## Conclusions and Policy Implications ## Source details **Canonical URL:** [Conclusions and Policy Implications](https://www.imf.org/-/media/files/publications/cr/2019/1lvaea2019002.pdf) ## Other formats - [Markdown version](/-/media/files/publications/cr/2019/1lvaea2019002.pdf.md) - [Structured JSON version](/-/media/files/publications/cr/2019/1lvaea2019002.pdf.json) --- ### Overview and context - Latvia is relatively well integrated in global value chains (GVCs); participation is lower compared to other Baltic countries but has increased over time. - The analysis employs a structural framework developed by Bems and Johnson (2017) and uses the 2016 vintage of the World Input-Output Database (WIOD, Timmer et al., 2015). - Given high persistency in the weights, the analysis assumes weights remain constant from 2014 through 2018. ### Latvia’s participation in GVCs — key patterns - Backward participation increase driven mainly by manufacturing of basic metals, rubber and plastic products, transport equipment, and chemical products, as well as air transport. - Latvia incorporates foreign value added mainly from Russia, Germany, Lithuania, Sweden and Finland. - Forward participation concentrated in manufacturing of wood and basic metals, land transport and transport via pipeline, wholesale trade (except of motor vehicles and motorcycles), and crop and animal production. - Forward GVC participation is mainly with Estonia, Lithuania, Russia, and Belarus. - Implication: Latvia’s competitiveness is likely sensitive to price changes in Russia and Germany and demand from Estonia and Lithuania. ### REER measures and methodological approach - Two REER measures compared: - Conventional REER: weights based on gross trade flow shares (double export weights from Armington CES demand). - Value-Added REER (VA-REER): replaces conventional weights with weights based on trade in value added; derived from structural model distinguishing gross output and value-added and accounting for bilateral value-added flows. - VA-REER constructed using WIOD and Bems and Johnson (2017) framework; full-input-output version assumes Leontief production (elasticity of substitution across inputs and between inputs and value added = 0), placing more weight on final goods and lower weight on strong bilateral input linkages. ### Comparative findings on weights and REER movements - Value added weights of Latvia’s traditional trading partners are lower than conventional weights: - VA weights of Lithuania and Estonia, Germany, Sweden, and Russia are about 2.3 percentage points lower than their conventional rates. - USA and China’s weight are 2 percentage points higher under VA weights. - VA-REER has appreciated more rapidly than the conventional REER since 2010. - Correlation between VA-REER and unit labor cost (ULC) is 0.9, suggesting competitiveness in supplying domestic value added is highly dependent on labor cost increases. ### Decomposition of the VA-REER vs Conventional REER gap - The REER gap decomposes into a price component and a weight component. - About 70 percent of the REER gap is explained by the price component; the remaining 30 percent by the weight component. - Further decomposition of the gap: - Value added weights account for 21 percent of the gap. - Elasticities (role of rigid production chains) account for 9 percent of the gap. - The price component is driven primarily by differences between Latvia’s GDP deflator and CPI. ### Price divergences and drivers - Cumulative changes since 2010: - GDP deflator has grown cumulatively by 23 percent since 2010. - CPI has increased by 13.6 percent during the same period. - Unit labor costs have remained a steady driver of final output prices, increasing by about 5 percent annually since 2012. - Early gap opening attributed to a rise in prices of capital goods; prices of capital goods largely stabilized since 2012. - Rapid growth in Latvia’s terms of trade may have induced an increase in profit margins, reflected in the unit profits component of the GDP deflator. ### Exposure to shocks in a GVC environment — empirical estimates - Empirical strategy: local projection approach à la Jordá (2005) with panel data of 27 European countries over 2003–13; control variables include inflation, real GDP per capita, net FDI inflows and external demand (external demand measured as weighted growth rate of trading partners). - Estimated impulse responses (panel results): - A 10-percentage point appreciation (increase in the VA REER index) leads to a statistically significant reduction in value-added export growth by 0.4 percentage point in the first year. - The effect cumulates to 0.5 percentage point in the second year. - Robustness: inflation excluded as a control in a robustness check with similar results. - Implication: appreciations of the VA-REER meaningfully reduce value-added export growth and can affect real GDP growth through trade channels. ### The Effect of VA-REER Shocks on Real Value-Added Export Growth - Impact of VA-REER appreciation on value-added (VA) export growth: - A 10-percentage point appreciation in the VA REER index leads to a reduction in VA export growth by 0.6 percentage point in the first year and cumulates up to 1 percentage point in the third year in countries that are above the sample median of GVC participation index. - Regressions include the full list of control variables, as well as country fixed effects and year fixed effects. Year 0 is the year of the shock. The local projection method was corrected following Teulings and Zubanov (2014). Dashed lines show the 95-percent confidence interval. - Transmission from VA export growth to real GDP growth: - Using the Local projection specification, a 1 percentage point increase in real value-added export is associated with a 0.3 percentage point increase in real GDP growth cumulatively over the 4 years. - Using these estimates, a 10 percent appreciation in VA-REER could reduce growth rate by 0.2 percentage point (calculation noted: 0.3 * (-0.5)). - For countries above the sample median of the GVC participation index, the implied impact is larger and could reach 0.3 percentage point loss in growth. - The reduced-form effect of VA-REER on real GDP growth was also estimated and found no statistically significant effect. No statistically significant effect of conventional REER on value added export growth or real GDP growth was found. ### Latvia-specific estimates and historical context - Since 2010, Latvia’s VA REER has appreciated by roughly 5 percent. - Using the model estimates, this implies: - a reduction in value added export by 0.2 percent, and - a reduction in growth by 0.1 percentage point. - Absent the appreciation in VA REER since 2010 (and the rise in ULC owing to strong pass-through), growth could have been higher by 0.1 percentage point in Latvia. - The modest effect in Latvia is attributed to relatively low GVC participation over the period in the sample. ### Transmission of a tariff shock through global value chains (GVCs) - A tariff penalizes assemblers and suppliers in a value chain, amplifying trade costs and potentially affecting competitiveness across the entire value chain. Tariffs on goods can also spill over to the service sector. - Trade tensions can impact global economic growth through direct factors (higher trade costs) and indirect factors (lower business confidence, weaker private sector investment, tighter financial conditions). - The structural model of Bems and Johnson (2017) is used to estimate short-run partial-equilibrium impacts of tariff-induced relative price changes on gross trade and value-added flows. The model: - assumes a single price of output per country, - proceeds sequentially to calculate tariff-induced price changes and demand responses, - uses elasticities from the structural model, - holds countries’ real expenditure levels constant, - does not account for long-run supply chain realignment. ### Latvia’s exposure to trade tensions and modeled tariff scenarios - Bilateral weights implied by the VA REER are higher than conventional ones for China, USA, and UK, indicating weaker input linkages and competition mainly on final goods rather than on inputs. - Accounting for trade in value-added, Latvia is more exposed to external trade shocks originating in China, USA, and UK than captured by gross trade. - Model estimates: a 5.9-percent tariff imposed by the US on its imports (Layer 1), with retaliation from all countries using the same tariff, would lead to a reduction of 0.2 percent in Latvia’s value added (four times larger than the reduction in gross turnover flows). - Layer definitions and associated aggregate tariff magnitudes reported: - Layer 1: 5.9% - Layer 2: 3.8% - Layer 3: 5.0% - Cumulative: 14.7% - Latvia’s exposure is relatively moderate compared to the European (EU28) average; Germany exhibits the largest exposure (domestically produced value added reduction about 50 percent higher than the average EU impact), while Latvia’s impact is slightly smaller than the EU average. ### Policy implications and risks highlighted - The VA REER index accounting for input-output linkages suggests Latvia may be less competitive than indicated by a standard REER index based on gross trade. - Rapid labor market tightening that raises unit labor costs could erode competitiveness in supplying domestic value added and reduce growth given the high sensitivity of VA-REER to labor costs (ULC correlation = 0.9). - Trade-tension induced tariff hikes could have significant cost for Latvia, especially in terms of value added produced in the country, given GVC linkages. - Given VA weights shift importance toward USA and China relative to some European partners, shifts in demand or prices in those economies can have disproportionate effects on Latvia’s competitiveness when measured on a value-added basis. - Because about 70 percent of the REER gap is driven by price differentials (GDP deflator vs CPI), policies that influence labor costs, profit margins, and terms of trade are central for competitiveness in value-added terms. ### Policy measures to mitigate exposure and improve competitiveness in GVCs - Enhance product sophistication or quality and pursue export market diversification to mitigate exposure to trade shocks in GVCs. - Backward GVC