## HIGH-FREQUENCY MONITORING OF THE SALVADORIAN ECONOMY ## Source details **Canonical URL:** [HIGH-FREQUENCY MONITORING OF THE SALVADORIAN ECONOMY](https://www.imf.org/-/media/files/publications/cr/2018/cr18152-elsalvadorsi.pdf) ## Other formats - [Markdown version](/-/media/files/publications/cr/2018/cr18152-elsalvadorsi.pdf.md) - [Structured JSON version](/-/media/files/publications/cr/2018/cr18152-elsalvadorsi.pdf.json) --- ### Introduction and objective - Purpose: develop a simple nowcast model for an early assessment of the Salvadorian economy to complement the Central Reserve Bank of El Salvador (BCR) nowcast. - Motivation: quarterly GDP statistics in El Salvador are released by the Central Bank with a 3-month delay. - Objective: build a disaggregated nowcasting tool using a bottom-up bridge model from the production side to nowcast GDP by supply components (sectoral nowcasts aggregated to GDP). - Main pre-release finding: estimated GDP growth rate in 2017Q4 is 2.4 percent y/y, implying average GDP growth of 2.3 percent in 2017; result consistent with official statistics released on March 23, 2018. ### Nowcasting approaches, data and variable treatment - Methods reviewed: autoregressive models, bridge equations, MIDAS regressions, vector autoregressive models (standard and mixed-frequency), dynamic factor models. - Trade-offs: - Monthly-data models generally outperform autoregressive models based on quarterly data. - Dynamic factor models produce more accurate nowcasts. - Bridge and MIDAS simpler and easier to interpret but less accurate. - BCR implementation: Kalman filter/dynamic factor model (Stock and Watson (1991) style as implemented by Camacho and Quiroz (2011)); uses 17 monthly indicators; provides a coincident indicator and short-term forecast for aggregate GDP only. - Data: - Monthly high-frequency variables over 2005-2017; dataset limited by revised national accounts and monthly economic activity indicator starting in 2005. - Coverage: real sector, monetary and financial sector, external sector, firm expectations (FUSADES), U.S. indicators. - Treatment: - Analysis uses annual growth rates (first difference of logged variables relative to same month in previous year). - Variables in U.S. dollars deflated using CPI or PPI. - Monthly series averaged up to the quarter when needed; seasonal adjustment not needed because analysis is on annual growth rates. - For unemployment rates, interest rates, and survey indicators, first differences used rather than growth rates. - Variable selection: started with a large set (Appendix Table A.1); retained variables with high explanatory power and statistically significant relationships with supply components (listed in Table 1). ### Supply-side bridge model specification - Aggregation: GDP nowcasted as weighted sum of nowcasted growth rates of sectoral GDP components, with weights equal to sector shares. - Sector weights (shares in parentheses): - Agriculture, farming forestry and fishing (6 percent) - Construction (6 percent) - Manufacturing and mining (17 percent) - Electricity, gas and water supply (4 percent) - Transportation and storage (5 percent) - Information and communication (4 percent) - Commerce (14 percent) - Hotels and restaurants (3 percent) - Financial services and insurance (7 percent) - Real estate (8 percent) - Government and social services (17 percent) - Other services (10 percent) - Regression framework: - For each supply component, linear OLS regressions link annual growth rate of the component to quarterly transformations of related monthly indicators. - Transformations by data type: - Flow variables: X is the sum of the monthly observations in the quarter. - Rates and indices: X is the average of the monthly observations in the quarter. - Stock variables: X is the end-of-quarter value (value in the third month). - Equations estimated by OLS using quarterly historical data for 2005-2014; 2015-2017 used to evaluate out-of-sample predictions. - Publication delays and updates: - High-frequency indicators have 1-, 2- or 3-month publication delays; data arrive non-synchronously. - Nowcasts are successively updated within the quarter as additional monthly observations become available. - Regression coefficients kept fixed; revisions driven solely by newly released indicator values. ### Variables used for sectoral nowcasts (selected mapping and publication delays) - Agriculture, Hunting, Forestry & Fishing: IVAE_A (2 months); IMP_INT_A (1 month) - Construction: