## wpiea2025018-print-pdf

## Source details

**Canonical URL:** [wpiea2025018-print-pdf](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025018-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025018-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025018-print-pdf.pdf.json)

---

### Annex I — Constant Share Analysis and Export Dynamics
- Manufacturing and economic structure
  - After regaining independence, Baltic countries had broadly comparable shares of manufacturing in total output; subsequently:
    - Gross value added (GVA) of manufacturing declined rapidly in Estonia and Latvia, while remaining elevated in Lithuania.
    - Manufacturing contribution to GVA has started to decline in Lithuania in more recent years; significant differences persist across countries.
  - Economic structure (2023) sectors highlighted: Industry (manufacturing, mining and quarrying, electricity, water and other utilities), Trade & Hospitality, Public Admin, Education & Defence, Professional services, Real Estate, Construction, ICT, Finance & Insurance, Arts & Recreation, Agriculture.

- Exports, trade openness, and sectoral composition
  - Exports as a percentage of GDP increased steadily in all three economies over past decades.
  - Trade openness (exports + imports as a percentage of GDP) in 2023:
    - Estonia: 154.9 percent
    - Lithuania: 149 percent
    - Latvia: 138 percent
    - EU average: about 100 percent
  - Lithuania: particularly strong export growth driven by services exports since 2010; larger manufacturing share but lower technological content of exports than Estonia and Latvia.
  - Lithuania export-share evolution:
    - By end-2019, Lithuanian export share was 0.17 percent of global exports, over three times higher than in early 2000.
    - Lithuanian market share in the EU rose from around 0.3 in 2014 to close to 0.4 percent after firms diversified away from Russia/Belarus.

- Technological content and re-exports
  - High-technology exports (percent of manufactured exports) higher in Estonia and Latvia than in Lithuania.
  - Re-exports and exports to Russia/Belarus materially affect measured export dynamics in Lithuania, often involving little value added.

- Real exchange rate (REER) and pass-through
  - Large inflation differentials and high nominal labor costs drove appreciation of the REER across the Baltics.
  - Lithuania exhibited real exchange rate depreciation when measured by export prices (muted pass-through from import to export prices).
  - REER change decomposition (2021Q4–2023Q4) attributes movements to:
    - HICP differential with EA-20.
    - Changes with non-EA partners (bilateral exchange rate effects).

- Corporate balance sheets and resilience
  - Baltic nonfinancial corporations preserved sizable profit shares (gross operating surplus/GVA) over last three decades.
  - Pandemic public support improved leverage, solvency, and liquidity.
  - Net financial wealth: Lithuanian corporations stronger than Estonia, Latvia, and other EU members in recent years and further improved lately.
  - Financial indices reported: NFC Profit Shares (ratio); NFC Liquidity: Current Ratio; NFC Net Financial Liabilities (Percent of GDP); NFC Leverage: Debt to Equity.

- Recent export market share dynamics (post-pandemic 2021Q3–2023Q4)
  - Export market share changes:
    - Estonia: -23 percent
    - Latvia: -6 percent
    - Lithuania: -7 percent
  - Estonia’s larger loss mainly reflects competitiveness effects; Latvia and Lithuania losses moderate and concentrated outside the EU.

- Constant market share decomposition interpretation
  - Decomposition components: intensive margin (competitiveness effect), extensive margin (composition/destination-market effect), and interaction.
  - Subperiod patterns:
    - 2010Q1–2013Q4 (post-GFC): large competitive gains plus moderate export market growth; Latvia and Lithuania gained market shares; Estonia’s gains contained.
    - 2014Q1–2021Q2: Baltics remained relatively competitive, competitiveness outweighed weaker destination-market dynamics.
    - 2021Q3–2023Q4 (post-pandemic): divergence — Lithuania’s moderate loss largely reflects demand developments; Latvia modest competitive gains; Estonia larger loss driven by competitiveness.

- Open questions linking to competitiveness
  - Why did common shocks and similar inflation/REER appreciation produce different export market share behavior across Baltics?
  - Potential explanations: muted pass-through from import to export prices (production/export composition), stronger corporate balance sheets aiding shock absorption.
  - Next step: use Balassa-Samuelson framework to construct TFP-consistent REER and assess competitiveness via REER gaps (actual − TFP-based).

---

### Section 3.2 — Data, Methodology, and Key Quantified Outcomes
- Data and measurement
  - Quarterly data: 1995 to mid-2023.
  - Output and capital: millions of constant 2015 euros.
  - Labor: thousands of hours worked per quarter.
  - Quarterly capital stock: constructed from AMECO annual stock with depreciation and multiplied by industrial capacity utilization to obtain effective capital stock, kt.
  - Labor input: employees et × average hours ht.
  - Labor share, αt: smooth trend of compensation of employees to GDP (time-varying).

- TFP estimation (growth-accounting)
  - Solow residual from Cobb-Douglas: at = yt − (1 − α) kt − α lt (notation preserved).
  - Time-varying labor share αt introduces an extra term in GDP growth decomposition to capture labor-share changes.

- Cointegration and Balassa-Samuelson assessment
  - ADF tests for integration; Johansen cointegration tests find at least one cointegrating relationship between TFP and REER in most cases.
  - Exception: Latvia — no cointegration with actual REER; used HP-trend REER instead.
  - DOLS used for cointegrating regressions with automatic lead/lag selection by SIC.

- Constructing TFP-based REER and REER gap interpretation
  - REER gap = actual REER − TFP-based REER.
    - Negative gap → undervalued relative to long-term norm → competitive advantage.
    - Positive gap → overvalued relative to TFP-based norm → competitiveness erosion.
  - Figures referenced: Figure 3.1 (TFP estimates) and Figure 3.2 (Actual and TFP-Based REER, 1995-2023).
  - Country sample windows: Latvia: 2002Q2-2023Q2; Lithuania: 1998Q3-2023Q2.

- Potential GDP and cyclical vs structural decomposition
  - Potential GDP via multivariate Kalman filter combining quarterly GDP with: monthly confidence indices, unemployment rate, industrial capacity utilization.
  - Structural (trend) TFP, ât, from production function applied to filtered k̄t, ēt, h̄t.
  - Sample partitioning: Pre-GFC 1995-2008; Post-GFC 2009-2019; Post-pandemic 2020-2023Q2.

- Key empirical quantified outcomes
  - Potential GDP growth decline (pre-GFC → post-GFC):
    - Fell from above 5 percent (pre-GFC) to around:
      - 1 percent in Latvia (post-GFC),
      - 2 percent in Estonia (post-GFC),
      - between 2.5 and 3 percent in Lithuania (post-GFC).
  - Structural gross TFP (GTFP = TFP growth + change in labor share) dynamics:
    - Estonia: GTFP growth declined from above 1 percent (pre-GFC) to negative growth (post-GFC).
    - Lithuania: trend GTFP growth accelerated over time toward positive territory.
    - Latvia: GTFP growth accelerated after GFC but decelerated post-pandemic and turned negative recently.
  - TFP and REER gap patterns:
    - Pre-GFC: Estonia and Latvia had prolonged negative REER gaps (competitive edge); Lithuania’s gap turned negative only in 2005-2007.
    - Post-GFC: competitive advantage eroded for Estonia and Latvia; negative gaps smaller and positive gaps more frequent.
    - Post-pandemic: rapid REER appreciation led to decoupling; Latvia and especially Estonia saw declining TFP push REER gaps to positive territory; Lithuania’s stronger TFP and more undervalued REER provided buffers.
    - Late 2021 onward: Estonia experienced a pronounced positive REER gap associated with TFP decline and protracted REER appreciation, linked to export market share losses and weaker activity.