participation: increase use of imported intermediate goods with high-technological content to improve product sophistication; Latvia currently incorporates significant foreign value added but less than peer countries. - Forward GVC participation: promote upstream activities (exporting intermediate goods, re-exports, non-transport services) to generate productivity and employment gains; improve allocation and incentives for innovation through better access to credit and skilled labor with knowledge of foreign markets. - Attract FDI inflows to spur GVC participation through intra-firm trade channels; FDI can increase trade in intermediate and final goods and is an important determinant of backward participation. FDI targeting export processing facilities could improve product diversity and sophistication. - Prevent long-term misalignment between wage growth and productivity to preserve competitiveness in supplying domestic value added. *Conclusions and Policy Implications — Republic of Latvia, International Monetary Fund (Content unit: 1lvaea2019002)* ### Conclusions and Policy Implications _________________________________________________ 14 ### Conclusions and Policy Implications ### Overview and context - Latvia is relatively well integrated in global value chains (GVCs); participation is lower compared to other Baltic countries but has increased over time. - The analysis employs a structural framework developed by Bems and Johnson (2017) and uses the 2016 vintage of the World Input-Output Database (WIOD, Timmer et al., 2015). - Given high persistency in the weights, the analysis assumes weights remain constant from 2014 through 2018. ### Latvia’s participation in GVCs — key patterns - Backward participation increase driven mainly by manufacturing of basic metals, rubber and plastic products, transport equipment, and chemical products, as well as air transport. - Latvia incorporates foreign value added mainly from Russia, Germany, Lithuania, Sweden and Finland. - Forward participation concentrated in manufacturing of wood and basic metals, land transport and transport via pipeline, wholesale trade (except of motor vehicles and motorcycles), and crop and animal production. - Forward GVC participation is mainly with Estonia, Lithuania, Russia, and Belarus. - Implication: Latvia’s competitiveness is likely sensitive to price changes in Russia and Germany and demand from Estonia and Lithuania. ### REER measures and methodological approach - Two REER measures compared: - Conventional REER: weights based on gross trade flow shares (double export weights from Armington CES demand). - Value-Added REER (VA-REER): replaces conventional weights with weights based on trade in value added; derived from structural model distinguishing gross output and value-added and accounting for bilateral value-added flows. - VA-REER constructed using WIOD and Bems and Johnson (2017) framework; full-input-output version assumes Leontief production (elasticity of substitution across inputs and between inputs and value added = 0), placing more weight on final goods and lower weight on strong bilateral input linkages. ### Comparative findings on weights and REER movements - Value added weights of Latvia’s traditional trading partners are lower than conventional weights: - VA weights of Lithuania and Estonia, Germany, Sweden, and Russia are about 2.3 percentage points lower than their conventional rates. - USA and China’s weight are 2 percentage points higher under VA weights. - VA-REER has appreciated more rapidly than the conventional REER since 2010. - Correlation between VA-REER and unit labor cost (ULC) is 0.9, suggesting competitiveness in supplying domestic value added is highly dependent on labor cost increases. ### Decomposition of the VA-REER vs Conventional REER gap - The REER gap decomposes into a price component and a weight component. - About 70 percent of the REER gap is explained by the price component; the remaining 30 percent by the weight component. - Further decomposition of the gap: - Value added weights account for 21 percent of the gap. - Elasticities (role of rigid production chains) account for 9 percent of the gap. - The price component is driven primarily by differences between Latvia’s GDP deflator and CPI. ### Price divergences and drivers - Cumulative changes since 2010: - GDP deflator has grown cumulatively by 23 percent since 2010. - CPI has increased by 13.6 percent during the same period. - Unit labor costs have remained a steady driver of final output prices, increasing by about 5 percent annually since 2012. - Early gap opening attributed to a rise in prices of capital goods; prices of capital goods largely stabilized since 2012. - Rapid growth in Latvia’s terms of trade may have induced an increase in profit margins, reflected in the unit profits component of the GDP deflator. ### Exposure to shocks in a GVC environment — empirical estimates - Empirical strategy: local projection approach à la Jordá (2005) with panel data of 27 European countries over 2003–13; control variables include inflation, real GDP per capita, net FDI inflows and external demand (external demand measured as weighted growth rate of trading partners). - Estimated impulse responses (panel results): - A 10-percentage point appreciation (increase in the VA REER index) leads to a statistically significant reduction in value-added export growth by 0.4 percentage point in the first year. - The effect cumulates to 0.5 percentage point in the second year. - Robustness: inflation excluded as a control in a robustness check with similar results. - Implication: appreciations of the VA-REER meaningfully reduce value-added export growth and can affect real GDP growth through trade channels. ### Policy implications and risks highlighted - Rapid labor market tightening that raises unit labor costs could erode competitiveness in supplying domestic value added and reduce growth given the high sensitivity of VA-REER to labor costs (ULC correlation = 0.9). - Trade-tension induced tariff hikes could have significant cost for Latvia, especially in terms of value added produced in the country, given GVC linkages. - Given VA weights shift importance toward USA and China relative to some European partners, shifts in demand or prices in those economies can have disproportionate effects on Latvia’s competitiveness when measured on a value-added basis. - Because about 70 percent of the REER gap is driven by price differentials (GDP deflator vs CPI), policies that influence labor costs, profit margins, and terms of trade are central for competitiveness in value-added terms. *Conclusions and Policy Implications — Republic of Latvia, International Monetary Fund (Content unit: 1lvaea2019002)* ### 69.8 versus the subsample below this sample median. The results provide evidence that a 10-percentage point ### The Effect of VA-REER Shocks on Real Value-Added Export Growth ### Impact of VA-REER appreciation on value-added (VA) export growth - A 10-percentage point appreciation in the VA REER index leads to a reduction in VA export growth by 0.6 percentage point in the first year and cumulates up to 1 percentage point in the third year in countries that are above the sample median of GVC participation index. - Regressions include the full list of control variables, as well as country fixed effects and year fixed effects. Year 0 is the year of the shock. The local projection method was corrected following Teulings and Zubanov (2014). Dashed lines show the 95-percent confidence interval. ### Transmission from VA export growth to real GDP growth - Using the Local projection specification, a 1 percentage point increase in real value-added export is associated with a 0.3 percentage point increase in real GDP growth cumulatively over the 4 years. - Using these estimates, a 10 percent appreciation in VA-REER could reduce growth rate by 0.2 percentage point (calculation noted: 0.3 * (-0.5)). - For countries above the sample median of the GVC participation index, the implied impact is larger and could reach 0.3 percentage point loss in growth. - The reduced-form effect of VA-REER on real GDP growth was also estimated and found no statistically significant effect. No statistically significant effect of conventional REER on value added export growth or real GDP growth was found. ### Latvia-specific estimates and historical context - Since 2010, Latvia’s VA REER has appreciated by roughly 5 percent. - Using the model estimates, this implies: - a reduction in value added export by 0.2 percent, and - a reduction in growth by 0.1 percentage point. - Absent the appreciation in VA REER since 2010 (and the rise in ULC owing to strong pass-through), growth could have been higher by 0.1 percentage point in Latvia. - The modest effect in Latvia is attributed to relatively low GVC participation over the period in the sample. ### Transmission of a tariff shock through global value chains (GVCs) - A tariff penalizes assemblers and suppliers in a value chain, amplifying trade costs and potentially affecting competitiveness across the entire value chain. Tariffs on goods can also spill over to the service sector. - Trade tensions can impact global economic growth through direct factors (higher trade costs) and indirect factors (lower business confidence, weaker private sector investment, tighter financial conditions). - The structural model of Bems and Johnson (2017) is used to estimate short-run partial-equilibrium impacts of tariff-induced relative price changes on gross trade and value-added flows. The model: - assumes a single price of output per country, - proceeds sequentially to calculate tariff-induced price changes and demand responses, - uses elasticities from the structural model, - holds countries’ real expenditure levels constant, - does not account for long-run supply chain realignment. ### Latvia’s exposure to trade tensions and modeled tariff scenarios - Bilateral weights implied by the VA REER are higher than conventional ones for China, USA, and UK, indicating weaker input linkages and competition mainly on final goods rather than on inputs. - Accounting for trade in value-added, Latvia is more exposed to external trade shocks originating in China, USA, and UK than captured by gross trade. - Model estimates: a 5.9-percent