IVAE_CONS (2 months) - Manufacturing Industry and Mining: IVAE_IP (2 months); ISSS_IP (2 months); IMP_INT_MAN (1 month) - Electricity, Gas and Water: PROENER (3 months); IVAE_IP (2 months) - Transportation and Storage: TCRGPO R (3 months); IMP_K_TRAN (1 month) - Commerce: IVAE_CTRH (2 months); TAX_VAT (1 month); ISSS_CTRHIC (2 months); DECOMEMPL (1 month) - Hotels and Restaurants: IVAE_CTRH (2 months); ENTPSAJ (3 months); TAX_VAT (1 month); ISSS_CTRHIC (2 months) - Information and Communication: IVAE_IC (2 months) - Financial Services and Insurance: IVAE_FS (2 months) - Real Estate: IVAE_RE (2 months) - Government and Social Services: IVAE_GSS (2 months) - Other Services: ISSS_PUB (2 months); IVAE_OS (2 months) ### Nowcast computation and treatment of missing/recent data - Nowcast computed by applying historical OLS coefficients to new data releases. - Treatment by variable type: - Flow: quarter sums used directly. - Index: growth rates computed from averages rather than sums. - Rate: compute differences based on averages. - Stock regressor: growth rate calculated using last data point available (end-of-quarter value). - Missing data at the end of the sample addressed via the described sums/averages approach; alternatives noted include univariate forecasting of regressors or Kalman filter/state-space approaches (Hahn and Skudelny, 2008). ### Fit and out-of-sample performance - Out-of-sample nowcasts for each sector over 2015Q1-2017Q3 show good fit, partly reflecting consistency between quarterly national accounts and the monthly economic activity indicator. - Sectors with poorer fit: - Electricity, Gas and Water - Transportation and Storage - Hotels and Restaurants - Poorer fit attributed to absence of disaggregated monthly activity indicators; pooling with other sectors reduces precision. - Improvement potential: availability of more high-frequency variables with shorter publication delays would improve sectoral fit and predictions. ### Nowcast results and revision chronology (2017Q4 example) - Nowcasted GDP growth for 2017Q4: 2.4 percent (nowcast as of January 31, 2018). - Given actual values for Q1-Q3, nowcasted GDP growth for 2017 with data as of January 31, 2018: 2.3 percent. - Sectoral nowcast evolution (percent) — Nowcast Periods Oct-17 (t-2), Nov-17 (t-1), Dec-17 (t), Jan-18 (t+1): - Agriculture, Hunting, Forestry & Fishing: 1.0, 1.0, 0.7, 0.4 - Construction: 5.3, 5.3, -4.1, 0.3 - Manufacturing Industry & Mining: 3.9, 3.9, 2.4, 3.7 - Electricity, Gas & Water: 0.5, 0.5, -1.5, 4.8 - Transportation and Storage: 3.4, 3.4, 4.2, 0.5 - Information and Communication: 1.2, 1.2, 6.0, 5.5 - Commerce: 4.0, 3.9, 3.5, 5.9 - Hotels and Restaurants: 2.5, 3.5, 2.7, 2.9 - Financial Institutions and Insurance: 3.8, 3.8, 2.0, 1.9 - Real Estate: 2.0, 2.0, 2.1, 2.2 - Government and Social Services: 1.3, 1.3, 0.8, 0.5 - Other Services: 2.5, 2.5, 5.6, 3.3 - Nowcasted GDP Growth: 2.7, 2.8, 2.1, 2.4 - Narrative of revisions: - Nov-2017: nowcast rises from 2.7 to 2.8 percent driven by Hotels and Restaurants acceleration, partly offset by Commerce slowdown. - Dec-2017: nowcast falls to 2.1 percent driven by slowdowns in Construction and Electricity, Gas and Water; partially offset by Transportation, Information and Communication, and Other Services. - Jan-2018: nowcast rises to 2.4 percent due to positive signals in Construction and Electricity, Gas and Water, and some industrial production improvements; primary sector, Financial Services and Insurance, and Government Services remained weak. - Lower Government Services growth may reflect fiscal consolidation efforts toward end-2017. ### Conclusions and suggested improvements for the nowcast system - Conclusions: - A simple disaggregated bridge nowcasting model can produce timely GDP estimates consistent with official statistics (2017 nowcast: 2.3 percent). - The disaggregated approach helps interpret drivers of GDP and the incremental impact of new monthly information. - Model is complementary to BCR’s dynamic factor approach and technically simple to implement. - Suggested next steps: - (i) Consider alternative methods to handle missing end-of-sample data (e.g., univariate forecasts for regressors). - (ii) Consider alternative bridge equation specifications, including lags of regressors and changing equations over the forecast cycle (Hahn and Skudelny (2008)). - (iii) Replicate methodology using demand-side components now that revised national accounts include demand decomposition. *Source: cr18152-elsalvadorsi (IMF staff content).