subsidies, business start-up subsidies, in-work benefits.
  - Public works programs (PWP): short-term employment for temporarily unemployed.
- Effectiveness evidence (Card et al. meta-analysis of 207 studies covering 857 programs):
  - JSA: relatively short-run impact for disadvantaged participants.
  - Job training programs: better medium-run outcomes for long-term unemployed.
  - Employment subsidies, especially in public sector: negligible or negative impacts at all horizons.
  - Impact varies by participant: larger for female workers and long-term unemployed; smaller for older and young workers.
  - Duration: ALMPs small short-run impact; larger positive impact over medium- and long-run.
  - More effective during slow growth and higher unemployment.
- Potential negative unintended consequences:
  - Substitution effect: displacement of existing workers by program participants.
  - Skill acquisition effect: low wage subsidies disincentivize further human capital acquisition.
  - Stigmatizing: programs signaling low productivity can reduce participants' job opportunities (Brown and Koettl, 2015).

Active Labor Market Programs in Lithuania (current state)
- ALMP expenditure low: "0.3 percent of GDP in 2016" relative to OECD countries.
- Participation low: "3.7 percent of the unemployed having participated in training programs in 2016."
- ALMP spending not responsive to cyclical conditions; heavily reliant on European funds (almost two-thirds in 2015).
- Dependence on EU funding led to specified beneficiary groups in labor code: older workers, long-term unemployed, youth, disabled. New labor code (2017) added unqualified persons.
- ALMP measures: support for learning, mobility, assisted recruitment, job creation.
- New labor code changes: promote self-employment and internship, encourage self-education and non-formal adult education, provide support for mobility; abolished some programs (public works, job rotation, subsidies for individual activity according to business certificate).
- Employment subsidies have become the main ALMP measure.
- Training expenditure fluctuated; increased recently after very low levels during 2011–14.
- Training programs tend to focus on low-skilled curricula (driving, construction, cooking, beauty services) despite oversupply in these areas.
- The 2012 training voucher scheme exists but training programs remain centered on oversupplied skills.

### Policy implications and recommendations on ALMPs
- Increase ALMP spending and secure predictable, reliable public resources; funding should not be linked to European funds and should be responsive to cyclical labor market developments.
- Continue recent increase in training spending given the degree of skills upgrade needed.
  - "40 percent of the unemployed in 2015 having no professional qualification" while the "3 percent of participation rate in training is inadequate."
  - Training and re-training should target over-supply of low-skilled workers.
  - Participation rules could be adapted to allow long programs and those requiring expensive equipment to increase participation of low-skilled, older workers and rural workers.
- Reduce the relative importance of employment subsidies in ALMPs:
  - Concentrate subsidies on the most disadvantaged groups unlikely to find a job without support.
  - Channel extra resources toward training programs.
- Improve the training voucher system by providing information on available programs so recipients can assess fit; establish a systemic rating system based on labor market outcomes of previous participants.
- Strengthen comprehensive performance evaluation of ALMPs on a regular, long-term basis:
  - Current evaluation is quarterly and focuses on integration into labor market, registration at labor exchange, and direct benefits.
  - Program evaluation should be strengthened to improve design from a longer-term and broad-based perspective.
  - 2017 evaluation revealed vocational training and support for acquisition of work skills have larger effects in the longer run (6+ month), while subsidized employment had relatively short-duration impacts that disappeared over time; impacts stabilized one year after program participation.