tariff imposed by the US on its imports (Layer 1), with retaliation from all countries using the same tariff, would lead to a reduction of 0.2 percent in Latvia’s value added (four times larger than the reduction in gross turnover flows). - Layer definitions and associated aggregate tariff magnitudes reported: - Layer 1: 5.9% - Layer 2: 3.8% - Layer 3: 5.0% - Cumulative: 14.7% - Latvia’s exposure is relatively moderate compared to the European (EU28) average; Germany exhibits the largest exposure (domestically produced value added reduction about 50 percent higher than the average EU impact), while Latvia’s impact is slightly smaller than the EU average. ### Conclusions and policy implications - The VA REER index accounting for input-output linkages suggests Latvia may be less competitive than indicated by a standard REER index based on gross trade. - Recent rise in unit labor cost may have been a drag on Latvia’s ability to supply domestic value added on world markets; preventing long-term misalignment between wage growth and productivity would help preserve competitiveness. - Trade-tension-induced tariff hikes are likely to have moderate costs for Latvia in terms of value added produced in the country. - Policy measures to mitigate exposure and improve competitiveness in the context of GVCs: - Enhance product sophistication or quality and pursue export market diversification to mitigate exposure to trade shocks in GVCs. - Backward GVC participation: increase use of imported intermediate goods with high-technological content to improve product sophistication; Latvia currently incorporates significant foreign value added but less than peer countries. - Forward GVC participation: promote upstream activities (exporting intermediate goods, re-exports, non-transport services) to generate productivity and employment gains; improve allocation and incentives for innovation through better access to credit and skilled labor with knowledge of foreign markets. - Attract FDI inflows to spur GVC participation through intra-firm trade channels; FDI can increase trade in intermediate and final goods and is an important determinant of backward participation. FDI targeting export processing facilities could improve product diversity and sophistication. *Source: IMF Staff estimates.* ### Introduction ### Introduction ### Overview and motivation - Productivity growth in Latvia slowed after the 2008–09 global financial crisis (GFC) from its pre-crisis high levels, a pattern consistent with a global post-GFC slowdown. - Latvia’s post-GFC productivity growth remains better than the euro area average, but boosting productivity growth is identified as the ultimate way to address long-term demographic challenges and achieve faster real convergence with the average income in western European countries. - Understanding factors affecting productivity growth is necessary to design policies to raise it. ### Literature and prior findings - Export entry increases productivity and employment for Latvian and Estonian firms, particularly for firms upstream in global value chains (Benkovskis et al., 2017). - Participation in EU co-funded projects raises firms' productivity with a lag (Benkovskis et al., 2018). - During the GFC, Latvian firms achieved higher productivity by using labor more efficiently and maintaining output with a smaller workforce; post-crisis productivity gains were driven by “catch-up” of relatively less productive firms (IMF, 2016a). - Productivity gaps between Central, Eastern, and Southeastern Europe and advanced Europe are largely due to structural and institutional obstacles and allocative inefficiencies (IMF, 2016b). - The prevalence of financially distressed (“zombie”) firms has risen since the GFC and is associated with lower investment and employment growth of non-zombie firms and reduced productivity-enhancing capital reallocation (Adalet McGowan et al., 2017a; IMF, 2018b). - Reforms lowering barriers to corporate restructuring are associated with higher productivity growth of laggard firms and aggregate productivity gains (Adalet McGowan et al., 2017b). ### Research question and main findings (overview) - The paper investigates firm-level factors associated with corporate productivity growth. - Main empirical findings: - Firms with a higher share of innovative (intangible) assets in total assets and better access to finance have higher productivity growth. - Smaller and younger firms generally exhibit faster productivity growth. - The presence of zombie firms is associated with lower productivity growth of other firms in the same sector. ### Organization of the paper - Section B describes empirical framework and data. - Section C presents analysis. - Section D concludes. ### Data and empirical framework - Data: annual firm-level Orbis database for non-financial corporates (NFCs) in Latvia, Estonia, Slovak Republic, and Slovenia, covering the 2010–15 period. - Latvian firms constitute about 20 percent of the total number of firms in the sample. - Micro firms (fewer than 10 employees) account for the vast majority of firms in all four countries. - Latvia has a very large share of micro and small enterprises: 96.2 percent of total (noted as slightly smaller compared to the other countries in the sample and the average EU share). - Due to data availability, pre-2010 data are not used because of a structural break. - Sectoral patterns: - Wholesale and retail trade have the highest share in all countries; in Latvia wholesale and retail trade show relatively high TFP growth. - Information and communication services is small but rapidly growing. - Agriculture has negative productivity growth in Latvia and other countries. - Manufacturing generally has higher-than-average productivity growth across countries. - TFP estimation: - Cobb-Douglas production function: Ln Yit = α + β LnKit + (1- β) LnLit + uit; the residual uit is log TFP. - Estimation by 2-digit NACE industry following De Loecker and Warzynski (2012) and Ackerberg, Caves, and Frazer (2015). - For Latvia, due to data availability, TFP is estimated using turnover revenue on the left-hand side and the cost of goods sold on the right-hand side as in De Loecker and Eeckhout (2017). - Firm-level data confirm a post-GFC slowdown in productivity growth. ### Regression specification for TFP growth - Baseline specification: - ∆TFPit = α + βSit-1 + δNon-zombieit-1 + θNon-zombieit-1*Zombie shareit-1 + γRecoveryit-1 + ui + vit - ∆TFPit measures TFP growth for firm i from year t-1 to year t. - Firm fixed effects and year dummies included. - Key explanatory variables (lagged to help control for endogeneity): - Debt-to-asset ratio (debt = long-term + short-term financial debt). Average debt-to-asset ratio is 17 percent. A quadratic term of debt is included. - Ratio of intangible to total fixed assets. Eighty two percent of firms in the dataset have zero intangible assets; the average ratio is 1 percent. Latvia’s average intangible assets are lower than peer countries; Latvian R&D expenditure is 0.5 percent of GDP. - Productivity gap to industry leaders (measured at NACE2 two-digit levels). The average TFP gap to industry leaders is about 100 percent (90 percent for Latvian firms). - Firm size: log of number of employees and its quadratic form. About one quarter of firms report only one employee; average number of employees is 17. - Firm age: firms less than 5 years are considered young. - Definition of zombie firms (Adalet McGowan et al., 2017a): - A firm is a zombie if aged 10 years or older AND has an interest coverage ratio (ICR = EBIT/interest paid) of less than one for three consecutive years. - All firms younger than 10 years are treated as non-zombies. - Latvia has a higher share of zombie firms than other countries in the sample; the share is particularly high in construction. - Sector-level variables: - Zombie shareit-1: share of zombie firms in the sector (2-digit NACE). - Non-zombieit-1 * Zombie shareit-1: interaction capturing impact of zombie-firm congestion on other firms. - Insolvency recovery rate (Recoveryit-1): cents on the dollar from World Bank Doing Business. - Recovery rate stands at 41 percent for Latvia in 2018. - EU average recovery rate reported as 63.4 percent. - The World Bank’s Doing Business Report shows recovery rate at 48.2 percent in 2016; IMF (2019) finds recovery rate was 23 percent in 2016. ### Key empirical results - Access to finance and debt: - Higher debt-to-asset ratio is positively associated with TFP growth and statistically significant at the 1 percent level. - An increase in debt financing of firm assets by 10 percentage points is associated with higher productivity growth by about 0.5 percentage points (at the average debt-to-asset ratio of 17 percent with respect to the quadratic term). - The quadratic term on debt is negative (only statistically significant in Table 4, Column (1)) suggesting excessively high debt is associated with lower productivity growth, though the implied threshold is very large. - Intangible assets and R&D: - The ratio of intangible fixed assets in total fixed assets is positive and statistically significant. - Raising Latvia’s intangible assets from an average of 0.4 percent to Slovenia’s 1.4 percent of total assets would increase productivity growth by 0.1 percentage point. - Latvia’s R&D expenditure at 0.5 percent of GDP is only a quarter of the EU average, indicating scope to increase intangible assets. - Policy note: Specific tax incentives to replace the R&D tax allowance that expired in 2018 could encourage R&D investment, but incentives need careful design (e.g., appropriate system for validating R&D tax credit). - Convergence to frontier: - The productivity gap to industry leaders is statistically significant at the 1 percent level. - The TFP gap would close by 18 percent per year (Table 4, Column (2)). - Example: for a firm with TFP equal to 52.6 percent of the industrial leader (TFP gap of 90 percent), its TFP growth would be 0.16 percentage points higher than the leading firms. - Firm size and age: - For micro firms, larger number of employees is associated with higher productivity growth. - Once firm size goes beyond micro firms, larger firms generally have lower productivity growth (after controlling for other factors). - Average number