* ### References _______________________________________________________________________________ 17 ### HIGH-FREQUENCY MONITORING OF THE SALVADORIAN ECONOMY ### Introduction - Purpose: develop a simple nowcast model for an early assessment of the Salvadorian economy to complement the Central Reserve Bank of El Salvador (BCR) nowcast. - Motivation: quarterly GDP statistics in El Salvador are released by the Central Bank with a 3-month delay. - Main finding (pre-release): the estimated GDP growth rate in the 4th quarter of 2017 is 2.4 percent y/y, leading to an average GDP growth rate of 2.3 percent in 2017. This is in line with official statistics released on March 23, 2018. ### Nowcasting approaches and rationale - Common methods reviewed: autoregressive models, bridge equations, MIDAS regressions, vector autoregressive models (standard and mixed-frequency), dynamic factor models. - Trade-offs noted: - Models using monthly data generally outperform autoregressive models based only on quarterly data. - Dynamic factor models produce more accurate nowcasts relative to other specifications. - Bridge and MIDAS equations are simpler and easier to interpret and communicate, although they reduce forecast accuracy relative to more sophisticated techniques. - BCR implementation: - Uses a Kalman filter/dynamic factor model (Stock and Watson (1991) style as implemented by Camacho and Quiroz (2011)). - Considers 17 monthly indicators and provides a coincident indicator and short-term forecast for aggregate GDP, but not for components. ### Objective of this exercise - Develop a disaggregated nowcasting tool using a bottom-up bridge model from the production side to nowcast GDP by supply components (sectoral nowcasts aggregated to GDP). - Rationale: disaggregated (sectoral) approach facilitates interpretation, communication, and identification of drivers of GDP forecast. ### Data description - Frequency and sample: monthly high-frequency variables over the period 2005-2017; dataset limited by availability of revised national accounts and monthly economic activity indicator starting in 2005. - Coverage: real sector, monetary and financial sector, external sector, indicators of expectations of local firms, indicators for the U.S. economy; includes both “hard” indicators and “soft” indicators (FUSADES enterprise survey); financial variables included. - Treatment: - Analysis uses annual growth rates. - Variables in U.S. dollars deflated using CPI or PPI (trade variables). - Monthly series averaged up to the quarter when needed. - Seasonal adjustment not needed because analysis is on annual growth rates. - Growth rates computed as first difference of logged variables relative to same month in previous year. - For unemployment rates, interest rates, and survey indicators, first differences rather than growth rates are calculated. - Variable selection: - Start with a large set of monthly indicators (see Appendix Table A.1). - Selection based on economic judgement and statistical testing: only variables with high explanatory power and statistically significant relationships with supply components are retained (listed in Table 1). ### Methodology: Supply-side bridge model - Approach: - Bottom-up bridge model from the production side: GDP growth nowcasted as weighted sum of nowcasted growth rates of sectoral GDP components, with weights equal to sector shares. - Sector weights (shares in parentheses): - Agriculture, farming forestry and fishing (6 percent) - Construction (6 percent) - Manufacturing and mining (17 percent) - Electricity, gas and water supply (4 percent) - Transportation and storage (5 percent) - Information and communication (4 percent) - Commerce (14 percent) - Hotels and restaurants (3 percent) - Financial services and insurance (7 percent) - Real estate (8 percent) - Government and social services (17 percent) - Other services (10 percent) - Regression framework: - For each supply component, linear OLS regressions link annual growth rate of the component to aggregate/quarterly transformations of related high-frequency indicators. - Transformations by data type: - Flow variables: ܺ is the sum of the monthly observations in the quarter. - Rates and indices: ܺ is the average of the monthly observations in the quarter. - Stock variables: ܺ is the end-of-quarter value (value in the third month). - Equations estimated by OLS using quarterly historical data for 2005-2014; 2015-2017 used to evaluate out-of-sample