### Natural interest rate (r*), interest rate gap, and monetary stance for Lithuania
- Conceptual framing:
  - Natural interest rate (r*) defined as the rate that equates domestic and foreign saving and investment when output is at potential and inflation approaches its long-term trend; r* can differ across countries in a currency union.
  - The analysis uses the gap between the natural rate and the ECB’s rate to assess monetary policy stance for Lithuania.
- Quantitative r* findings:
  - Over the past two decades, estimates of r* generally show a decline of about "two and a half percentage points."
  - GDP growth averaged "8.1 percent between 1999 and before 2009" and "3.4 percent after the crisis."
  - TFP growth was remarkably high pre-crisis and halved afterwards; crisis produced a step decline in r* around 2008–09.
  - Pre-crisis average median estimate of r* is around "five percent" compared to the post-crisis average of "slightly above two percent."
  - In the open-economy extended model, some estimates show a drop of Lithuanian median r* estimates to around "negative six percent" during the GFC under certain specifications.
  - In the recovery and euro adoption period, median r* estimates for Lithuania between "2014 and 2018" are estimated to be "0.4 percentage points higher than in the largest four economies in the euro area."
- Model and market-based divergences:
  - Laubach-Williams (LW) framework and extensions emphasize trend growth and trading partners’ growth (notably Russia); common-factor and market-implied approaches emphasize euro-area-wide factors.
  - The EA common factor shows a decline in r* to "0.8 percent at the end of 2018."
  - Nelson-Siegel affine arbitrage-free term structure model (Nelson-Siegel variant) key estimate: "Lithuania’s r* is expected to hover around negative one percent in coming years."
  - Caveat: term-structure model estimates are significantly lower than standard LW model; short time series and gaps in government bond yield data warrant caution.
- The interest rate gap:
  - Defined as the difference between the forward-looking real rate of interest and the natural interest rate (nominal interest rate minus inflation expectations).
  - Main finding: the interest rate gap is estimated to be persistently negative for Lithuania, "implying highly accommodative monetary policy stance."
  - The extended LW model (model b/) implies slightly less accommodative stance than the original LW model (model a/); extended model better captures tightening in financial conditions during the crisis.
  - Both LW models point to a more accommodative monetary policy stance than market-derived estimates.
- Drivers and channels:
  - Sharp fall in r* around the crisis largely driven by a decrease in productivity growth and external factors owing to Lithuania’s small and very open economy status.
  - External factors in extended model: REER gap, trading partner growth, ten-year government bond spread against the German bund (risk premium).
  - Trading partner growth contributed the most to r* dynamics in Lithuania; e.g., Russia’s GDP dropped by "8 percent" during 2009–2010, which materially affected Lithuania.
  - Capital market-based estimates show divergence around the crisis and convergence toward euro-area r* in the run-up to euro adoption.

### Financial conditions and the PLS-chained FCI
- Measurement and methodology:
  - A chained-index partial least square (PLS) approach used to derive a Financial Conditions Index (FCI), summarizing information from many indicators and including a funding (target) variable F.
  - PLS selects variables that maximize covariance between dependent (target) and independent variables; PLS-chained extends the index as new indicators become available.
- Indicators with highest loadings:
  - External: euro area shadow rate, euro swap rate, non-sovereign AAA bond yields.
  - Domestic: Lithuania ten-year bond yield and general government interest-expenses-to-revenue ratio.
- Dynamics and interpretation:
  - PLS-chained FCI shows financial conditions tightened during the 2009 crisis and loosened since the end of 2016.
  - Negative loading coefficient of the negative interest rate gap shows loosening monetary policy has contributed to loosening financial conditions and vice versa.
- Comparison with Bank of Lithuania FCI:
  - Two FCI estimates mostly aligned except since 2016: Bank of Lithuania’s FCI shows continuing decline since 2016 (implying moderation of the financial cycle), while PLS index shows loosening conditions.
  - Bank of Lithuania evidence for a tightening cycle includes higher rejection rate of loan applications by banks, lower risk appetite by banks in surveys, and slowdown in lending to non-financial corporates.
  - Other indicators mixed: continued high growth in mortgages and real estate prices.
  - PLS-chained FCI attributes a bigger role to external (euro-area-wide) factors.