of employees is 16 for Latvian firms; 75 percent of Latvian firms have 8 or fewer employees. - Evidence suggests younger firms have higher productivity growth (Table 4, Column (3)). - Zombie firms and sectoral effects: - The non-zombie dummy alone is statistically insignificant. - The interaction term Non-zombieit-1 * Zombie shareit-1 is negative and statistically significant at the 1 percent level, indicating prevalence of zombie firms lowers productivity growth of other (non-zombie) firms in the sector. - Possible channels: zombie firms occupy productive resources and congest the sector, distorting prices and profits, discouraging new firm entry and investment, and raising sectoral risk premia. - Access to finance is harder for non-zombie firms in sectors with higher zombie shares: - Probit regression indicates that when explanatory variables are at average values, a 1 percentage point increase in the share of zombie firms in the sector would lower the probability of new borrowing for non-zombie firms by 1.04 percentage points (Table 5, Column (2)). ### Policy implications and considerations highlighted in the analysis - Facilitating access to finance can raise productivity growth, but excessive leverage may be harmful. - Policies that increase intangible assets (supporting R&D and innovation) could help raise productivity growth. - Consider carefully designed R&D incentives to ensure benefits to genuine R&D activities (e.g., an appropriate system for validating R&D tax credit). - Evidence suggests R&D tax subsidies are most effective when policy settings also encourage exit or downsizing of weak firms to free up resources. - Addressing the prevalence of zombie firms could raise productivity growth by: - Improving insolvency and corporate restructuring frameworks to lower barriers to restructuring and encourage efficient reallocation of resources. - Enhancing the insolvency recovery process (Latvia’s recovery rate: 41 percent in 2018) relative to EU averages. - Tackling zombie-firm congestion may also improve non-zombie firms’ access to financing and investment incentives. ### Conclusion (summary points) - Firm-level determinants of productivity growth in Latvia include access to finance, intangible assets, firm size and age, and sectoral prevalence of zombie firms. - Policies that expand access to appropriate finance, foster R&D and intangible investment, and improve insolvency and corporate restructuring regimes are indicated as potential avenues to boost aggregate productivity growth and support convergence. *Prepared by Yi Wu.* ### 17. Furthermore, productivity growth is higher in country/year with higher insolvency ### 17. Furthermore, productivity growth is higher in country/year with higher insolvency recovery rate ### Key empirical finding - An increase in the recovery rate by one standard deviation, or 9 percentage points, is associated with an increase of TFP growth by 0.4 percentage points. - The correlation between TFP growth and its lag is -0.16. ### Robustness checks and estimation approach - Wooldridge test for autocorrelation (Wooldridge 2002) does not reject the null hypothesis of no first-order autocorrelation for TFP growth, suggesting no need to include lagged TFP growth in the right-hand side of the regression (dynamic panel). - Fixed effects (equation 2) is considered a better specification than dynamic panel. - Dynamic panel estimated using the Blundell and Bond (1998) two-step system GMM yields results broadly similar to the baseline results. - Under some dynamic panel specifications, the intangible asset ratio and firm age are statistically insignificant, and the insolvency recovery rate becomes negative. - Note: if there is endogeneity caused by omitted variables or reverse causality, the GMM will not correct it (as lagged variables in level and differences are used as instruments). - Additional robustness checks (results available upon request) include: - using debt-to-equity ratio instead of debt-to-asset ratio; - using (log) assets as the measure of firm size; - dropping Slovenia from the sample (whose insolvency recovery rate increased significantly in 2015). - Various additional robustness checks yield broadly similar results. ### Concluding findings on firm characteristics and productivity - Firms with better access to finance and higher share of innovative assets had higher productivity growth. - Younger firms generally had faster productivity growth. - Productivity growth seems to slow as firms become larger (beyond micro firm). - Evidence points to productivity catch-up for firms that are far from the technological frontier. - Zombie firms negatively affect the productivity growth of other firms in the same sector, possibly by: - holding up productive resources; - reducing other firms’ ability to borrow; - preventing the entry of new firms. - Lower insolvency recovery rate is associated with lower productivity growth. ### Policy implications and recommendations - Policies that support access to financing, R&D and innovation, and improve the insolvency regime could potentially help raise productivity growth. - R&D policy recommendations for Latvia: - Latvia’s R&D spending lags behind other EU countries and there is ample room for improvement. - Better tax incentives that benefit genuine R&D activities could help support R&D spending. - The design of R&D incentives needs to guard against the risk of foregoing tax revenue without a commensurate rise in innovation. - Important aspects include the scope for eligible R&D, the eligibility requirements, and carry-back and carry-forward provisions. - The bulk of R&D spending usually occurs in large corporations, as young firms often do not generate enough profit to benefit from non-refundable tax incentives. - Cash refunds, carry forwards, or payroll withholding tax credits for R&D related wages could be more helpful to young firms. - Better access to seed financing and well-designed direct support, such as grants and contracts, may also stimulate R&D for such firms. - Well-functioning product, labor, and risk capital markets are essential to support R&D and other knowledge-based capital (OECD, 2013). - Insolvency regime recommendations for Latvia: - Improve insolvency regime to lower barriers to restructuring so zombie firms can be liquidated more quickly, allowing better allocation of resources and easier entry of new firms. - Faster restructuring or liquidation of financially distressed firms and better recovery rates for creditors would help reduce borrowing risk premiums and improve firms’ incentives to invest. - Bankruptcy laws that do not overly penalize business failure could encourage greater investment in knowledge-based assets. - Prevention tools (early warning mechanisms and pre-insolvency regimes) are also desirable. ### Selected summary statistics (from Tables 3a and 3b) - TFP growth: mean 0.004; Std. Dev. 0.17; Min -7.50; Max 6.38; measurement units ratio. - Debt-to-asset ratio: mean 0.17; Std. Dev. 0.23; Min 0; Max 1.0; measurement units ratio. - Intangible assets/total assets: mean 0.01; Std. Dev. 0.05; Min 0; Max 1.0; measurement units ratio. - TFP gap to frontier firms: mean 1.02; Std. Dev. 0.44; Min -6.09; Max 6.05; measurement units ratio. - Number of employees: mean 17.24; Std. Dev. 114.40; Min 1; Max 15000; measurement units unit. - Firm age: mean 12.37; Std. Dev. 6.99; Min 1; Max 118; measurement units year. - Insolvency recovery rate: mean 0.51; Std. Dev. 0.09; Min 0.32; Max 0.88; measurement units ratio. - Share of zombie firms in the sector: mean 0.01; Std. Dev. 0.01; Min 0.00; Max 0.30; measurement units ratio. - Latvia-specific summary statistics (Table 3b): - TFP growth: mean 0.01; Std. Dev. 0.18; Min -7.50; Max 6.38. - Debt-to-asset ratio: mean 0.18; Std. Dev. 0.24; Min 0.00; Max 1.00. - Intangible assets/total assets: mean 0.00; Std. Dev. 0.03; Min 0.00; Max 0.99. - TFP gap to frontier firms: mean 0.90; Std. Dev. 0.49; Min -6.09; Max 5.62. - Number of employees: mean 16.09; Std. Dev. 104.40; Min 1; Max 8537. - Firm age: mean 11.56; Std. Dev. 6.74; Min 2; Max 25. - Insolvency recovery rate: mean 0.47; Std. Dev. 0.02; Min 0.32; Max 0.48. - Share of zombie firms in the sector: mean 0.02; Std. Dev. 0.02; Min 0.00; Max 0.24. ### Selected regression results (Table 4: Latvia: TFP Growth, 2010–15) - Debt-to-asset ratio coefficients: - Column (1): 0.057*** (0.007) - Column (2): 0.042*** (0.007) - Column (3): 0.040*** (0.008) - Debt-to-asset ratio^2 coefficients: - Column (1): -0.023** (0.010) - Column (2): -0.006 (0.010) - Column (3): -0.010 (0.012) - Intangible assets/total assets coefficients: - Column (1): 0.052*** (0.018) - Column (2): 0.046*** (0.018) - Column (3): 0.053*** (0.020) - TFP gap to frontier firms: - Column (2): 0.183*** (0.001) - Column (3): 0.196*** (0.001) - Log number of employees: - Column (2): 0.026*** (0.002) - Column (3): 0.024*** (0.002) - Log number of employees squared: - Column (2): -0.006*** (0.000) - Column (3): -0.006*** (0.001) - Firm age: - Column (2): 0.001 (0.001) - Column (3): -0.002** (0.001) - Insolvency recovery rate: - Column (3): 0.045*** (0.012) - Non-zombie dummy (Column 3): -0.004 (0.004) - Non-zombie dummy * share of zombies in the sector (Column 3): -0.388*** (0.047) - Firm fixed effects: Yes (all columns). - Year dummies: Yes (all columns). - No. of firms: - Column (1): 125287 - Column (2): 119745 - Column (3): 102663 - No. of obs.: - Column (1): 387030 - Column (2): 368070 - Column (3): 294972 - R-squared: - Column (1): 0.01 - Column (2): 0.09 - Column (3): 0.10 - Note: All explanatory variables in lag. ***, **, * indicate significance at 1, 5, and 10% level. ### Selected regression results (Table 5: Latvia: Probability of New Borrowing for Non-Zombie Firms, 2010–15) - Share of zombie in the sector (in lag): - Column (1): -5.386*** (0.426) - Column (2): -4.776*** (0.495) - TFP gap to frontier firms (Column 2): -0.123*** (0.010) - Log number of employees (Column 2): 0.032*** (0.004) - Firm age (Column 2): -0.014*** (0.001) - Sector dummies (Nace2 two digits): Yes (Columns 1 and 2). - Year dummies: Yes (Columns 1 and 2). - No. of obs.: - Column (1): 170,862 - Column (2): 143,947 - Note: All explanatory variables in lag. ***, **, * indicate significance at 1, 5, and 10% level. *Source: Orbis and IMF staff calculations, as presented in the chapter.