predictions. - Publication delay and nowcast updates: - High-frequency indicators have 1-, 2- or 3-month publication delays; data arrive non-synchronously. - Nowcasts are successively updated within the quarter as additional monthly observations become available. - Example: October/November/December 2017 publication timing summarized in Table 2 and explained: a nowcast as of December 2017 incorporates more recent observations (e.g., October and November for 1-month delay series) than the November 2017 nowcast. ### Variables used for sectoral nowcasts (as listed) - Agriculture, Hunting, Forestry & Fishing: - IVAE_A (2 months) - IMP_INT_A (1 month) - Construction: - IVAE_CONS (2 months) - Manufacturing Industry and Mining: - IVAE_IP (2 months) - ISSS_IP (2 months) - IMP_INT_MAN (1 month) - Electricity, Gas and Water: - PROENER (3 months) - IVAE_IP (2 months) - Transportation and Storage: - TCRGPO R (3 months) - IMP_K_TRAN (1 month) - Commerce: - IVAE_CTRH (2 months) - TAX_VAT (1 month) - ISSS_CTRHIC (2 months) - DECOMEMPL (1 month) - Hotels and Restaurants: - IVAE_CTRH (2 months) - ENTPSAJ (3 months) - TAX_VAT (1 month) - ISSS_CTRHIC (2 months) - Information and Communication: - IVAE_IC (2 months) - Financial Services and Insurance: - IVAE_FS (2 months) - Real Estate: - IVAE_RE (2 months) - Government and Social Services: - IVAE_GSS (2 months) - Other Services: - ISSS_PUB (2 months) - IVAE_OS (2 months) ### Key findings and performance - Nowcast estimate for 2017Q4: 2.4 percent y/y GDP growth. - Nowcast implied average GDP growth for 2017: 2.3 percent. - Consistency: nowcast results are in line with official statistics released on March 23, 2018. ### Implementation notes and next steps - The bridge model focuses on supply-side disaggregation because, at project start, national accounts lacked demand-side disaggregation; with the revised national accounts released on March 23, 2018 (which include demand decomposition), nowcasting by demand component is now possible and is a natural next step. - The model exploits non-synchronous data arrival to produce intra-quarter updates and to assess incremental information content of newly released monthly observations. *Prepared by Ana Lariau.* ### 12. The nowcast is computed by applying the coefficients estimated with the historical ### 12. The nowcast is computed by applying the coefficients estimated with the historical ### Methodology - Nowcast computed by applying coefficients estimated with historical data to new data releases. - Nowcast example: growth rate of GDP in sector ݆ in the last quarter of 2017 (ܲܦܩ∆　 ෣ ଶ଴ଵ଻ொସ ௝ ). - Estimated coefficients denoted by ߙො, ߚ መ and Θ ෡ ௜ for Equation (1). - Given non-synchronous arrival of data, the nowcast is updated throughout the quarter as additional information becomes available. - Treatment of variable types: - If a dependent variable is a flow, calculations in equations (2)-(4) apply directly. - If a dependent variable is an index, growth rates in brackets in equations (2)-(4) are computed with averages rather than sums. - If a dependent variable is a rate, compute the difference (rather than the growth rate), based on averages and not sums. - If a regressor is a stock variable, its growth rate is calculated with the last data point available rather than with sums or averages. - Note: calculations in equations (2)-(4) are a simplest way to address missing data at the end of the sample due to publication lags; alternatives include univariate forecasting techniques for regressors (Hahn and Skudelny, 2008) or Kalman filter/state-space approaches. ### Fit and out-of-sample performance - Out-of-sample nowcasts for each sector over 2015Q1-2017Q3 show good fit, partly reflecting consistency between quarterly national accounts and the monthly economic activity indicator when used. - Sectors with poorer fit include: - Electricity Gas and Water - Transportation and Storage - Hotel and Restaurants - Poorer fit in these sectors attributed to absence of disaggregated monthly activity indicators; pooling with other sectors reduces precision. - Improvement potential: availability of more high-frequency variables with shorter publication delays would improve sectoral fit and predictions. ### Nowcast Results (2017Q4 and 2017 annual) - GDP growth in 2017Q4 is nowcasted at 2.4 percent. - Given actual values for Q1-Q3, the nowcasted GDP growth for 2017 with data as of January 31st, 2018, stands at 2.3 percent. - Consistency: result consistent with the actual