### Methods, model specifications, and caveats
- Laubach-Williams (LW) framework and extensions:
  - LW links r* to trend growth; estimated jointly with potential output using a Kalman filter and Bayesian techniques.
  - Model includes a forward-looking IS curve, a Phillips curve (extended to open-economy, backward-looking relating core inflation to output gap and import-price gap), and trend growth as a random walk with persistent shocks.
  - Quarterly annualized output, core inflation, inflation expectations, and import inflation used; EONIA used as the nominal interest rate to capture euro-area policy as exogenous.
  - LW model modified to incorporate IMF WEO estimates of potential growth as an additional signal to improve plausibility for a small open economy.
- Market-implied approaches:
  - Bayesian VAR with an euro-area common factor (Del Negro et al. 2018): jointly estimates country-specific trends of real short-term interest rates and a common euro-area trend using inflation, nominal short rates (three-month), and nominal long rates (10-year); extracts trend real short rate as a low-frequency measure of r*.
  - Term-structure approaches (Christensen et al. 2011) and other capital-market driven methods provide investor-implied information about r*.
- Interpretations and cautionary notes:
  - Large uncertainty and volatility around r* estimates warrant caution for real-time policy decisions; estimates differ by methodology and are sensitive to model specification and external shocks.
  - Model-implied sharp declines in r* during the GFC reflect extreme model dynamics when output gaps widen and monetary policy cannot restore output due to nominal-rate lower bounds.
  - Convergence of market-implied r* estimates toward euro-area trends in the run-up to euro adoption reflects reduced redenomination risk and market-perceived integration with the euro area.

### Conclusions and policy-relevant summaries
- Trend and magnitude:
  - After the 2009–10 crisis, estimates of trend growth and r* show a step decline before remaining mostly flat.
  - "Over two decades between 1999 and 2018, r* estimates have dropped by around 2 percentage points, which is consistent with the post-crisis decline in potential output."
- Monetary stance and cross-border drivers:
  - Controlling for external factors results in more volatile r* estimates and reflects Lithuania’s small, open and volatile nature.
  - "The monetary policy stance for the past six years has been more accommodative for Lithuania than for the ‘core’ EA countries by roughly 0.5 percentage points" and manifests the more advanced cyclical position of the Lithuanian economy relative to the euro area—the focus of ECB’s mandate.
  - This gap has contributed to loosening financial conditions.
  - "However, the main determinants of financial conditions, that seem to have loosened after 2016, appear to be euro-area-wide financial factors."

*Prepared by Kanghoon Keah; July 12, 2019 — REPUBLIC OF LITHUANIA, INTERNATIONAL MONETARY FUND*

### References ____________________________________________________________________________ 11

### 1ltuea2019003 - References ____________________________________________________________________________ 11

### Labor Market in Lithuania
- Labor market characterized by "large flexibility, but persistent structural issues."
- Wages are very sensitive to unemployment; real wage growth fell to -9 percent in 2009 from a pre-crisis peak of 14 percent in 2007.
- Low density of trade union and employer organization and rare occurrence of collective bargaining; wage setting largely at firm level.
- Real wages and productivity traditionally closely linked; sectoral deviations can be persistent.
- Structural unemployment historically high but "appears to be gradually falling."
- NAIRU estimated to have been "gradually decreasing to around 7 percent" per a multivariate factor (MVF) model.
- Beveridge curve: shifted outward during the crisis and partially shifted back in during recovery, indicating deterioration in labor market efficiency during the crisis with only partial recovery.
- Focus of the paper: evolution of skill mismatches since the crisis and international experience on active labor market policies (ALMPs).

### Skills Mismatch in Lithuania
- Aggregate measures indicate Lithuania "has one of the highest skill mismatches in Europe."
- Skills mismatch index follows Estevao and Tsounta (2011) framework (Box 1) and increased sharply during the crisis, only partially recovering since.
- Lithuania suffers relative labor shortage for high-skilled workers and surplus of low- and medium-skilled workers.
- Sectoral examples:
  - Skill-intensive sectors with shortages: ICT and finance.
  - Less-skill-intensive sectors with excess labor: construction and trade.
- Causes cited: legacy pre-crisis boom in low-skills sectors (notably construction), shortcomings in the education system, high emigration, restrictions on non-EU workers, limited participation in life-long learning (OECD, 2018a).
- Finding workers with the right skills a significant constraint for "over 40 percent of firms." (European Investment Bank, EIB Investment Survey, 2015)
- Vacancy rates and wage growth patterns:
  - Skill-intensive sectors (finance, communication): high wage growth and high vacancy rates → shortages.
  - Less-skill-intensive sectors (transports, construction): high wage growth but low job vacancy rates → no labor shortages; high wage growth signals labor market tightness operating below NAIRU.
- EU classification of occupations:
  - High skill needs ("shortage occupations"): ICT professionals, engineers, etc. (EC, 2017)
  - Excess supply ("surplus occupations"): street salesperson, childcare workers, teachers’ aides, machine operators, transport clerk.
- OECD Skills for Jobs Database indicators (Box 2):
  - Field-of-study mismatch in Lithuania: 35 percent (compared to EU average of 32 percent).
  - Lithuania high in overqualification and low in underqualification relative to other EU countries (OECD, 2017).
  - Lithuania has relatively low knowledge and ability needs; technology-related skill needs relatively small; low-technology skills more in demand.