* ### References ### References ### Selected citations - Acemoglu, D., U. Akcigit, N. Bloom, and W. Kerr, 2013, “Innovation, Reallocation and Growth,” NBER Working Papers, No. 18993. - Ackerberg, D. A., Caves, K., and G. Frazer, 2015, “Identification Properties of Recent Production Function Estimators.” Econometrica 83(6): 2411–2451. - Adalet McGowan, Muge, Dan Andrews, and Valentine Millot, 2017a, “The Walking Dead? Zombie Firms and Productivity Performance in OECD Countries,” OECD Economics Department Working Papers No. 1372. - Adalet McGowan, Muge, Dan Andrews, and Valentine Millot, 2017b, “Insolvency Regimes, Technology Diffusion and Productivity Growth: Evidence from Firms in OECD Countries,” OECD Economics Department Working Papers No. 1425. - Adalet McGowan, Muge, Dan Andrews, and Valentine Millot, 2018, “Insolvency Regimes, Zombie Firms and Capital Reallocation,” OECD Economics Department Working Papers No. 1399. - Adler, Gustavo, Romain Duval, Davide Furceri, Sinem Kiliç Çelik, Ksenia Koloskova, and Marcos Poplawski-Ribeiro, 2017, “Gone with the Headwinds: Global Productivity,” IMF Staff Discussion Note 17/04. - Anderson, Gareth and Mehdi Raissi, 2018, “Corporate Indebtedness and Low Productivity Growth of Italian Firms,” IMF Working Paper 18/33. - Benkovskis, Konstantins, Jaan Masso, Olegs Tkacevs, Priit Vahter, and Naomitsu Yashiro, 2017, “Export and Productivity in Global Value Chains: Comparative Evidence from Latvia and Estonia,” OECD Economics Department Working Papers No. 1448. - Benkovskis, Konstantins, Olegs Tkacevs, and Naomitsu Yashiro, 2018, “Importance of EU Regional Support Programmes for Firm Performance,” Bank of Latvia Working Paper 2018/01. - Blundell, R. and S. Bond, 1998, “Initial Conditions and Moment Restrictions in Dynamic Panel Data models," Journal of Econometrics, vol. 87(1), pages 115–143, August. - Budina, N., P. Deb, F. Diez, J. Fan, J. Luzi, F. Misch, and A. Shabunina, 2018, “Drivers of Firm Productivity Growth and Misallocation,” mimeo. - Caballero, Ricardo J., Takeo Hoshi, and Anil K. Kashyap, 2008, “Zombie Lending and Depressed Restructuring in Japan,” American Economic Review 98(5), 1943–77. - De Loecker, J. and Warzynski, F., 2012. “Markups and Firm-Level Export Status,” American Economic Review 102(6), 2437-71. - De Loecker, J. and Eeckhout, J., 2017, “The Rise of Market Power and the Macroeconomic Implications,” NBER Working Paper No. 23687. - Díez, Federico J., Jiayue Fan, and Carolina Sanchez-Villegas, 2018, “Orbis Dataset: Cleaning Procedure and TFP Estimation,” mimeo. - Duval, Romain, Gee Hee Hong, and Yannick Timmer, 2017, “Financial Frictions and the Great Productivity Slowdown,” IMF Working Paper No. 17/129. - Hausmann, Ricardo, 2017, “Here’s How We Can Boost Research and Development,” World Economic Forum. - Huergo, Elena and Jordi Jaumandreu, 2004, “Firms’ Age, Process Innovation and Productivity Growth,” International Journal of Industrial Organization 22(4), 541-59. - International Monetary Fund (IMF), 2016a, “Post-Crisis Adjustment in Latvia: Evidence from Firm Level Data,” IMF Country Report No. 16/171. - International Monetary Fund (IMF), 2016b, “Central, Eastern, and Southeastern Europe: How to Get Back on the Fast Track,” Regional Economic Issues. - International Monetary Fund (IMF), 2016c, “Firm-Level Productivity and Its Determinants: The Irish Case,” IMF Country Report No 16/257. - International Monetary Fund (IMF), 2018a, “Labor Productivity in the Czech Republic: Stocking and Policies,” IMF Country Report No. 18/188. - International Monetary Fund (IMF), 2018b, “Productivity Growth in Asia: Boosting Firm Dynamism and Weeding out the Zombies,” Regional Economic Outlook: Asia and Pacific, Background Paper No. 3. - International Monetary Fund (IMF), 2019, “Latvia—Evaluation of the Insolvency Framework”, Technical Assistance Report. - OECD, 2013, “Supporting Investment in Knowledge Capital, Growth and Innovation.” - OECD, 2018, “OECD Review of National R&D Tax Incentives and Estimates of R&D Tax Subsidy Rates, 2017.” - European Union, 2018, Annual Report on European SMEs 2017/2018. - The World Bank, 2017, Latvia Tax Review. - Wooldridge, J. M., 2002, Econometric Analysis of Cross Section and Panel Data. Cambridge, MA: MIT Press. ### Financial integrity and stability: key findings and statistics (Latvia) - Monitoring the de-risking process requires effective prudential and conduct regulation to support the re-orientation of banks servicing foreign clients (BSFCs) and mitigate remaining and emerging risks associated with new business models and changing liquidity risk profiles. - Main qualitative findings: - Need for a deeper understanding of ML/TF risks, more effective risk-based supervision accompanied by proportionate and dissuasive sanctions, and greater use of financial intelligence in ML/TF investigations. - Liquidity stress test performed independently of the forthcoming FSAP confirms a sufficient degree of resilience of the sector, as banks enter the exercise from a position of relative strength, reducing fundamental vulnerability to funding shocks if ML/TF concerns were to flare up again. - Supervisory assessment of liquidity risk should expand the current indicator-based framework to integrate ML/TF risks and their impact on banks’ viability in the development of stress test scenarios. - Sector structure and scale: - Total sector assets: €21.0 billion at end-2018. - BSDCs (Swedbank, Luminor Bank, SEB banka, and Citadele banka) account for more than three-quarters of total sector assets. - BSFCs historically relied on non-resident deposits; at their peak in early 2016, about 40 percent of transactional flows of foreign clients originated from former or current CIS member countries and OFCs. - Correspondent banking relationships (CBRs) and international context: - Latvia’s cumulative change in CBRs: -13.7 percent (after adjusting for shrinking banking sector). - Comparators: Baltics average -19.2 percent; EU average -14.4 percent. - De-risking outcomes and transaction volumes: - Since end-2017, non-resident deposits declined by more than 60 percent (€4.8 billion); their share decreased to about 20 percent of total deposits. - Three-quarters of the decline in non-resident deposits were mostly USD from offshore jurisdictions that seem to have left the banking sector altogether. - Incoming and outgoing interbank payments of foreign clients declined by a factor of four across all banks due to the steep reduction in USD transactions. - Since 2016, the total volume of incoming and outgoing payments (gross values) transacted by BSFCs declined from 49.7 to 10.9 percent of GDP by end-March 2019; BSDCs transactional business stood at 7.5 percent of GDP. - Shell entity customers across the banking sector decreased from 19,590 at end-2017 to 1,971 in May 2018. - Deposit volume of shell entities declined from €2.7 billion to €379 million over the same period. - Triggering events and supervisory response: - ABLV Bank’s closure in early 2018 following proposals by FinCEN prompted intensified reform and re-orientation of BSFCs away from transaction-based business toward services supporting the real economy (specialized lending, asset/wealth management, e-commerce). - As part of SREP, all BSFCs submitted and obtained approval for new business plans and are implementing them; ML/TF risk assessment is a component of SREP (quantitative since 2009; qualitative component added in 2018). - Legislative and institutional reforms (highlights): - Amendments to the Law on the Prevention of Money Laundering and Terrorism and Proliferation Financing (AML/CFT Law) limiting engagement with certain types of shell entities and prohibiting engagement with shells meeting defined characteristics. - Amendments to extend application to proliferation financing and introduce the offence of proliferation financing (PF). - Replacement of unusual transaction reports (UTRs) with threshold reporting requirements for transactions conducted through financial institutions. - Amendments to the Law on Sanctions providing a mechanism to implement UN designations without waiting for an EU regulation; entered into force on July 4, 2019. - Full transposition of the Fifth EU Anti-Money Laundering Directive, opting for public disclosure of beneficial ownership information as of 1 March, 2018, and amendments providing for exclusion from the Enterprise Registry for failure to submit beneficial ownership information. - Amendments of the FCMC Act and the Credit Institutions Law broadening the FCMC’s remit to include prevention of money laundering and combatting terrorism and proliferation financing in supervision and liquidation of banks, revision of Board appointment and removal procedures, and open recruitment/Parliament approval of the FCMC Board. - As of July 4, 2019, the current Chairman and his Deputy tendered resignations, effective as of July 15, 2019. - Institutional capacity and supervisory activity: - The Financial Intelligence Unit (FIU) significantly expanded operational resources and recruited staff to fill strategic analyst positions that were vacant at the time of the MER; the FIU developed new typologies (e.g., virtual currencies in 2018) and updated the NRA. - The FCMC increased on-site inspections: 14 in 2018 and 9 in the first half of 2019 (compared to 6 in 2017). *Source: 1lvaea2019002 - References* ### Box 1. MONEYVAL’s Assessment of Latvia’s AML/CFT Regime ### Box 1. MONEYVAL’s Assessment of Latvia’s AML/CFT Regime ### Overall assessment - MONEYVAL (Council of Europe, 2018) raised concerns about Latvia’s anti-money laundering and combating the financing of terrorism (AML/CFT) regime. - The regime was deemed effective in some areas, such as cooperating with foreign counterparts, but significant deficiencies were identified, putting the country at risk of being designated (grey-listed) as a jurisdiction with strategic AML/CFT deficiencies. - MONEYVAL’s Mutual Evaluation Report (MER) acknowledged a general understanding of ML/TF risks but found an uneven appreciation of the specific ML/TF threats posed by large cross-border banking flows. - Overall, Latvia was found to lack targeted measures to address ML/TF threats arising from the high concentration of non-resident deposits and