GDP growth displayed by the revised statistics published on March 23, 2018. - Table 3 (El Salvador: GDP Growth Nowcast for 2017Q4 as of January 31, 2018) — Quarter / GDP Growth (percent) / Type of Figure: - 2017 Q1 — 3.4 — Actual - 2017 Q2 — 0.3 — Actual - 2017 Q3 — 3.1 — Actual - 2017 Q4 — 2.4 — Nowcast - 2017 — 2.3 — Actual + Nowcast ### Revision process and drivers (chronology of nowcast updates) - Regression coefficients remain unchanged across updates; nowcast revisions are driven solely by incremental information from new data releases. - Evolution of nowcast for 2017Q4 (Nowcast Period t-2 t-1 t t+1; Date Oct-17, Nov-17, Dec-17, Jan-18) — sectoral nowcasts (percent): - Agriculture, Hunting, Forestry & Fishing: 1.0, 1.0, 0.7, 0.4 - Construction: 5.3, 5.3, -4.1, 0.3 - Manufacturing Industry & Mining: 3.9, 3.9, 2.4, 3.7 - Electricity, Gas & Water: 0.5, 0.5, -1.5, 4.8 - Transportation and Storage: 3.4, 3.4, 4.2, 0.5 - Information and Communication: 1.2, 1.2, 6.0, 5.5 - Commerce: 4.0, 3.9, 3.5, 5.9 - Hotels and Restaurants: 2.5, 3.5, 2.7, 2.9 - Financial Institutions and Insurance: 3.8, 3.8, 2.0, 1.9 - Real Estate: 2.0, 2.0, 2.1, 2.2 - Government and Social Services: 1.3, 1.3, 0.8, 0.5 - Other Services: 2.5, 2.5, 5.6, 3.3 - Nowcasted GDP Growth: 2.7, 2.8, 2.1, 2.4 - Narrative of revisions: - November 2017: nowcast increases slightly from 2.7 percent to 2.8 percent due to acceleration in Hotels and Restaurants, partially offset by slowdown in Commerce. - December 2017: nowcast revised downwards to 2.1 percent driven by slowdowns in most sectors, particularly Construction and Electricity Gas and Water; partially offset by improvements in Transportation, Information and Communication, and Other Services. - January 2018: nowcast revised up to 2.4 percent due to positive signals in Construction and Electricity, Gas and Water, and to a lesser extent Industrial Production; several services activities also improved. Persistent slowdowns observed in the primary sector, Financial Services and Insurance, and Government Services. - Lower growth rate of Government Services may reflect fiscal consolidation efforts towards end-2017. ### Conclusions and suggested improvements - Main conclusions: - The note proposes a simple nowcasting model for high-frequency monitoring of the Salvadorian economy based on a bridge model using information for 2005-2017 from a large set of higher-frequency, earlier-published variables. - Nowcasted GDP growth in 2017 is 2.3 percent, consistent with BCR data published in March 2018. - The disaggregated bridge model is technically simple and helps understand drivers of the economy and impacts of new information on each GDP component; it is complementary to the model used by the BCR. - Suggested future improvements / lines of work: - (i) Consider alternative methods to deal with missing data at the end of the sample period, e.g., using univariate techniques to produce forecasts of the regressors. - (ii) Consider alternative specifications of the bridge equations, by incorporating lags of the regressors and by changing the estimated equation over the forecast cycle as suggested by Hahn and Skudelny (2008). - (iii) Replicate methodology using demand-side components rather than supply-side components to provide a more comprehensive short-run view and to determine whether weak GDP growth is due to subdued growth of a specific demand component or structural problems/shocks to a particular economic sector. ### Annex I — Data description (selected highlights) - Data period for bridge model: period 2005-2017 for many series; comprehensive list spans monthly and quarterly series with varying start dates and end at Dec-17 or Nov-17 for some series. - Key high-frequency indicators used (sample entries): - IVAE (Monthly BCR) — Jan-05 to Dec-17 - IVAE_A (Agriculture) — Jan-05 to Dec-17 - IVAE_CONS (Construction) — Jan-05 to Dec-17 - ISSS contributors: Total — Jan-99 to Nov-17 (Monthly ISSS) - CPI (Consumer Price Index) — Jan-90 to Dec-17 (Monthly DIGESTYC, published by BCR) - REM (Remittances) — Jan-91 to Dec-17 (Monthly BCR) - PUB_SPEND (Government spending) — Jan-94 to Dec-17 (Monthly Ministry of Finance, published by BCR) - CREDIT (Total credit) — Jan-02 to Dec-17 (Monthly BCR) - Selected external/US indicators included (e.g., PIBTUSA US Quarterly GDP Mar-90 to Dec-17; FEDFE Federal Funds Effective Rate Jan-90 to Dec-17). - Full variable list and start/end dates are provided in Table AI.1 (Data Description). *Source: Fund staff estimates and Central Reserve Bank of El Salvador as presented in the content unit.