- Role of growth components across subsamples
  - Capital accumulation: major driver for Estonia and Lithuania over full sample; contribution declined post-GFC in Latvia.
  - Labor: limited or drag on growth (especially Latvia).
  - TFP:
    - Estonia: TFP significantly contributed pre-GFC but faded and became negative post-GFC; decline in average TFP growth explains almost 90 percent of fall in potential GDP growth (see Table 3.1).
    - Latvia and Lithuania: TFP increasingly a growth driver for Latvia and especially Lithuania; Lithuania’s structural TFP acceleration since GFC helped shield it from shocks.
  - Scarring: potential GDP estimates imply scarring after the GFC for all Baltics, with a more pronounced impact on Estonia; pandemic scarring similar but less pronounced.

---

### Section 3.5 — Key Takeaways (Quantified Snippets and Main Findings)
- Analytical outputs: estimates of potential GDP, GDP growth decompositions, and TFP for three Baltics (1995-2023); TFP-based REER used to measure competitiveness.
- Exact quantitative decomposition snippet preserved from source:
  - TotalTrendCyclicalTotalTrendC
    yclicalTotalTrendCyclical
    GDP-4.3-3.1-1.2-4.1-4.70.6-3.1-2.0-1.0
    TFP-5.2-2.7-2.52.01.30.78.78.50.2
    K-1.1-2.31.1-7.0-7.60.6-8.9-8.1-0.9
    L1.31.30.10.20.9-0.7-0.5-0.1-0.4
    Δα0.70.70.00.70.70.0-2.4-2.30.0
    LithuaniaEstoniaLatvia

- Main findings
  - Potential GDP growth has fallen since GFC, largely due to steady decline in TFP growth, especially in Estonia.
  - Estonia: TFP growth decline since GFC; level of TFP has dropped since 2020; decline includes a structural component likely linked to scarring.
  - Lithuania: acceleration in TFP growth in recent years contrasts with Estonia.
  - REER and competitiveness:
    - Common real appreciation occurred, but long-term TFP growth differences produced divergent competitiveness outcomes.
    - Estonia experienced the largest positive wedge (actual − TFP-based REER) earlier and wider than Latvia and Lithuania, contributing to a competitive disadvantage and protracted downturn.

---

### Section 4.5 — Allocative Efficiency, Regulation, and Policy Implications
- Regulatory landscape
  - Baltic product market regulation relatively light vs other advanced economies.
  - Labor market regulation varies:
    - Latvia: more stringent labor market regulation than other Baltics and other advanced economies per OECD Employment Protection Legislation (2019): higher severance pay for low-tenure employees; stricter unfair dismissal definitions; generous reinstatement availability.
    - Dismissal costs under regular contracts higher in Latvia than other OECD economies.
  - Indicators cited: Product market regulation (2018); Employment protection legislation (2019); Financial market liberalization (2014).

- Allocative efficiency findings and sectoral patterns
  - Allocative efficiency measures standardized to USA benchmark; Lithuania is an outlier due to administrative data bias toward small/micro firms.
  - Resource misallocation reduced TFP growth on average across all three Baltics during the sample.
  - Estonia: allocative efficiency worsened generally over time, with limited short-lived recovery post-GFC; real estate industry saw deterioration before GFC.
  - Latvia: allocative efficiency slightly improved after 2016 but recovery limited; some industries worsened in 2012-2015.
  - Lithuania: resource misallocation generally worsened over time.
  - Sectoral pattern: productivity loss from misallocation more pronounced in services than goods.

- Policy-relevant recommendations (preserved wording and priorities)
  - Structural reforms to improve allocative efficiency and productivity:
    - Reform product markets to reduce regulatory barriers.
    - Liberalize financial markets to improve access to finance, particularly for small firms.
    - Pursue labor market reforms to balance job protection with labor market flexibility.
  - Rationale: Improving allocative efficiency through these reforms would support productivity growth and enhance competitiveness for Estonia, Latvia, and Lithuania.

---

### Annex II & III — Estimation Details and Allocative Efficiency Methodology
- Johansen cointegration tests — rank selection and deterministic cases
  - Rank selection at 0.05 level using MacKinnon-Haug-Michelis (1999) critical values.
  - Deterministic Cases defined (Cases 1–5 with variants).
  - Rank selection results (reported exactly):
    - ln(TFP) Trace: 1 1 1 1 1 2 2
    - ln(REER) Max-Eigen: 1 1 1 1 1 2 2
    - ln(TFP) Trace: 1 1 1 1 1 2 2
    - ln(REER-HP) Max-Eigen: 1 1 1 1 1 2 2
    - ln(TFP) Trace: 2 1 1 2 2 2 2
    - ln(REER) Max-Eigen: 2 1 1 2 2 0 0
  - Endogenous variables and samples (exact):
    - Countries: Estonia, Latvia, Lithuania
    - Samples: 1995Q4-2023Q2; 2002Q4-2023Q2; 1999Q2-2023Q2
    - No. of Observations: 111; 83; 97

- DOLS cointegration estimates (Table All.2) — key coefficients and fit statistics (exact)
  - Samples and Observations:
    - Estonia: 1996Q4-2023Q2, Observations 107
    - Latvia: 2002Q2-2023Q2, Observations 85
    - Lithuania: 1998Q4-2023Q2, Observations 99
  - Ln(TFP) coefficient:
    - Estonia: 0.47; Std. Error 0.10; t-Statistic 4.77; Prob. 0.000
    - Latvia: 0.30; Std. Error 0.07; t-Statistic 4.46; Prob. 0.000
    - Lithuania: 0.03; Std. Error 0.01; t-Statistic 2.49; Prob. 0.01
  - C (constant) coefficient:
    - Estonia: 117.87; Std. Error 29.79; t-Statistic 3.96; Prob. 0.000
    - Latvia: -30.85; Std. Error 6.56; t-Statistic -4.70; Prob. 0.000
    - Lithuania: -4.93; Std. Error 1.49; t-Statistic -3.31; Prob. 0.00
  - @TREND coefficient:
    - Estonia: 0.32; Std. Error 0.02; t-Statistic 18.80; Prob. 0.000
    - Latvia: 0.05; Std. Error 0.02; t-Statistic 2.87; Prob. 0.01
  - R-squared:
    - Estonia: 0.96
    - Latvia: 0.48
    - Lithuania: 0.98
  - Adjusted R-squared:
    - Estonia: 0.96
    - Latvia: 0.47
    - Lithuania: 0.98
  - S.E. of regression:
    - Estonia: 2.48
    - Latvia: 6.04
    - Lithuania: 1.59
  - Long-run variance:
    - Estonia: 18.70
    - Latvia: 136.26
    - Lithuania: 2.69
  - Mean dependent var:
    - Estonia: -3.06
    - Latvia: -1.97
    - Lithuania: -0.87
  - S.D. dependent var:
    - Estonia: 12.45
    - Latvia: 8.30
    - Lithuania: 10.37
  - Sum squared resid:
    - Estonia: 595.72
    - Latvia: 2,989.53
    - Lithuania: 236.88