Box 1. Aggregate Skills Mismatch Index (summary)
- Index measures gap between labor supply and demand by education attainment using EULFS data categorized as:
  - Low skilled: less than primary, primary and lower secondary education.
  - Medium-skilled: upper secondary and post-secondary non-tertiary education.
  - High-skilled: tertiary education (tertiary adjusted for quality using World Economic Forum data).
- Aggregate Skills Mismatch_t = sum_{j=1}^3 (S_{j,t} - M_{j,t})^2
  - where S_{j,t} is percentage of population with skill level j at time t (skills supplied)
  - and M_{j,t} is percentage of employees with skill level j at time t (skills demanded).

Box 2. OECD’s Skills for Jobs Indicators (summary)
- Two sets of indicators:
  - Skill needs indicators: shortage/surplus of skills and occupations.
  - Mismatch indicators: qualification mismatch and field-of-study mismatch.
- Qualification mismatch: share of workers under- or over-qualified relative to modal educational attainment for each occupation.
- Field-of-study mismatch: measures mismatch between type of education/knowledge and job requirements.

### Impacts of Skills Mismatch on Productivity
- Skills mismatch negatively affects productivity via:
  - Misallocation of the workforce.
  - Insufficient accumulation of firm-specific knowledge.
- Individual consequences: higher unemployment risk, lower wages, lower job satisfaction, poorer career prospects.
- Evidence and mechanisms:
  - Over-qualified workers participate less in training → hinders firm-specific knowledge accumulation (Verhaest and Omey, 2006).
  - Skill shortages negatively affect technology adoption and investments in tangible and intangible assets (Forth and Mason, 2006).
  - Under-education detrimental to firm productivity (Kampelmann and Rycx, 2012).
  - Over-qualification reduces productivity via less efficient resource allocation; under-qualification reduces allocative efficiency and within-firm productivity (McGowan and Andrews, 2015).
- Empirical finding: panel data for 26 European countries (1995–2016) found skills mismatch index had a negative and statistically significant impact on total factor productivity (TFP) and labor productivity (IMF Country Report No. 18/242).
- Lithuania: increase in skill mismatches after the crisis associated with a significant decline in productivity.

### Active Labor Market Policy (ALMP) in Lithuania
- Definition and ALMP categories:
  - Job search assistance (JSA): job centers, labor exchanges, CV/interview help.
  - Vocational education and training (VET) schemes: classroom training, on-the-job training, apprenticeships.
  - Employment subsidies: wage subsidies, hiring subsidies, business start-up subsidies, in-work benefits.
  - Public works programs (PWP): short-term employment for temporarily unemployed.
- JSA and VET address skills mismatch directly; employment subsidies and PWP support employment more broadly.
- Evidence on ALMP effectiveness (Card et al. meta-analysis of 207 studies covering 857 programs):
  - JSA: relatively short-run impact for disadvantaged participants.
  - Job training programs: better medium-run outcomes for long-term unemployed.
  - Employment subsidies, especially in public sector: negligible or negative impacts at all horizons.
  - Impact varies by participant: larger for female workers and long-term unemployed; smaller for older and young workers.
  - Duration: ALMPs small short-run impact; larger positive impact over medium- and long-run.
  - More effective during slow growth and higher unemployment.
- Potential negative unintended consequences of ALMPs:
  - Substitution effect: displacement of existing workers by program participants.
  - Skill acquisition effect: low wage subsidies disincentivize further human capital acquisition.
  - Stigmatizing: programs signaling low productivity can reduce participants' job opportunities (Brown and Koettl, 2015).