transactional flows in BSFCs. ### Identified weaknesses - Weaknesses identified in the MER included: - (i) AML/CFT supervision and the analytical abilities of the Financial Intelligence Unit (FIU), - (ii) the application of internal controls by financial institutions, - (iii) the prioritization of ML/TF offences by law enforcement and prosecution authorities. - The MER also expressed doubts as to the adequacy of customer due diligence being carried out by BSFC banks, citing poor quality of beneficial ownership information (BOI) (often taken at face value by banks when on-boarding new customers) as a significant vulnerability for ML/TF as well as proliferation financing (PF). ### Banking sector implementation and reporting - The insufficient implementation of AML/CFT controls in the banking sector amplified concerns about the supervisory regime. - At the time of the MER: - most banks had not incorporated ML/TF risks into their enterprise risk management systems (i.e., banks could assess the risk of their customers, but could not speak to their own risk exposure), - no banks characterized their ML/TF risks exposure as being high. - This incongruous risk assessment was reflected in the level and quality of suspicious transaction reporting, which demonstrated a poor risk understanding or weak implementation of AML/CFT control on the part of BSFCs. - Reported figures from past years suggest defensive reporting by Latvian banks (given the high number of reports received relative to the number of disseminations by the FIU to law enforcement), although limitations in the FIU’s operational analysis may have also played a contributing factor. ### FIU capacity and operational analysis - At present, the FIU has 8 strategic analysts and 16 operational analysts. ### Assessment ratings context - Assessment Ratings of Effectiveness (note in source): - Includes only European countries for which assessments under the 2013 FATF Methodology are available. - Based on MONEYVAL assessments for ALB, CZE, HUN, LVA, LTU, SRB, SVN, and UKR; FATF assessments for AUT, BEL, CHE, DNK, ESP, GBR, IRE, ISL, NOR, PRT, and SWE; and IMF/FATF assessment for ITA. - Ratings reflect the extent to which a country's measures are effective. The assessments are conducted on the basis of 11 immediate outcomes, which represent key goals that an effective AML/CFT system should achieve. - Country ordering shown in source (verbatim): SRB ISL LVA LTU SVN HUN NOR ALB UKR AUT CZE DNK BEL IRE SWE PRT CHE ITA ESP ISR GBR - Effectiveness legend shown in source (verbatim): HighSubstantialModerateLow *Source: MONEYVAL (Council of Europe, 2018); Financial Action Task Force (FATF), Consolidated Assessment Ratings, April 17, 2019.* ### Box 1. MONEYVAL’s Assessment of Latvia’s AML/CFT Regime (concluded) ### Box 1. MONEYVAL’s Assessment of Latvia’s AML/CFT Regime (concluded) ### FIU reporting and case initiation - STRs received, Transactions involved, Disseminations, STRs included in dissemination: - 2015: 7,267; 17,525; 340; 584 - 2016: 5,008; 18,712; 231; 439 - 2017: 7,722; 25,517; 225; 393 - Source: Latvia Financial Intelligence Unit. Note: STR=Suspicious Transaction Report. - FIU disseminations to law enforcement and criminal proceedings started: - 2017: FIU disseminated 225 reports; Criminal proceedings started 83; Ratio 36.9% - 2018: FIU disseminated 171 reports; Criminal proceedings started 59; Ratio 34.5% - 2019*: FIU disseminated 99 reports; Criminal proceedings started 36; Ratio 36.4% - Source: Latvia Financial Intelligence Unit. Note: */ until July 1, 2019. ### MONEYVAL / MER findings on legal framework and enforcement - Legal and sanctions framework: - Latvia’s legal framework was deemed not sufficiently precise to ensure implementation of targeted financial sanctions (TFS). - Risk ratings: Proliferation Financing (PF) = “medium high”; Terrorist Financing (TF) = “low”. - At MER time, inconsistencies existed among EU Regulations, the AML/CFT Law and the Law on Sanctions (scope of freezing obligation, persons obligated, FIU’s indefinite freezing without court order). - Instances noted of banks repeating breaches under administrative agreements. - Penalties after MER: In 2017, the FCMC fined three banks a total of €641,000 for systematic failures to detect payments to the DPRK (first penalties of this kind); penalties for contravention of DPRK sanctions were significantly lower than for other AML/CFT compliance breaches. - Quality of financial intelligence and prosecutorial outcomes: - FIU disseminations were of insufficient quality at MER time: incomplete reports and inaccurate beneficial owner identification. - FIU over-relied on BOI submitted by reporting entities and failed to classify predicate offences based on its own analysis. - Law enforcement preferred to collect financial intelligence independently. - In 2016: three times as many prosecutions were initiated based on FIU disseminations than other sources, yet convictions were the same number across categories—indicating a lower conviction rate from FIU-initiated proceedings. - Lack of training among police creates evidentiary challenges for money laundering enforcement. ### Findings: awareness, banking sector response, and residual risks - Overall ML/TF awareness: - Overall ML/TF awareness has been improving; authorities and industry broadly aware of main risks faced by Latvian banks. - Agencies vary in degree and granularity of risk analysis; not all stakeholders cognizant of specific modalities for illicit fund movement. - Banks’ compliance and business-model adjustments: - Banks moved to fully automated systems for customer due diligence (CDD) and risk assessment, including TF and PF. - Banks began incorporating ML/TF risks at strategic risk-management level. - De-risking metrics coarse (e.g., resident vs. non-resident) leading to account closures of companies with genuine economic activity. - Some banks pursued market niches (capital markets, trade financing for SMEs); others scaled back high-risk customer relationships. - TF/PF controls: - Controls for TF/PF rose in priority; since 2018 banks obliged to have sanctions policies and a dedicated sanctions compliance officer. - FIU received 142 STRs related to TF in 2018 (compared with 31 in 2017). - Non-resident accounts and concentration of risk: - Decline in non-resident deposits may have reduced ML/TF risk but remaining BSFCs still hold a considerable share of high-risk clients in remaining gross payment flows. - FCMC classified all BSFCs as high-risk banks as of end-March 2019. - Payment patterns to/from foreign customers remained similar at lower volume; potential roundtripping using Latvia as a transit country may continue. - Significant differences in reported trade flows between Latvia and CIS countries (overreported exports and underreported imports) provide circumstantial evidence of potential mis-pricing due to illicit transactional flows. - Shell companies and legal carve-outs: - 2018 AML/CFT law amendments carve out a category of shell entities banned from business with financial institutions. - Entities covered by prohibition represent a very small share of entities that fall within shell company definition; definition subject to interpretation. - Many shell companies whose accounts were closed do not fall under the prohibition; law has strong “signaling effect” but long-term effectiveness questionable. - Residency classification for corporate accounts is based on legal ownership; decline in non-resident deposits is reduction in accounts held by companies established abroad, not necessarily accounts of Latvian companies with foreign owners. - Supervisory approach and capacity: - Uncertainty about implementation of supervisory reforms led to excessive risk aversion and account closures of non-resident clients. - FCMC’s risk-based supervision needs improvement: risk categorization does not deterministically drive inspection plans; critical risk categorization lacks binding supervisory actions. - Due to resource constraints, supervision focused on targeted/thematic inspections; FCMC had not conducted full-scope inspection of all high-risk banks in last three years (two banks not subject to full-scope inspection during 2016–19). - Of 9 on-site inspections in first half of 2019, only one was a full-scope inspection. - None of the high-risk banks had more than one full-scope inspection in last three years despite policy of full-scope exams every 18 months. - Sanctions regime: - Number of sanctions increased markedly in 2018 but effectiveness, dissuasiveness and proportionality remain in question. - Regulator can enter administrative agreements to reduce pecuniary penalties; penalties have varied widely. - Examples: 2017 fines for CDD weaknesses leading to circumvention of sanctions ranged between €35,000 and €1.3 million; 2018 fines for CDD deficiencies ranged from €9,825 to €2.2 million. - Administrative sanction procedures can take up to a year to complete due to delays and limited resources. - Risks outside banking sector: - ML/TF risks outside banking sector poorly monitored: weak or non-existent supervision of cash collection service providers, financial leasing service providers, non-bank lenders, other payment service providers not supervised by FCMC. - Tax administration building capacity in real estate sector inspections; sanctions generally around €1,800 with maximum €10,000 and not effective, proportionate, or dissuasive. - FIU operational and strategic analysis: - FIU progressively strengthened systems to identify predicate offences and improve analytical ability, including external consultant support. - New IT systems expected to enhance operational capacity and strategic activities. - Percentage of FIU disseminations resulting in initiation of criminal proceedings has remained roughly constant over recent years (see dissemination table above). ### Policy implications and recommended actions - Overarching challenge: - Need to demonstrate effectiveness of the enhanced AML/CFT regime in the near term; consider absorptive capacity of regulatory authorities and private sector; sequence activities to ease uncertainty and allow training and meaningful assessment of effectiveness. - Priority actions (excerpted and preserved): - Understanding AML/CFT risks: - Continue improving