- Decomposition of GDP growth — selected exact entries (percentage points; source tables AII.3–AII.5)
  - Estonia (1995-2008; 2009-2019; 2020-2023Q2) — selected rows (values reproduced exactly):
    - TFP Total: 30.7; Structural: 32.2; Cyclical: -1.6
    - Capital Total: 57.6; Structural: 0.2; Cyclical: -2.6
    - Capital accumulation Total: 47.9; Structural: 47.9; Cyclical: n.a
    - Capacity utilization Total: 9.7; Structural: 12.3; Cyclical: -2.6
    - Labor Total: -3.1; Structural: -5.2; Cyclical: 2.1
    - GDP Total: 70.9; Structural: 73.0; Cyclical: -2.1
    - Average Annual Growth GDP: 5.1; 5.2; -0.1; 2.4; 2.1; 0.3; 0.8; 2.2; -1.4
  - Latvia (1998Q3-2008; 2009-2019; 2020-2023Q2) — selected rows:
    - TFP Total: 3.5; Structural: 4.0; Cyclical: -0.5
    - Capital Total: 46.8; Structural: 51.3; Cyclical: -4.5
    - Capital accumulation Total: 52.2; Structural: 52.2; Cyclical: n.a
    - Capacity utilization Total: -5.4; Structural: -0.9; Cyclical: -4.5
    - Labor Total: 6.6; Structural: 0.2; Cyclical: 6.4
    - GDP Total: 38.9; Structural: 37.2; Cyclical: 1.7
    - Average Annual Growth GDP: 6.0; 5.7; 0.3; 1.0; 1.3; -0.2; 1.9; 1.0; 0.9
  - Lithuania (1998Q3-2008; 2009-2019; 2020-2023Q2) — selected rows:
    - TFP Total: -48.5; Structural: -49.0; Cyclical: 0.6
    - Capital Total: 115.0; Structural: 114.0; Cyclical: 1.0
    - Capital accumulation Total: 97.3; Structural: 97.3; Cyclical: n.a
    - Capacity utilization Total: 17.7; Structural: 16.6; Cyclical: 1.0
    - Labor Total: 2.0; Structural: 2.2; Cyclical: -0.2
    - GDP Total: 56.1; Structural: 54.6; Cyclical: 1.5
    - Average Annual Growth GDP: 5.2; 5.1; 0.1; 2.1; 2.4; -0.4; 2.1; 3.0; -0.9

- Methodology for allocative efficiency (Annex III) — framework preserved exactly
  - Firm-level Cobb-Douglas with country-sector capital share 훼훼_cccc; Y_cccc, A_cccc, K_cccc, L_cccc denote output, technology, capital, labor; subscripts c, s, i, t denote country, sector, firm, year.
  - Aggregation with CES (휎휎_cc elasticity of substitution).
  - Distortions: capital/labor effective-cost wedges 휏휏_K and 휏휏_L; output tax 휏휏_Y; composite 휏휏_ccccc function of distortions.
  - Profit maximization under monopolistic competition leads to equilibrium expressions (Equations (4)–(10)); distortions reduce aggregate TFP via wedge AE in Equation (10).
  - Relationship: "For each unit decline in allocative efficiency, there will be a one-percentage point decline in TFP growth."
  - Aggregation and decomposition:
    - Aggregate firm-level allocative efficiency to sector level; then to country using EUKLEMS value-added shares.
    - Annual TFP growth from AMECO country-level TFP index.
    - Decompose TFP growth into innovation and allocative efficiency components per Equation (11).

*Italic: Source — IMF staff estimates and analysis in "Competitiveness and Productivity in the Baltics: Common Shocks, Different Implications", Working Paper No. WP/2025/018 (content unit: wpiea2025018-print-pdf).*

### Annex I. The Constant Share Analysis Decomposition .....................................................................

### Annex I. The Constant Share Analysis Decomposition

### Overview
- Title: The Constant Share Analysis Decomposition
- Location in source PDF: page 34

*IMF Working Paper: Competitiveness and Productivity in the Baltics: Common Shocks, Different Implications — Annex I (page 34).*

### 2.5 Production and export composition varies significantly across the Baltics

### 2.5 Production and export composition varies significantly across the Baltics

### Manufacturing and economic structure
- After regaining independence, the three Baltic countries exhibited broadly comparable shares of manufacturing in total output; subsequently:
  - Gross value added (GVA) of manufacturing declined rapidly in Estonia and Latvia, while remaining elevated in Lithuania.
  - Manufacturing contribution to GVA has started to decline in Lithuania in more recent years, but significant differences persist in economic structure across the three countries.
- Economic structure (2023) components highlighted include: Industry (manufacturing, mining and quarrying, electricity, water and other utilities), Trade & Hospitality, Public Admin, Education & Defence, Professional services, Real Estate, Construction, ICT, Finance & Insurance, Arts & Recreation, Agriculture.

### Exports, trade openness, and sectoral composition
- Exports as a percentage of GDP have steadily increased in all three economies over past decades.
- Trade openness (exports + imports as a percentage of GDP) in 2023:
  - Estonia: 154.9 percent
  - Lithuania: 149 percent
  - Latvia: 138 percent
  - EU average: about 100 percent
- Lithuania experienced particularly strong export growth, driven by services exports since 2010.
- Despite Lithuania’s larger share of manufacturing, the technological content of its exports is lower than that of Estonia and Latvia, reflecting:
  - Greater reliance on more traditional manufacturing output.
  - Product specialization and lower GVC participation relative to Estonia.
  - Possible explanation for Lithuania’s position as a price taker in export markets and consistency with the terms-of-trade puzzle discussed earlier.
- Lithuania’s export share evolution:
  - By the end of 2019, the Lithuanian export share was 0.17 percent of global exports, over three times higher than in early 2000.
  - Lithuanian companies diversified towards other markets after 2015 sanctions, with market share in the EU increasing from around 0.3 in 2014 to close to 0.4 percent.

### Technological content and re-exports
- High-technology exports (percent of manufactured exports) are higher in Estonia and Latvia than in Lithuania.
- Re-exports and exports to Russia/Belarus affect measured export dynamics, notably in Lithuania where a large share of such trade involves little value added.

### Real exchange rate developments and pass-through
- The region experienced large inflation differentials relative to trading partners and high nominal labor costs; prices, wages and input costs shifted up, resulting in significant appreciation of the real effective exchange rate (REER) for the three countries.
- Lithuania experienced real exchange rate depreciation when measured in terms of export prices, reflecting a more muted pass-through from import to export prices.
- REER change decomposition (2021Q4–2023Q4) attributes REER movements to:
  - HICP differential with EA-20 (inflation differential with euro-area partners).
  - Changes with non-EA partners (bilateral exchange rate effects for some countries).
- REER (2023/19 Percent Change) components shown include export price based and ULC based contributions.