Active Labor Market Programs in Lithuania (current state)
- ALMP expenditure low: "0.3 percent of GDP in 2016" relative to OECD countries.
- Participation low: "3.7 percent of the unemployed having participated in training programs in 2016."
- ALMP spending not responsive to cyclical conditions; heavily reliant on European funds (almost two-thirds in 2015).
- Dependence on EU funding led to specified beneficiary groups in labor code: older workers, long-term unemployed, youth, disabled. New labor code (2017) added unqualified persons.
- ALMP measures: support for learning, mobility, assisted recruitment, job creation.
- New labor code changes: promote self-employment and internship, encourage self-education and non-formal adult education, provide support for mobility; abolished some programs (public works, job rotation, subsidies for individual activity according to business certificate).
- Employment subsidies have become the main ALMP measure.
- Training expenditure fluctuated; increased recently after very low levels during 2011–14.
- Training programs tend to focus on low-skilled curricula (driving, construction, cooking, beauty services) despite oversupply in these areas.
- The 2012 training voucher scheme exists but training programs remain centered on oversupplied skills.

### Policy Implications and Recommendations
- Increase ALMP spending and secure predictable, reliable public resources; funding should not be linked to European funds and should be responsive to cyclical labor market developments.
- Continue recent increase in training spending given the degree of skills upgrade needed.
  - 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 should target over-supply of low-skilled workers.
  - Participation rules could be adapted to allow long programs and those requiring expensive equipment to increase participation of low-skilled, older workers and rural workers.
- Reduce the relative importance of employment subsidies in ALMPs:
  - Concentrate subsidies on the most disadvantaged groups unlikely to find a job without support.
  - Channel extra resources toward training programs.
- Improve the training voucher system by providing information on available programs so recipients can assess fit; establish a systemic rating system based on labor market outcomes of previous participants.
- Strengthen comprehensive performance evaluation of ALMPs on a regular, long-term basis:
  - Current evaluation is quarterly and focuses on integration into labor market, registration at labor exchange, and direct benefits.
  - Program evaluation should be strengthened to improve design from a longer-term and broad-based perspective.
  - 2017 evaluation revealed vocational training and support for acquisition of work skills have larger effects in the longer run (6+ month), while subsidized employment had relatively short-duration impacts that disappeared over time; impacts stabilized one year after program participation.

*Prepared by Kanghoon Keah; July 12, 2019 — REPUBLIC OF LITHUANIA, INTERNATIONAL MONETARY FUND*

### References

### References and The Interest Rate Gap and Financial Conditions in Lithuania

### References (selected)
- 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.
- Card, D., J. Klube, and A. Weber, 2009, “Active Labor Market Policy Evaluations: A Meta-Analysis,” CESifo Working Paper, No. 2570.
- 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.
- Estevao, M., 2003, “Do ALMP Increase Employment?”, IMF Working Paper 03/234.
- Forth, J., and G. Mason, 2006, “Do ICT Skill Shortages Hamper Firms’ Performance? Evidence from UK Benchmarking Surveys,” NIESR 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 Linked 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.
- 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.”
- OECD, 2018b, “OECD Reviews of Labor Market and Social Policies: Lithuania.”
- OECD, 2017, “Getting Skills Right for Jobs Indicators.”
- OECD, 2016a, “Getting Skills Right: Assessing and Anticipating Changing Skill Needs.”
- OECD, 2016b, “OECD Economic Surveys: Lithuania 2016.”
- Verhaest, D. and E. Omey, 2006, “The Impact of Overeducation and Its Measurement,” Social Indicators Research, Vol. 77, Issue 3, 419-448.