understanding and analysis of ML/TF threats/vulnerabilities to ensure effective de-risking. - Conduct deeper analysis of available data (present and historical) to enhance assessment of evolving vulnerabilities and guide a forward-looking NRA and targeted risk mitigation strategy. - Increase FIU’s operational capabilities to strengthen analytical products and maintain detailed statistics. - Strengthening AML/CFT supervision: - Adopt a more refined risk-based model to guide FCMC supervisory activities, broaden risk factors to identify critical risks, link supervisory actions to risk categorization, and shift focus toward full-scope inspections on a risk-based approach. - Provide aid to bank and non-bank financial institutions in developing risk assessments, mitigation measures, and internal controls. - Strengthening the sanctions regime: - Evaluate use of administrative agreements for reduction of penalties; consider a more streamlined procedure for imposing administrative sanctions. - Ensuring accurate beneficial ownership information: - Focus AML/CFT supervision on effective implementation of updates to improve availability of companies’ BOI; monitor regulatory and institutional framework for BOI collection, registration, and verification under company law. - Improving preventive measures and reporting: - FIU and supervisory authorities should ensure appropriate quality and amount of STR reporting, with special focus on higher risk entities such as BSFCs and TFS implementation. - Encourage nuanced risk-based approach rather than outright bans on customer types; accompany with thorough understanding of risk factors, stronger internal controls, and proactive regulatory guidance. - Improving quality and use of financial intelligence: - FIU to continue to improve analytical capabilities, recruit and train staff, apply adequate IT solutions, and collaborate with law enforcement, prosecutors, and judges. - Enhancing ML/TF enforcement: - FIU should work with law enforcement and prosecutors to strengthen investigative and prosecutorial bodies, improve ability to lay charges and secure convictions, and develop sentencing guidelines to ensure criminal sanctions are effective, proportionate, and dissuasive. - Ensure independence of regulatory authorities: - Strengthen governance and accountability of FCMC with safeguards so removal procedures for Board members or Chairperson do not undermine regulator independence. - Encouraging greater strategic coordination and cooperation: - Regulatory authorities should work toward a common national vision; Financial Sector Development Board is the main forum for developing action plan to address MONEYVAL recommendations. - FCMC and FIU cooperation strengthened but should further share mutual accountability in setting strategic goals; finalize a Memorandum of Understanding initiated in June 2019. - FIU to be key participant in re-orientation of BSFCs given its role in strategic analysis. - Note: These areas should be harmonized with any action plan developed with the FATF or MONEYVAL. ### Liquidity Stress Test: purpose, design, and scenarios - Purpose: - Assess how de-risking and ML/TF vulnerabilities affect liquidity conditions under stress; examine system-wide resilience to sudden sizable withdrawals of funding; identify threats to financial stability potentially triggered by escalation of ML/TF risks. - Stress test covers 94 percent of total banking sector (including foreign subsidiaries/branches but excluding banks that exited Latvia by end-2018). - Methodology: - Top-down implied cash flow (ICF) tests using publicly available 2018 statutory financial filings. - Two tests: cumulative five-day ICF and non-cumulative 30-day ICF (methodology in Jobst, Ong, and Schmieder (2017 and forthcoming)). - Tests capture market liquidity risk (inflows from maturing/repo-able/saleable assets), funding liquidity risk (restricted access to funding markets), unscheduled cash flows from credit lines, and net derivatives flows. - Liquidity ratio = cash inflows (including asset sale proceeds, securities lending, repos) divided by cash outflows; ratio > 100 percent = liquidity surplus, ratio < 100 percent = deficient liquidity position. - Sample: tests applied to all 4 BSDCs and 7 (out of total 11) BSFCs on a solo basis, representing 95.0 percent of banking sector's total assets. - Scenarios: - Main Scenario (“historical scenario”): - Calibrated to (i) actual deposit run-off experienced by BSFCs, (ii) valuation change of Latvian government debt securities, and (iii) rise in external finance premium after ECB’s FOLF decision on ABLV Bank in February 2018. - Examples of calibrations: valuation haircut for government bonds and other public sector securities set at -5.2 percent; valuation haircut for other debt securities and marketable loans set at -12.9 percent; deposit run-off rate applied as daily run-off rate of 5.5 percent during four weeks after FOLF decision (consistent with 27.7 percent over 30 days). - Background Scenarios: - Regulatory Scenario: mapping LCR to ICF tests. - Quasi-ECB scenarios: mapped from ECB’s LiST (baseline, adverse, extreme). - IMF-Standard Scenario: conventional FSAP scenario used in recent FSAPs for European countries. - Assumptions cover valuation haircuts, amortization/renewal rates, drawdown rates on contingent claims, intragroup funding behavior, and treatment of derivatives. ### Liquidity Stress Test findings and vulnerabilities - Aggregate sector results: - Liquidity risk for most banks is low; banks exceed minimum statutory liquidity ratios and hold large stocks of liquid assets at low encumbrance levels. - Overall excess liquidity amounts to 20 percent of total assets (stated as “excess liquidity amounts to 20 percent of total assets, which is the highest level in any euro area member country”). - Sample LCRs and capital: - Capital adequacy (CET1): BSDCs 20.3 percent vs. BSFCs 20.4 percent (weighted-average CET1). - Weighted-average LCR at end-2018: BSFCs 375.0 percent vs. BSDCs 247.0 percent. - For stress testing sample partial coverage of BSFCs raises average sample LCR from 269.0 to 271.3 percent. - ICF test outcomes: - Five-day ICF test: - Despite contraction of unsecured funding by up to 21 percent, most banks can compensate for negative net cash outflows over 5 days; implied ICF liquidity ratio ≥ 100 percent for most banks. - Average net cash shortfall of banks with deficient liquidity position: maximum of 4.0 percent of liquid assets and 12.5 percent of total assets under the historical scenario (worst aggregate outcome). - 30-day ICF test: - Extending horizon to 30 days generates no overall liquidity shortage under all but most extreme scenarios (loss of unsecured funding up to 49 percent under adverse and extreme ECB scenarios). - Average net cash shortfall declines to 2.1 percent of liquid assets and 6.5 percent of total assets due to larger number of marginally failing banks. - No bank had a deficient liquidity position under the main historical scenario. - Residual vulnerabilities and structural concerns: - Banks vulnerable to large outflow shocks due to limited alternative term funding sources; dependence on retail deposits from other euro area countries and local corporate deposits which are market-sensitive. - Banks with large mortgage and consumer loan books and high maturity mismatches may find liquid buffers insufficient under sizable deposit declines. - Lack of diversity of funding with longer maturities and high carrying cost of large cash balances present medium-term challenges for BSFCs re-orienting to domestic lending. - BSDCs: retail-dominated deposit base but dependency on collateral value of less liquid assets and susceptibility to intragroup funding outflows and contingent claims. - BSFCs: main risk driver is shrinking deposit base; liquidity improves substantially if customer deposits remain stable. - Some banks hold significant intragroup funding obligations and contingent claims that could amplify liquidity-to-solvency feedback under stress. - Robustness and caveats: - Results robust across scenario variations and asset encumbrance assumptions; differences reflect haircuts and callback/drawdown rates. - Stress test is static, uses public data as of end-2018, excludes some recent developments, and assumes all banks face escalating liquidity risk simultaneously. - Limitations include coarse maturity buckets, exclusion of separate FX liquidity risk testing, and limited modeling of interconnectedness and solvency-liquidity feedback effects. - Mitigating factors not incorporated in full: central bank refinancing as lender of last resort, reallocation of deposits within banking sector, and possible non-simultaneous shocks across banks. ### Liquidity stress test policy implications and recommendations - Summary conclusion: - Stress test confirms sufficient degree of resilience of the sector: banks are profitable, hold high levels of capital and liquidity, and asset encumbrance is relatively low. - Only a very severe deterioration of market and funding liquidity risk would cause material aggregate net cash outflows over up to 30 days. - Some banks remain vulnerable to large outflow shocks, potentially triggered by renewed AML/CFT concerns. - Recommended supervisory and policy follow-ups: - Refine qualitative and quantitative treatment of liquidity risk in the SREP and integrate AML/CFT considerations into stress test scenario development. - Widen specification of associated liquidity risks beyond indicator-based approaches to a comprehensive system-wide framework to support effective risk-based supervision and macroprudential surveillance. - Incorporate nuance concerning: - (i) varying asset quality of counterbalancing capacity and eligibility of liquid assets as central bank collateral; - (ii) adverse deposit composition effects on behavioral cash outflows given de-risking; and - (iii) interlinkages between liquidity and solvency risks influenced by interconnectedness and network effects in the banking sector. - Consider incorporation of ML/TF risks within stress testing framework using granular prudential information (e.g., contingent claims and intragroup transactions under severe stress; concentration of high-risk clients in lending and payment services). - Supplementary longer-term funding sources recommended to reduce cash-flow mismatches for BSFCs as they re-orient towards domestic lending. *Source: IMF staff summary of Box 1. MONEYVAL’s Assessment of Latvia’s AML/CFT Regime (concluded).