### Corporate balance sheets and resilience
- Baltic nonfinancial corporations preserved sizable profit shares across the last three decades.
  - Profit share measured as gross operating surplus/gross value added.
- Pandemic-related unprecedented public support helped contain leverage and improve solvency and liquidity.
- Net financial wealth (total financial assets minus liabilities):
  - Lithuanian corporations have been consistently stronger than Estonia, Latvia, and other EU member states in recent years, and this position has further improved lately.
- Financial indices for non-financial corporations include:
  - NFC Profit Shares (ratio)
  - NFC Liquidity: Current Ratio (current financial assets/current financial liabilities)
  - NFC Net Financial Liabilities (Percent of GDP)
  - NFC Leverage: Debt to Equity (ratio measured as (debt securities+loans)/total equity)

### Recent export market share dynamics
- Export market shares grew substantially prior to the Global Financial Crisis (GFC); after 2008:
  - Estonia’s share flattened.
  - Lithuania’s and, to a lesser extent, Latvia’s shares continued to rise until recently.
- Post-2019 and post-pandemic developments:
  - All three Baltic economies experienced losses of export market shares in the post-pandemic period (2021Q3 to 2023Q4):
    - Estonia: -23 percent
    - Latvia: -6 percent
    - Lithuania: -7 percent
  - Lithuania and Latvia’s losses were moderate and largely concentrated outside the EU; Lithuania’s correction followed a decade of steady gains.
  - Estonia’s larger loss is associated with competitiveness effects rather than primarily destination-market demand changes.

### Constant market share decomposition and interpretation
- A constant market share methodology decomposes market share changes into:
  - Intensive margin (competitiveness effect): change attributable to higher penetration margins.
  - Extensive margin (composition effect): change attributable to change in size of destination markets (foreign demand).
  - Interaction of both factors.
- Historical periods:
  - Post-GFC (2010Q1–2013Q4): large competitive gains plus moderate export market growth; Latvia and Lithuania gained market shares, Estonia’s progress was more contained.
  - 2014Q1–2021Q2: Baltics remained relatively competitive, outweighing less favorable destination market dynamics; developments broadly comparable across the three countries.
  - Post-pandemic (2021Q3–2023Q4): divergence across the region:
    - Lithuania: moderate loss of export share largely reflects demand developments in trading partners (sanctions impact on Russia and Belarus, significant share of re-exports with little value added).
    - Latvia: modest competitive gains despite some demand effects.
    - Estonia: larger loss of market share is mostly explained by a competitiveness effect.

### Key open questions and links to competitiveness analysis
- Unresolved issues highlighted:
  - Why, despite common shocks and similar inflation and REER appreciation, did the Baltic countries experience very different export market share behavior?
  - To what extent do these divergences reflect conjunctural developments versus deeper structural differences?
  - Has allocative efficiency played a role?
- Early intuition offered:
  - A more muted pass-through from import to export prices (tied to production and export composition) and stronger corporate balance sheets may help explain different abilities to absorb supply-side shocks.
- The next section aims to provide a systematic approach to competitiveness in the Baltics using the Balassa-Samuelson hypothesis and constructing a TFP-consistent REER to assess competitiveness via the gap between actual REER and TFP-based REER.

*IMF Working Paper chapter excerpt.*

### 3.2 Data and methodology

### 3.2 Data and methodology

### Data and sample
- Quarterly data from 1995 to mid-2023.
- Output and capital: measured in millions of constant 2015 euros.
- Labor: measured in thousands of hours worked per quarter.
- Quarterly capital stock: constructed by applying quarterly investment flows to annual stock data from the European Commission’s Annual Macro-Economic Database (AMECO) and estimates of depreciation rates; then multiplied by a measure of industrial capacity utilization to obtain the effective capital stock, kt.
- Labor input: constructed as number of employees (et) multiplied by average hours worked per employee (ht).
- Labor share, αt: a smooth trend of the ratio of compensation of employees to GDP is used (time-varying rather than a fixed calibrated value).

### TFP estimation (growth-accounting approach)
- TFP obtained as a Solow residual from a standard Cobb-Douglas production function on quarterly data:
  - logged form: at = yt − (1 − α) kt − α lt (notation preserved as in source).
- Components yt and kt are in millions of constant 2015 euros; lt is in thousands of hours per quarter.
- Time-varying labor share αt is used for all three countries to capture observed changes (notably increases in Latvia and Lithuania since around 2015).
- Using a smooth, time-varying labor share introduces an extra term in the GDP growth decomposition that captures the effect of changes in the labor share (see source for exact expression).

### Cointegration between TFP and REER (Balassa-Samuelson assessment)
- The estimated TFP series is plugged into a cointegrating equation with the real effective exchange rate (REER) to assess the Balassa-Samuelson hypothesis.
- Both series tested for integration using Augmented Dickey-Fuller tests (constant and deterministic linear trend; lags selected by Schwartz information criterion).
- Johansen cointegration tests find at least one cointegrating relationship between TFP and REER for different specifications (with exogenous regressors and/or short-term dynamics).
  - Exception: for Latvia, no cointegrating relationship was found when using the actual REER series; the HP-trend of the REER series was used instead.
- Dynamic Ordinary Least Squares (DOLS) used for cointegrating regressions, with automatic lead/lag selection based on the Schwartz Information Criterion; selected specification chosen for positive sign and best in-sample R2 fit.

### Constructing TFP-based REER and interpreting REER gaps
- Fitted values from the cointegrating regressions are used to construct a TFP-based REER (the implied long-term relationship between TFP and REER).
- REER gap = actual REER − TFP-based REER.
  - A negative gap (actual below TFP-based) indicates an undervalued REER relative to its long-term norm → competitive advantage.
  - A positive gap indicates an overvalued REER relative to the TFP-based norm → competitiveness erosion.
- Figures referenced: Figure 3.1 (TFP estimates) and Figure 3.2 (Actual and TFP-Based REER in the Baltics, 1995-2023). (Country sample windows noted: Latvia: 2002Q2-2023Q2; Lithuania: 1998Q3-2023Q2.)

### Potential GDP and decomposition of cyclical vs structural drivers
- Potential GDP estimated via a multivariate Kalman filter (state-space model) combining quarterly GDP with high-frequency indicators:
  - Monthly confidence indices (consumer, industry, construction, retail), unemployment rate, industrial capacity utilization.
  - The unobserved trend in the state-space model is identified as potential GDP (ŷt).
- Structural (trend) TFP, ât, obtained by applying the production function to filtered (smooth trend / HP-filter or Kalman-filtered) series of effective capital (k̄t), employment (ēt), and hours worked (h̄t).
- The approach separates cyclical (short-term) and structural (low-frequency) components of TFP and GDP.

### Sample partitioning for analysis
- Sample divided into three subperiods:
  - Pre-GFC: 1995-2008.
  - Post-GFC: 2009-2019.
  - Post-pandemic: 2020-2023Q2.