### The interest rate gap and financial conditions: key themes and findings
- Motivation and context
  - Despite a common nominal policy rate in the euro area, domestic monetary conditions differ across countries because of different domestic characteristics, stages in the business cycle, and financial transmission mechanisms.
  - The natural interest rate (r*) is defined as the rate that equates domestic and foreign saving and investment when output is at potential and inflation approaches its long-term trend; r* can differ across countries in a currency union.
  - This analysis uses the gap between the natural rate and the ECB’s rate to assess monetary policy stance for Lithuania.

- Major conceptual frameworks used
  - Two strands to operationalize r*:
    - DSGE-style approach: models r* as the counterfactual interest rate in the absence of nominal frictions; often volatile; linear combinations of transitory shocks (Barsky et al. 2014).
    - Reduced-form / semi-structural approach (Laubach-Williams and extensions): models r* as closely related to trend growth with time-varying preference parameters (Laubach and Williams 2003; Holston et al. 2017).
  - Market-data approaches: capital-market-driven estimates (Del Negro et al. 2018; Christensen et al. 2011) that reflect investor expectations of r*.
  - The paper contrasts several approaches: Laubach-Williams original and extended models (Pescatori and Turunen 2015), and capital-market models (Del Negro et al. 2018; Christensen et al. 2011).

- Quantitative findings on Lithuania’s r* and growth
  - Over the past two decades, estimates of r* generally show a decline of about two and a half percentage points.
  - GDP growth averaged 8.1 percent between 1999 and before 2009 and only 3.4 percent after the crisis.
  - Total factor productivity (TFP) growth was remarkably high pre-crisis and halved afterwards; the crisis produced a step decline in r* around 2008–09 rather than a gradual decline.
  - Pre-crisis average median estimate of r* is around five percent compared to the post-crisis average of slightly above two percent.
  - In the open-economy extended model, some estimates show a drop of Lithuanian median r* estimates to around negative six percent during the GFC under certain specifications.
  - In the recovery and euro adoption period, median r* estimates for Lithuania between 2014 and 2018 are estimated to be 0.4 percentage points higher than in the largest four economies in the euro area.

- Drivers and channels
  - The sharp fall in Lithuania’s r* around the crisis is largely driven by a decrease in productivity growth and by external factors due to Lithuania’s small and very open economy status.
  - External factors considered as drivers of r* in the extended model: REER gap (deviation of REER from its long-term average), trading partner growth, and the ten-year government bond spread against the German bund (proxy for risk premium).
  - Among external factors, trading partner growth contributed the most to r* dynamics in Lithuania; during 2009–2010 Russia’s GDP dropped by 8 percent, which materially affected Lithuania.
  - Capital market-based estimates (Del Negro et al. 2018 joint estimation with euro-area common factors) show periods of divergence between Lithuania and euro-area r* around the crisis and convergence in the run up to euro adoption, signaling market views of Lithuanian integration into the euro area.

- Monetary stance, transmission, and financial conditions
  - Using the gap between domestic r* and the ECB policy rate suggests policy has been largely accommodating for Lithuania since at least 2011.
  - Accommodative (exogenous) ECB monetary policy has contributed to relaxing financial conditions in Lithuania regardless of cyclical conditions, given Lithuania’s higher potential growth (and thus higher r*) as a converging economy.
  - However, monetary policy stance does not always translate into looser financial conditions: the Financial Condition Index (FCI) indicates financial conditions tightened after the 2009 crisis and only clearly recovered since 2016.
  - The loading coefficient on the interest rate gap in the FCI is relatively low; euro-area-wide financial factors play a larger role in Lithuania’s financial conditions than domestic interest-rate gaps alone.
  - Bank of Lithuania’s domestic-factor-based FCI shows different dynamics compared to the broader FCI used in this analysis.