* ### Annex I. Table 1. Latvia: Detailed Assumptions for Implied Cash Flow Tests―Liquid Assets (Valuation Haircuts) ### Annex I. Table 1. Latvia: Detailed Assumptions for Implied Cash Flow Tests―Liquid Assets (Valuation Haircuts) ### Annex I. Detailed Scenario Assumptions - Sources: BCBS; ECB; Jobst and others (2017); and IMF staff calculations. - Note: LCR=liquidity coverage ratio. The following (additional) assumptions were made regarding the calibration of the parameters for the various scenarios ― for the LCR scenario: all non-public assets are rated AA or below (otherwise, haircut would be 15 percent only); for the ECB scenarios: - (i) all securities have a residual maturity of 5-10 years (for the 30-day test), - (ii) public sector assets with a credit RWA of 20 percent are assumed to be rated "BBB" (CQS=3), and - (iii) all equities qualify as high-quality liquid asset (HQLA). - */ Calibrated to historical experience during ABLV Bank episode (and applied to all banks); **/ Common calibration used in the liquidity stress test of the IMF Financial Sector Assessment Program (FSAP). - 1/ In excess of minimum reserve requirements. - 2/ E.g., repo/reverse repo. ### Liquid assets (valuation haircuts) — key calibration notes - All non-public assets: rated AA or below for LCR scenario (otherwise haircut would be 15 percent only). - ECB scenario adjustments: residual maturity 5-10 years for 30-day test; public sector assets with credit RWA of 20 percent → rated "BBB" (CQS=3); equities considered HQLA. - Calibration flags: */ and **/ retained as in source. --- ### Annex I. Table 2. Latvia: Detailed Assumptions for Implied Cash Flow Tests―Cash Inflows (Callback and Drawdown Rates) ### Sources and note - Sources: BCBS; ECB; Jobst and others (2017); and IMF staff calculations. - Note: LCR=liquidity coverage ratio. The following (additional) assumptions were made regarding the calibration of the parameters for the ECB scenario: - (i) unsecured lending to the public and non-financial private sector (households, SMEs, retail, and wholesale) received call-back rates according to the category “monies due not reported above resulting from sight and non-maturing loans and advances” while unsecured lending to credit institutions and other financial customers as treated as monies due not reported as “monies due from secured lending,” - (ii) intragroup funding was assigned the call-back rates for “other cash inflows,” and - (iii) all secured lending is classified as “monies due from secured lending and capital market transactions collateralized." - */ Calibrated to historical experience during ABLV Bank episode (and applied to all banks); **/ Common calibration used in the liquidity stress test of the IMF Financial Sector Assessment Program (FSAP). - 1/ For the 5-day test under the LCR scenario, the run-off and drawdown rates are scaled consistent with the ratio of the parameter values for the 5-day and 30-day tests under the standard IMF scenario. - 2/ Transactions with liquid securities and bank loans (e.g., reverse repo and securities borrowing transactions). ### Cash inflows — calibration highlights - ECB scenario mapping: - Public/non-financial unsecured lending → call-back rates per “monies due not reported above resulting from sight and non-maturing loans and advances.” - Unsecured lending to credit institutions/other financial customers → treated as “monies due from secured lending.” - Intragroup funding → call-back rates for “other cash inflows.” - Secured lending → “monies due from secured lending and capital market transactions collateralized.” - Scaling for 5-day LCR test: run-off and drawdown rates adjusted consistent with 5-day vs 30-day standard IMF scenario ratios. - Calibration markers: */ and **/ as in source. --- ### Annex I. Table 3. Latvia: Detailed Assumptions for Implied Cash Flow Tests―Cash Outflows (Runoff and Drawdown Rates) ### Sources and note - Sources: BCBS; ECB; Jobst and others (2017); and IMF staff calculations. - Note: LCR=liquidity coverage ratio. The following (additional) assumptions were made regarding the calibration of the parameters under various scenarios ─ for the LCR scenario: - (i) 50/50 split between stable and less stable retail deposits; - (ii) 50/50 split between covered/uncovered deposits, - (iii) no non-retail operational deposits (so only “non-operational” parameters applied to non-retail categories), - (iv) secured funding with securities and loans that can be mobilized for secured funding /marketable securities are subject to an average run-off rate of “Level 2A assets” and “non-Level 2A assets”/”RMBS eligible for Level 2B” and “Other Level 2B assets”/”Other funding transactions,” and - (v) secured funding with potentially re-usable securities received as collateral are considered “other funding transactions”; - for the ECB scenarios: - (i) operational deposits are defined as in the LCR scenario, - (ii) all retail (incl. SME) deposits are treated as operational sight deposits, - (iii) other short-term funding through maturing bonds (issued by the institution) and other sources is mapped to the run-off rate of “other transactions,” and - (iv) all secured funding is mapped to “liabilities from secured borrowing and capital markets driven transactions, collateralized.” - */ Calibrated to historical experience during ABLV Bank episode (and applied to all banks); **/ Common calibration used in the liquidity stress test of the IMF Financial Sector Assessment Program (FSAP). - 1/ For the 5-day test under the LCR scenario, the call-back and drawdown rates are scaled consistent with the ratio of the parameter values for the 5-day and 30-day tests under the standard IMF scenario. - 2/ Transactions with liquid securities and bank loans (e.g., reverse repo and securities borrowing transactions). ### Cash outflows — calibration highlights - LCR scenario specifics: - Retail deposits: 50/50 stable vs less stable. - Covered vs uncovered deposits: 50/50. - No non-retail operational deposits; non-retail categories use “non-operational” parameters. - Secured funding: mobilizable securities/loans mapped to Level 2A/Level 2B average run-off rates per description. - Re-usable securities received as collateral → “other funding transactions.” - ECB scenario mapping: - Retail (incl. SME) deposits → operational sight deposits. - Maturing bonds and other short-term funding → run-off rate of “other transactions.” - Secured funding → “liabilities from secured borrowing and capital markets driven transactions, collateralized.” - Scaling for 5-day LCR test: call-back and drawdown rates adjusted consistent with 5-day vs 30-day standard IMF scenario ratios. - Calibration markers: */ and **/ as in source. --- ### Annex I. Table 4. Latvia: Detailed Assumptions for Implied Cash Flow Tests―Net Derivative Flows (Discount Rates) ### Sources and note - Sources: BCBS; ECB; Jobst and others (2017); and IMF staff calculations. - Note: LCR=liquidity coverage ratio. The following (additional) assumptions were made for the ECB scenarios: - net contractual cash flows from derivatives are assessed at current market values of derivatives amount payables/receivables and mapped to “FX swaps maturing” (for currency derivatives) and “other transactions (for interest rate and other derivatives).” - */ Calibrated to historical experience during ABLV Bank episode (and applied to all banks); **/ Common calibration used in the liquidity stress test of the IMF Financial Sector Assessment Program (FSAP). - 1/ For the 5-day test under the LCR scenario, the run-off and drawdown rates are scaled consistent with the ratio of the parameter values for the 5-day and 30-day test under the standard IMF scenario. - 2/ Excluding credit derivatives. ### Net derivative flows — calibration highlights - ECB scenario mapping: - Derivative net contractual cash flows → assessed at current market values and mapped to: - “FX swaps maturing” for currency derivatives, - “other transactions” for interest rate and other derivatives. - Scaling for 5-day LCR test: run-off and drawdown rates adjusted consistent with 5-day vs 30-day standard IMF scenario ratios. - Credit derivatives: excluded (2/). - Calibration markers: */ and **/ as in source. --- ### Annex II. AML/CFT Considerations in FCMC’s Supervisory Review and Evaluation Process (SREP) ### Overview of SREP AML/CFT integration - Since 2012, the FCMC has set specific Pillar II capital requirements for banks servicing foreign clients (BSFCs) to ensure prudent risk management. - In 2018, additional provisions were introduced under the Pillar II capital review to reflect a risk-based assessment of continued ML/TF risks. ### Aspects considered in SREP (listed) - Business model: - (i) business strategy and drivers of profitability (geographies, products, clients), and - (ii) use of legal entities/channels to do business (particularly in jurisdictions with high ML/TF risks); - Internal governance and controls: - (i) level of managerial awareness for ML/TF risks and degree of [strategic/operational] engagement in mitigating actions, and - (ii) effectiveness of internal controls/reporting system; - Operational risk: - (i) understanding/addressing sources of events leading to ML/TF-related sanctions/measures, and - (ii) adequacy of provisions and economic capital for known litigation risk (including sanctioning/investigations procedures); - Liquidity risk: - (i) bank reliance on funding sources more exposed to ML/TF risk, and - (ii) capacity of servicing payment obligations if access to settlement services (e.g., in USD) through correspondent banking relationships were suspended due to ML/TF sanctions; and - Credit risk: - (i) relevant indicators assessed under the review of business model as well as internal governance and control (see above), - (ii) secured lending collateralized with cash, and - (iii) implementing KYC approach in loan underwriting process. *International Monetary Fund staff calculations and underlying calibration notes as presented in source.* --- _Source: https://www.imf.org/-/media/files/publications/cr/2019/1lvaea2019002.pdf_