### Key empirical findings and quantified outcomes
- Potential GDP growth decline (post-GFC compared with pre-GFC):
  - Fell from above 5 percent in the pre-GFC period to around:
    - 1 percent in Latvia (post-GFC),
    - 2 percent in Estonia (post-GFC),
    - between 2.5 and 3 percent in Lithuania (post-GFC).
- Structural "gross TFP" (GTFP = TFP growth + change in labor share) dynamics:
  - Estonia: GTFP growth declined from above 1 percent (pre-GFC) to negative growth (post-GFC).
  - Lithuania: trend GTFP growth accelerated over time toward positive territory.
  - Latvia: GTFP growth accelerated after the GFC but decelerated in the post-pandemic period and has become negative more recently.
- TFP and REER gap patterns:
  - Pre-GFC: Estonia and Latvia experienced prolonged negative REER gaps (actual REER below TFP-based REER) → competitive edge; Estonia’s negative gaps largely due to rapid TFP growth; Latvia’s due initially to a depreciating REER and low TFP growth. Lithuania’s REER gap turned negative only in 2005-2007.
  - Post-GFC: Competitive advantage eroded for Estonia and Latvia as TFP growth declined; negative REER gaps became smaller/shorter-lived and positive gaps became larger/more frequent.
  - Recent years (post-pandemic): Rapid real exchange rate appreciation caused a decoupling between actual and TFP-based REER for all three countries:
    - Latvia and especially Estonia: declining TFP growth pushed REER gaps into positive territory (overvaluation), with Estonia experiencing the largest positive gaps.
    - Lithuania: stronger TFP growth and a more undervalued REER provided larger buffers and helped retain export shares.
  - Late 2021 onward: Significant divergence in Estonia; REER gap turned positive as TFP deceleration became an outright decline combined with protracted real exchange rate appreciation—linked to reduced ability to absorb shocks, marked decline in export market shares, and weaker economic activity. Positive REER gaps materialized later and were smaller in Latvia and Lithuania.
- Role of growth components across subsamples (growth decomposition):
  - Capital accumulation: major driver of income convergence and a dominant contributor to GDP growth for Estonia and Lithuania over the full sample; contribution significantly declined post-GFC in Latvia.
  - Labor contribution: limited across the Baltics and acted as a drag on growth (especially in Latvia).
  - TFP contribution:
    - Estonia: TFP contributed significantly pre-GFC but faded and became negative post-GFC; decline in average TFP growth in Estonia explains almost 90 percent of the fall in potential GDP growth (see Table 3.1).
    - Latvia and Lithuania: TFP increasingly became a driver of growth for Latvia and especially Lithuania; Lithuania experienced acceleration in structural TFP growth since the GFC, helping shield it from recent external shocks.
- Scarring and potential GDP:
  - Estimates of potential GDP suggest scarring effects for all Baltics following the GFC, with a more pronounced impact of recent external shocks on Estonia; scarring effects after the pandemic are similar but less pronounced.

*Source: IMF staff calculations and analysis in "3.2 Data and methodology" (wpiea2025018-print-pdf).*

### 3.5 A few key takeaways

### 3.5 A few key takeaways

### Overall analytical focus
- Estimates produced: potential GDP, GDP growth decompositions, and TFP for the three Baltic countries.
- Uses: assess drivers of actual and potential GDP growth over 1995-2023 and the nexus between real exchange rate and TFP implied by the Balassa-Samuelson hypothesis.
- Competitiveness metric: an estimated TFP-based REER was used to measure competitiveness in the Baltics.
- Note: "Annex II shows a detailed decomposition, distinguishing between cyclical and structural factors."

### Key quantitative decomposition snippet (preserved exactly as in source)
- TotalTrendCyclicalTotalTrendC
yclicalTotalTrendCyclical
GDP-4.3-3.1-1.2-4.1-4.70.6-3.1-2.0-1.0
TFP-5.2-2.7-2.52.01.30.78.78.50.2
K-1.1-2.31.1-7.0-7.60.6-8.9-8.1-0.9
L1.31.30.10.20.9-0.7-0.5-0.1-0.4
Δα0.70.70.00.70.70.0-2.4-2.30.0
LithuaniaEstoniaLatvia

### Main findings on potential GDP and TFP
- Potential GDP growth in the Baltics has fallen since the GFC, largely due to a steady decline in TFP growth, especially for Estonia.
- Some decline in potential GDP growth is consistent with income convergence and decelerating capital accumulation, but:
  - The largest contributor to the reduction in potential GDP growth for Estonia has been a decline in TFP growth.
  - This decline is significantly more pronounced in Estonia than in Latvia and Lithuania.
  - Lithuania has experienced an acceleration in TFP growth in recent years (in contrast to Estonia).
- Estonia specifics:
  - TFP growth declined since the GFC.
  - Unlike Latvia and Lithuania, the level of TFP in Estonia has dropped since 2020.
  - The decline in TFP growth includes a structural component, likely linked to scarring effects of recent shocks.

### Real exchange rate, competitiveness, and country heterogeneity
- All three Baltic countries experienced significant price increases relative to trading partners and rapid real exchange appreciation following common external shocks.
- Long-term differences in TFP growth produced notable differences in external competitiveness and GDP growth, affecting each country's ability to absorb common shocks.
- Pre-GFC patterns:
  - Estonia’s fast TFP growth underpinned a competitive advantage, even with real exchange rate appreciation.
  - Latvia’s competitiveness, and to a lesser extent Lithuania’s, were mainly driven by periods of depreciating REER amidst low TFP growth.
- Post-GFC patterns:
  - Decelerating TFP growth combined with protracted REER appreciation eroded competitiveness across the Baltics, most strongly for Estonia.
  - Estonia faced the most pronounced decline in TFP growth among the Baltics; recent significant REER appreciation compounded declining TFP, turning into a competitive disadvantage.
- Competitive implications for Estonia:
  - A positive wedge between the actual and TFP-based REER emerged earlier, grew faster, and became wider in Estonia than in the other Baltics.
  - This wider wedge has affected Estonia’s competitive position more than in Latvia and Lithuania.
  - The loss of competitiveness may be an important factor in Estonia’s protracted economic downturn.

*IMF Working Papers — Competitiveness and Productivity in the Baltics: Common Shocks, Different Implications*

### 4.5 Exploring the relationship between allocative efficiency and regulation

### 4.5 Exploring the relationship between allocative efficiency and regulation

### Regulatory landscape and labor market protection
- Baltic economies have relatively light product market regulation compared to other advanced economies.
- There is variation in labor market regulation across the Baltics:
  - Latvia has more stringent labor market regulation compared to the rest of the Baltic economies and other advanced economies.
  - According to the OECD Employment Protection Legislation indicator, Latvia exhibits:
    - higher severance pay for low tenure employees;
    - stricter definitions of unfair dismissal including the exclusion of non-performance related reasons;
    - generous availability for re-instatement after an employee’s dismissal.
  - It is more costly to dismiss an individual worker under a regular contract in Latvia than other OECD economies.
- Indicators cited in the chapter:
  - Product market regulation (2018)
  - Employment protection legislation (2019)
  - Financial market liberalization (2014)

### Linkages between regulation, market liberalization, and allocative efficiency
- Allocative efficiency measures in the chapter:
  - The calculated allocative efficiency values are standardized using the USA as the benchmark.
  - Lithuania is noted as an outlier because administrative data for Lithuania represents small and micro firms to a greater extent.
- Structural reform implications:
  - Structural reforms may help improve allocative efficiency, support productivity growth, and retain competitiveness for the Baltic economies.
  - Less regulation in product markets and more liberalization in financial and labor markets are generally associated with better allocative efficiency (IMF, 2024).
- Comparative position:
  - Indicators of product market regulation and financial market liberalization place Baltic economies in a favorable position compared to other advanced and emerging market economies.
- Identified frictions and concerns:
  - Previous studies suggest distortions in the capital market hampered productivity growth of firms in Latvia and Lithuania (Benkovskis, 2015; Foda et al, 2024).
  - There is concern over tight credit conditions and limited access to finance by small firms.
  - Labor market measures that protect jobs in downturns may come at the cost of labor market flexibility.
    - Example: recent research on Estonia suggests government programs such as job retention schemes in response to the pandemic may have hampered efficient labor allocation and led to productivity losses (Meriküll and Paulus, 2024).