### Methods and model specifications (summary)
- Laubach-Williams (LW) framework and extensions
  - LW conceptual link: r* is related to trend growth and is estimated jointly with potential output using a Kalman filter and Bayesian techniques.
  - Equations include a forward-looking IS curve, a Phillips curve (extended here to an open-economy, backward-looking Phillips curve relating core inflation to the output gap and the gap between import prices and core inflation), and trend growth modeled as a random walk with persistent shocks.
  - The model used quarterly annualized output, core inflation, inflation expectations, and import inflation; EONIA is used as the nominal interest rate to capture euro-area policy as exogenous to Lithuania.
  - To improve plausibility of output-gap estimates for a small open economy, the LW model is modified to incorporate IMF WEO estimates of potential growth as an additional signal.

- Market-implied approaches
  - Bayesian VAR with a euro-area common factor (Del Negro et al. 2018):
    - Jointly estimates country-specific trends of real short-term interest rates and a common euro-area trend using inflation, nominal short rates (three-month), and nominal long rates (10-year) with trend components as random walks.
    - Extracts a trend real short rate as a low-frequency measure of r* and allows common euro-area factors and country-specific trends.
  - Term-structure approaches (Christensen et al. 2011) and other capital-market driven methods are also used to provide investor-implied information about r*.

### Interpretations and cautionary notes
- Large uncertainty and volatility around r* estimates warrant caution in their use for real-time policy decisions; estimates differ by methodology and are sensitive to model specification and external shocks.
- Model-implied sharp declines in r* during the GFC reflect extreme model dynamics when output gaps widen and monetary policy cannot restore output due to limits on how low nominal rates can go.
- Convergence of market-implied r* estimates toward euro-area trends in the run-up to euro adoption reflects reduced redenomination risk and market-perceived integration with the euro area.

*Source: 1ltuea2019003 - References; 1ltuea2019003 - The Interest Rate Gap and Financial Conditions in Lithuania (excerpts).*

### 0.8 percent at the end of 2018. The EA common factor also shows this decline with almost identical

### 1ltuea2019003 - 0.8 percent at the end of 2018. The EA common factor also shows this decline with almost identical

### Natural interest rate (r*) estimates and dynamics
- The EA common factor shows a decline in r* to "0.8 percent at the end of 2018."
- Laubach-Williams (LW) model vs. common-factor approach:
  - Unlike the LW model, the common-factor approach produces a more gradual decline in r* and a rebound after the crisis that "is not sustained, especially after 2014."
  - The LW model emphasizes potential growth and trading partners’ growth (notably Russia); the common-factor approach emphasizes euro-area-wide factors in determining r* in Lithuania.
- Cross-country and sample-period patterns:
  - Prior to euro adoption, r* in Lithuania is persistently higher than in the core EA countries; the gap widens slightly in the early 2000s and particularly from 2007 to 2009.
  - In the runup to euro adoption the gap starts to close as markets expected more integration and co-movement of business cycles between Lithuania and the rest of the currency union.
  - Throughout the sample period, both median estimates were within the other’s ‘credibility band’.

### Affine arbitrage-free term structure model (Nelson-Siegel variant)
- Method and interpretation:
  - A version of the Nelson-Siegel term structure model (Christensen et al., 2011) is used to fit the yield curve of Lithuanian government debt through time imposing an arbitrage-free condition across time and maturities.
  - The implied path of future nominal short rates (3-month treasury bill), consistent with this yield curve, is derived and used as a proxy for the (nominal) natural rate five-years ahead.
  - Inflation expectations are used to convert nominal rates into real rates.
- Key estimate:
  - "Lithuania’s r* is expected to hover around negative one percent in coming years."
- Caveats:
  - Estimates from this model are significantly lower than from the standard LW model.
  - Lithuania’s r* estimate registers a lower end-point than other core EA countries.
  - "With relatively short time series and gaps in the data for government bond yields at different maturities, Lithuania results should be taken with caution."

### The interest rate gap
- Definition:
  - The interest rate gap is measured as the difference between the forward-looking real rate of interest and the natural interest rate. (Footnote: Nominal interest rate minus inflation expectations.)
- Main findings:
  - The interest rate gap is estimated to be persistently negative for Lithuania, "implying highly accommodative monetary policy stance."
  - Different methodologies point to a monetary policy stance that has been highly accommodative for Lithuania, and more so than for “core” economies in the euro area, especially after the global financial crisis.
  - The extended LW model (model b/) that better captures Lithuanian characteristics implies slightly less accommodative stance than the original model (model a/). The extended model is the only one that captures tightening in financial conditions during the crisis.
  - Both LW models point to more accommodative monetary policy stance than market-derived estimates.