### Policy-relevant findings and recommendations
- Structural reforms that can improve allocative efficiency and productivity:
  - Reform product markets to reduce regulatory barriers.
  - Liberalize financial markets to improve access to finance, particularly for small firms.
  - Pursue labor market reforms to balance job protection with labor market flexibility.
- Rationale:
  - Improving allocative efficiency through these reforms would support productivity growth and enhance competitiveness for Estonia, Latvia, and Lithuania.

### Key empirical takeaways (summary from section 4.6)
- Resource misallocation has negatively affected total factor productivity (TFP) growth in all three Baltic economies on average throughout the sample periods.
- Estonia:
  - Allocative efficiency worsened generally over time, despite a limited and short-lived recovery after the global financial crisis (GFC).
  - In industries such as real estate, allocative efficiency deteriorated in the years leading up to the GFC.
- Latvia:
  - Allocative efficiency has slightly improved after 2016, but the recovery has been limited.
  - For some industries such as real estate and transport, allocative efficiency worsened more significantly during the early period of 2012-2015.
- Lithuania:
  - Resource misallocation has generally worsened over time.
- Sectoral pattern:
  - Productivity loss due to resource misallocation is more pronounced for services than for goods sectors.
- Overall policy conclusion:
  - Structural reforms in product, capital, and labor markets can help improve allocative efficiency, promote productivity growth, and enhance competitiveness for the Baltic economies.

*Source: IMF Working Paper — chapter 4.5, “Exploring the relationship between allocative efficiency and regulation.”*

### Annex II.  Additional Estimation Material

### Annex II. Additional Estimation Material

### Johansen cointegration tests — rank selection and deterministic cases
- Rank selection performed at the 0.05 level using critical values from MacKinnon-Haug-Michelis (1999).
- Deterministic Cases described:
  - Case 1: No deterministic terms.
  - Case 2: Cointegrating relationship includes a constant.
  - Case 3 (Johansen-Hendry-Juselius): Cointegrating relationship includes a constant. Short-run dynamics include a constant.
  - Case 4 (Johansen-Hendry-Juselius): Cointegrating relationship includes a constant and trend. Short-run dynamics include a constant.
  - Case 4 (alternative description): Cointegrating relationship includes a trend. Short-run dynamics include a constant.
  - Case 5 (Johansen-Hendry-Juselius): Both the cointegrating relationship and short-run dynamics include a constant and trend.
  - Case 5 (alternative description): Short-run dynamics include a constant and trend.
- Rank selection results (by Test and Deterministic Case), reported exactly as in source:
  - ln(TFP) Trace: 1 1 1 1 1 2 2
  - ln(REER) Max-Eigen: 1 1 1 1 1 2 2
  - ln(TFP) Trace: 1 1 1 1 1 2 2
  - ln(REER-HP) Max-Eigen: 1 1 1 1 1 2 2
  - ln(TFP) Trace: 2 1 1 2 2 2 2
  - ln(REER) Max-Eigen: 2 1 1 2 2 0 0
- Endogenous variables and sample details (reported exactly):
  - Countries: Estonia, Latvia, Lithuania
  - Samples: 1995Q4-2023Q2; 2002Q4-2023Q2; 1999Q2-2023Q2
  - No. of Observations: 111; 83; 97

### Cointegration between REER and TFP — DOLS estimates (Table All.2)
- Dependent variable: Ln(REER) for Estonia, Latvia, Lithuania.
- Method: DOLS for all three countries.
- Samples (adjusted) and Observations (adjusted):
  - Estonia: 1996Q4-2023Q2, Observations 107
  - Latvia: 2002Q2-2023Q2, Observations 85
  - Lithuania: 1998Q4-2023Q2, Observations 99
- Cointegrating equation deterministic factors:
  - Estonia: C, @TREND
  - Latvia: C, @TREND
  - Lithuania: C, @TREND (formatted in source with extra spacing)
- Automatic Leads: Estonia 0 1 1 0; Latvia 0 1 1 0; Lithuania (implied)
- Automatic Lags: Estonia 6 1 0; Latvia (implied)
- Criterion: SIC for all (reported as SIC SIC SIC)
- Key coefficient estimates and statistics (exactly as reported):
  - Ln(TFP) coefficient:
    - Estonia: 0.47; Std. Error 0.10; t-Statistic 4.77; Prob. 0.000
    - Latvia: 0.30; Std. Error 0.07; t-Statistic 4.46; Prob. 0.000
    - Lithuania: 0.03; Std. Error 0.01; t-Statistic 2.49; Prob. 0.01
  - C (constant) coefficient:
    - Estonia: 117.87; Std. Error 29.79; t-Statistic 3.96; Prob. 0.000
    - Latvia: -30.85; Std. Error 6.56; t-Statistic -4.70; Prob. 0.000
    - Lithuania: -4.93; Std. Error 1.49; t-Statistic -3.31; Prob. 0.00
  - @TREND coefficient:
    - Estonia: 0.32; Std. Error 0.02; t-Statistic 18.80; Prob. 0.000
    - Latvia: 0.05; Std. Error 0.02; t-Statistic 2.87; Prob. 0.01
  - Ln(REER-1) reported as 0.88 (context in table)
  - Ln(REER-1) coefficient (reported separately): 0.05; Std. Error 18.55; t-Statistic 0.00; Prob. (blank in source)
- Goodness-of-fit and residual statistics (exact values reported):
  - R-squared:
    - Estonia: 0.96
    - Latvia: 0.48
    - Lithuania: 0.98
  - Adjusted R-squared:
    - Estonia: 0.96
    - Latvia: 0.47
    - Lithuania: 0.98
  - S.E. of regression:
    - Estonia: 2.48
    - Latvia: 6.04
    - Lithuania: 1.59
  - Long-run variance:
    - Estonia: 18.70
    - Latvia: 136.26
    - Lithuania: 2.69
  - Mean dependent var:
    - Estonia: -3.06
    - Latvia: -1.97
    - Lithuania: -0.87
  - S.D. dependent var:
    - Estonia: 12.45
    - Latvia: 8.30
    - Lithuania: 10.37
  - Sum squared resid:
    - Estonia: 595.72
    - Latvia: 2,989.53
    - Lithuania: 236.88

### Decomposition of GDP growth — Estonia (Table AII.3), Latvia (Table AII.4), Lithuania (Table AII.5)
- Notes: All entries are in percentage points; growth rates calculated as difference in natural logarithms of original series. Values reproduced exactly as presented.

- Estonia (periods and selected lines; values exactly as presented):
  - Period headings (as in source): 1995-2008; 2009-2019; 2020-2023Q2
  - Selected rows (Total / Structural / Cyclical):
    - TFP:
      - Total: 30.7; Structural: 32.2; Cyclical: -1.6
      - 2009-2019 Total: 1.4; Structural: -3.6; Cyclical: 5.0
      - 2020-2023Q2 Total: -10.5; Structural: -1.6; Cyclical: -9.0
    - Capital:
      - Total: 57.6; Structural: 0.2; Cyclical: -2.6
      - 2009-2019 Total: 23.5; Structural: 23.1; Cyclical: 0.4
      - 2020-2023Q2 Total: 10.4; Structural: 7.0; Cyclical: 3.3
    - Capital accumulation:
      - Total: 47.9; Structural: 47.9; Cyclical: n.a
      - 2009-2019 Total: 20.8; Structural: 20.8; Cyclical: n.a
      - 2020-2023Q2 Total: 8.4; Structural: 8.4; Cyclical: n.a
    - Capacity utilization:
      - Total: 9.7; Structural: 12.3; Cyclical: -2.6
      - 2009-2019 Total: 2.7; Structural: 2.3; Cyclical: 0.4
      - 2020-2023Q2 Total: 2.0; Structural: -1.4; Cyclical: 3.3
    - Labor:
      - Total: -3.1; Structural: -5.2; Cyclical: 2.1
      - 2009-2019 Total: -2.0; Structural: -0.1; Cyclical: -1.8
      - 2020-2023Q2 Total: 3.9; Structural: 3.1; Cyclical: 0.8
    - Labor force:
      - Total: -0.3; Structural: -1.6; Cyclical: 1.2
      - 2009-2019 Total: 0.3; Structural: 1.6; Cyclical: -1.4
      - 2020-2023Q2 Total: 2.8; Structural: 2.2; Cyclical: 0.6
    - (-) Unemployment:
      - Total: 0.8; Structural: 0.2; Cyclical: 0.6
      - 2009-2019 Total: 1.6; Structural: 1.5; Cyclical: 0.1
      - 2020-2023Q2 Total: -1.0; Structural: -0.3; Cyclical: -0.7
    - Hours-worked:
      - Total: -3.6; Structural: -3.8; Cyclical: 0.2
      - 2009-2019 Total: -3.9; Structural: -3.3; Cyclical: -0.6
      - 2020-2023Q2 Total: 2.1; Structural: 1.2; Cyclical: 0.8
    - Δ Labor Share:
      - Total: -14.2; Structural: -14.2; Cyclical: 0.0
      - 2009-2019 Total: 3.5; Structural: 3.4; Cyclical: 0.1
      - 2020-2023Q2 Total: -1.1; Structural: -1.1; Cyclical: 0.0
    - GDP:
      - Total: 70.9; Structural: 73.0; Cyclical: -2.1
      - 2009-2019 Total: 26.4; Structural: 22.7; Cyclical: 3.6
      - 2020-2023Q2 Total: 2.7; Structural: 7.5; Cyclical: -4.8
    - Average Annual Growth:
      - GDP: 5.1; 5.2; -0.1; 2.4; 2.1; 0.3; 0.8; 2.2; -1.4 (presented across subperiods in source)
      - TFP: 2.2; 2.3; -0.1; 0.1; -0.3; 0.5; -3.0; -0.4; -2.6
      - Gross TFP (= TFP + Δα): 1.2; 1.3; -0.1; 0.4; 0.0; 0.5; -3.3; -0.7; -2.6
      - Capital: 4.1; 4.3; -0.2; 2.1; 2.1; 0.0; 3.0; 2.0; 1.0
      - Labor: -0.2; -0.4; 0.1; -0.2; 0.0; -0.2; 1.1; 0.9; 0.2
      - Δ Labor Share: -1.0; -1.0; 0.0; 0.3; 0.3; 0.0; -0.3; -0.3; 0.0

- Latvia (selected lines reproduced exactly as in table):
  - Period headings in source: 1998Q3-2008; 2009-2019; 2020-2023Q2
  - Selected rows and values:
    - TFP:
      - Total: 3.5; Structural: 4.0; Cyclical: -0.5
      - 2009-2019 Total: 37.9; Structural: 36.6; Cyclical: 1.3
      - 2020-2023Q2 Total: 8.8; Structural: 6.6; Cyclical: 2.2
    - Capital:
      - Total: 46.8; Structural: 51.3; Cyclical: -4.5
      - 2009-2019 Total: 6.6; Structural: 3.6; Cyclical: 3.0
      - 2020-2023Q2 Total: 0.7; Structural: 1.0; Cyclical: -0.3
    - Capital accumulation:
      - Total: 52.2; Structural: 52.2; Cyclical: n.a
      - 2009-2019 Total: -4.0; Structural: -4.0; Cyclical: n.a
      - 2020-2023Q2 Total: 1.5; Structural: 1.5; Cyclical: n.a
    - Capacity utilization:
      - Total: -5.4; Structural: -0.9; Cyclical: -4.5
      - 2009-2019 Total: 10.6; Structural: 7.6; Cyclical: 3.0
      - 2020-2023Q2 Total: -0.7; Structural: -0.4; Cyclical: -0.3
    - Labor:
      - Total: 6.6; Structural: 0.2; Cyclical: 6.4
      - 2009-2019 Total: -14.1; Structural: -7.0; Cyclical: -7.1
      - 2020-2023Q2 Total: 4.4; Structural: 3.3; Cyclical: 1.1
    - Labor force:
      - Total: 1.3; Structural: -0.2; Cyclical: 1.5
      - 2009-2019 Total: -6.9; Structural: -5.1; Cyclical: -1.8
      - 2020-2023Q2 Total: -1.2; Structural: -0.5; Cyclical: -0.7
    - (-) Unemployment:
      - Total: 1.3; Structural: 0.5; Cyclical: 0.7
      - 2009-2019 Total: 2.0; Structural: 1.8; Cyclical: 0.2
      - 2020-2023Q2 Total: 0.0; Structural: 0.6; Cyclical: -0.6
    - Hours-worked:
      - Total: 4.1; Structural: -0.1; Cyclical: 4.2
      - 2009-2019 Total: -9.2; Structural: -3.7; Cyclical: -5.5
      - 2020-2023Q2 Total: 5.6; Structural: 3.1; Cyclical: 2.5
    - Net L/K Share Shift:
      - Total: -18.0; Structural: -18.2; Cyclical: 0.2
      - 2009-2019 Total: -19.2; Structural: -19.4; Cyclical: 0.3
      - 2020-2023Q2 Total: -7.2; Structural: -7.2; Cyclical: 0.0
    - GDP:
      - Total: 38.9; Structural: 37.2; Cyclical: 1.7
      - 2009-2019 Total: 11.3; Structural: 13.9; Cyclical: -2.6
      - 2020-2023Q2 Total: 6.7; Structural: 3.7; Cyclical: 3.0
    - Average Annual Growth:
      - GDP: 6.0; 5.7; 0.3; 1.0; 1.3; -0.2; 1.9; 1.0; 0.9
      - TFP: 0.5; 0.6; -0.1; 3.4; 3.3; 0.1; 2.5; 1.9; 0.6
      - Gross TFP (= TFP + Δα): -2.2; -2.2; 0.0; 1.7; 1.6; 0.1; 0.5; -0.2; 0.6
      - Capital: 7.2; 7.9; -0.7; 0.6; 0.3; 0.3; 0.2; 0.3; -0.1
      - Labor: 1.0; 0.0; 1.0; -1.3; -0.6; -0.6; 1.2; 0.9; 0.3
      - Δ Labor Share: -2.8; -2.8; 0.0; -1.7; -1.8; 0.0; -2.1; -2.1; 0.0