### Financial conditions and the PLS-chained FCI
- Measurement approach:
  - The paper applies a chained-index partial least square (PLS) approach to derive a Financial Conditions Index (FCI), summarizing information from a large number of indicators and including a funding (target) variable F.
  - PLS selects variables that maximize covariance between dependent (target) and independent variables, addressing limitations of PCA when the most-volatile variable is not the most relevant.
  - The PLS-chained methodology extends the index as new indicators become available.
- Indicators with highest loadings:
  - External factors: euro area shadow rate, euro swap rate, and non-sovereign AAA bond yields.
  - Domestic factors: Lithuania ten-year bond yield and general government interest-expenses-to-revenue ratio.
- Dynamics and interpretation:
  - The PLS-chained FCI shows that financial conditions tightened during the 2009 crisis and loosened since the end of 2016.
  - The negative loading coefficient of the negative interest rate gap shows that loosening monetary policy has contributed to a loosening of financial conditions and vice versa.
- Comparison with Bank of Lithuania FCI:
  - The two FCI estimates are mostly aligned except since 2016. The Bank of Lithuania’s FCI shows continuing decline since 2016 (implying moderation of the financial cycle), while the PLS index shows loosening conditions.
  - Bank of Lithuania evidence for a tightening cycle includes a higher rejection rate of loan applications by banks, a lower risk appetite by banks according to surveys, and a slowdown in lending to non-financial corporates.
  - Other indicators are more mixed, such as continued high growth in mortgages and real estate prices.
  - The PLS-chained FCI gives a bigger role to external (euro-area-wide) factors.

### Conclusion (summary of implications)
- Trend and magnitude:
  - After the 2009–10 crisis, estimates of trend growth and r* show a step decline before remaining mostly flat.
  - "Over two decades between 1999 and 2018, r* estimates have dropped by around 2 percentage points, which is consistent with the post-crisis decline in potential output."
- Monetary stance and cross-border drivers:
  - Controlling for external factors results in more volatile r* estimates and reflects the small, open and more volatile nature of the Lithuanian economy.
  - "The monetary policy stance for the past six years has been more accommodative for Lithuania than for the ‘core’ EA countries by roughly 0.5 percentage points" and manifests the more advanced cyclical position of the Lithuanian economy relative to the euro area—the focus of ECB’s mandate.
  - This gap has contributed to loosening financial conditions.
  - "However, the main determinants of financial conditions, that seem to have loosened after 2016, appear to be euro-area-wide financial factors."

### Annex I — calibration of priors (Laubach-Williams model)
- Model calibration and key assumptions:
  - Quarterly data are used for Lithuania.
  - Output gap (IS curve): parameters drawn from constrained regression of output gap on lagged output gap and lagged interest rate gap. Interest rate gap proxied by HP-filtered real interest rate as the natural rate. Coefficients on two lagged interest rate gaps constrained to be equal.
  - Phillips Curve: current inflation determined by lagged inflation, output gaps, import inflation and oil price inflation. "The coefficients of eight quarters of lagged inflation sum to 1."
  - Potential output: assumed to follow a random walk. Trend year-on-year growth converted to quarterly growth.
  - Natural rate equation: natural rate is a function of trend growth and other domestic factors. "The coefficient of trend growth ... is assumed to be 1."
  - Dynamics of other domestic factors: priors for autocorrelation coefficients drawn from independent estimation; contributions to Z assumed equal and sum to one when other factors included.
  - Error-term priors: based on Pescatori and Turunen (2015).
- The Annex presents a table of prior and posterior distributions of key parameters used in the Laubach-Williams calibration for Lithuania (parameter names and distributions described).

*Source: IMF staff calculations and accompanying text in the provided chapter.*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2019/1ltuea2019003.pdf_