- Lithuania (selected lines reproduced exactly as in table):
  - Period headings in source: 1998Q3-2008; 2009-2019; 2020-2023Q2
  - Selected rows and values:
    - TFP:
      - Total: -48.5; Structural: -49.0; Cyclical: 0.6
      - 2009-2019 Total: 10.4; Structural: 11.1; Cyclical: -0.7
      - 2020-2023Q2 Total: 14.7; Structural: 13.7; Cyclical: 1.1
    - Capital:
      - Total: 115.0; Structural: 114.0; Cyclical: 1.0
      - 2009-2019 Total: 39.2; Structural: 39.5; Cyclical: -0.3
      - 2020-2023Q2 Total: 6.1; Structural: 8.9; Cyclical: -2.8
    - Capital accumulation:
      - Total: 97.3; Structural: 97.3; Cyclical: n.a
      - 2009-2019 Total: 33.5; Structural: 33.5; Cyclical: n.a
      - 2020-2023Q2 Total: 11.0; Structural: 11.0; Cyclical: n.a
    - Capacity utilization:
      - Total: 17.7; Structural: 16.6; Cyclical: 1.0
      - 2009-2019 Total: 5.7; Structural: 6.0; Cyclical: -0.3
      - 2020-2023Q2 Total: -4.9; Structural: -2.1; Cyclical: -2.8
    - Labor:
      - Total: 2.0; Structural: 2.2; Cyclical: -0.2
      - 2009-2019 Total: -4.6; Structural: -1.4; Cyclical: -3.2
      - 2020-2023Q2 Total: -1.0; Structural: 0.4; Cyclical: -1.4
    - Labor force:
      - Total: -5.0; Structural: -5.6; Cyclical: 0.6
      - 2009-2019 Total: -1.5; Structural: -0.4; Cyclical: -1.1
      - 2020-2023Q2 Total: 1.9; Structural: 1.3; Cyclical: 0.6
    - (-) Unemployment:
      - Total: 1.9; Structural: 2.1; Cyclical: -0.2
      - 2009-2019 Total: 0.6; Structural: 0.9; Cyclical: -0.3
      - 2020-2023Q2 Total: 0.0; Structural: -0.4; Cyclical: 0.4
    - Hours-worked:
      - Total: 5.1; Structural: 5.7; Cyclical: -0.6
      - 2009-2019 Total: -3.7; Structural: -2.0; Cyclical: -1.8
      - 2020-2023Q2 Total: -3.0; Structural: -0.5; Cyclical: -2.4
    - Net L/K Share Shift:
      - Total: -12.5; Structural: -12.5; Cyclical: 0.0
      - 2009-2019 Total: -22.4; Structural: -22.5; Cyclical: 0.1
      - 2020-2023Q2 Total: -12.3; Structural: -12.3; Cyclical: 0.0
    - GDP:
      - Total: 56.1; Structural: 54.6; Cyclical: 1.5
      - 2009-2019 Total: 22.6; Structural: 26.6; Cyclical: -4.1
      - 2020-2023Q2 Total: 7.5; Structural: 10.7; Cyclical: -3.2
    - Average Annual Growth:
      - GDP: 5.2; 5.1; 0.1; 2.1; 2.4; -0.4; 2.1; 3.0; -0.9
      - TFP: -4.5; -4.6; 0.1; 0.9; 1.0; -0.1; 4.2; 3.9; 0.3
      - Gross TFP (= TFP + Δα): -5.7; -5.7; 0.1; -1.1; -1.0; 0.0; 0.7; 0.4; 0.3
      - Capital: 10.7; 10.6; 0.1; 3.6; 3.6; 0.0; 1.8; 2.5; -0.8
      - Labor: 0.2; 0.2; 0.0; -0.4; -0.1; -0.3; -0.3; 0.1; -0.4
      - Δ Labor Share: -1.2; -1.2; 0.0; -2.0; -2.0; 0.0; -3.5; -3.5; 0.0

### Methodology — Deriving allocative efficiency (Annex III)
- Framework and assumptions (exact terminology preserved):
  - Cobb-Douglas production function at the firm level with 훼훼_cccc representing the country-sector specific capital share. Y_cccc, A_cccc, K_cccc, and L_cccc represent output, technology, capital, and labor at the firm level. Subscripts c, s, i, and t represent country, sector, firm, and year, respectively.
  - Aggregation with constant elasticity of substitution, where 휎휎_cc represents the elasticity of substitution. Lower case i indicates sector i.
  - Distortions on output, capital, and labor markets. Distortions on capital and labor markets increase the effective cost of capital and labor by 휏휏_K and 휏휏_L, respectively. 휏휏_Y represents a tax on output. 휏휏_ccccc is defined as a function of the distortions on capital, labor, and output markets.
  - Firms maximize profits under monopolistic competition by choosing K and L such that marginal revenue product equals marginal cost (Equations (4) and (5)). r_cccc and w_cccc are the cost of capital and labor at the sector level. Firms’ output price is a fixed markup over marginal cost.
  - Equation (6) gives the output in equilibrium; equation (7) shows marginal revenue products of capital and labor will not be equalized due to distortions.
  - In absence of distortions (휏휏_ccccc and 휏휏_cccc both equal 1), sector-level TFP aggregates the firm-level technology component A (Equations (8) and (9)). With distortions, total factor productivity becomes lower; the wedge is AE in Equation (10).
  - Relationship between allocative efficiency and TFP growth: "For each unit decline in allocative efficiency, there will be a one-percentage point decline in TFP growth."
- Aggregation and decomposition procedure:
  - Aggregate firm-level allocative efficiency to the sector level.
  - Aggregate sector-level allocative efficiency to the country level using sectoral value added shares from the EUKLEMS database.
  - Calculate annual TFP growth using the country-level TFP index from the AMECO database.
  - Decompose TFP growth into innovation and allocative efficiency components using the aggregated allocative efficiency based on firm-level data; the two components correspond to the two terms on the right-hand side of Equation (11).

*Source: IMF staff estimates, Annex II and Annex III, Competitiveness and Productivity in the Baltics: Common Shocks, Different Implications, Working Paper No. WP/2025/018*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025018-print-pdf.pdf_