* ### 1. El Salvador’s population has remained relatively young but this is set to change. Like ### 1. El Salvador’s population has remained relatively young but this is set to change. Like ### Demographic outlook and implications - Current demographic dividend: El Salvador “continues to enjoy a demographic dividend.” - Aging projections: - Share of the population over 64 relative to population ages 15–64 projected to rise from 13 percent currently to 28 percent in 2050 and 69 percent in 2100. - Median age projected to rise from 27 currently to 42 in 2050 and 53 in 2100. - Emigration effect: “Most of the emigrants have tended to be relatively young thereby pushing upwards the average age of those who stay behind.” ### Salvadoran pension system — historical background - Original system: “Heavily subsidized defined-benefit (DB) scheme” with very low contribution and high guaranteed replacement rates, which generated pronounced actuarial and fiscal imbalances; by mid-1990s “fiscally untenable.” - Late-1990s reform: transition to defined-contribution (DC) system based on individual accounts managed by private pension funds; phase-in for younger cohorts created fiscal “transition costs” as payroll contributions shifted away from public revenue while legacy DB entitlements were financed from fiscal accounts. Transition costs were expected to dissipate by 2030 in the original design. - Expected benefits of DC reform: long-term fiscal burden was assessed as among the lowest in Latin America and introduction of private accounts was expected to bolster low labor force participation and coverage. - Implementation problems: - Grandfathering decisions in 2003 and 2006 guaranteed DBs to early DC retiree cohorts, greatly increasing transition costs and straining fiscal accounts by 2016–17. - Low asset returns and linkage of returns on public pension bonds to LIBOR (government decision in 2006) depressed returns further, especially after the 2008 global financial crisis. - Coverage remained weak: affiliated individuals comprised only a quarter of the economically active population; worker contribution density was weak and declining. ### Pre-2017 and 2016 proposals - 2016 government “mixed” system proposal (abandoned): - Transfer of more than one-half of pension-related payroll contributions and assets to the public sector. - Flat public pension benefits corresponding to the contributory minimum pension. - Downsized DC pillar for higher earners only. - Intended to reduce short-to-medium-term fiscal transition costs but would generate significant permanent DB obligations; lacked political support. - Early-2017 private sector proposal: - Sought to protect privately-managed assets and relieve government of significant transition costs by re-directing part of payroll contributions to cohorts with guaranteed DBs (without recourse to government funds). - Criticism: could depress replacement rates for future DC pensioners due to reduced contributions to individual accounts. ### 2017 reform (legislated September 29, 2017) — rationale and political context - Urgent political compromise due to difficult financing projected for Q4 2017, adverse investor sentiment, ratings downgrades, and April 2017 missed payment on pension bonds. - Private sector’s proposal used as basis, with last-minute modifications adding government-backed guarantees to augment contributions to individual accounts. - Congressional support: 71 of 82 congressmen voted in favor. ### Key elements of the September 2017 reform Re-calibration of cash flows: - Increase in the pension payroll contribution rate from 13 to 15 percent. - Marginal cut in pension fund fees: from 2.2 to 2 percent in 2018 and to 1.9 percent starting from 2020. - Diversion of 5 percentage points of payroll contributions from individual account holders to a privately managed solidarity guarantee account (SGA): 3 percentage points temporarily diverted; 2 percentage points permanently set aside to finance longevity benefits. - Use of the SGA to pay: (i) public DBs to those who benefitted from top-ups in the 2000s and opted for the private system (optados), (ii) the DC system’s minimum pension guarantee, and (iii) longevity benefits. - Higher interest rate on government pension bonds to support pension fund returns: - Nominal interest rate on new issuances set at 6 percent. - Interest rate on the old stock to gradually increase from 2.5 to 4.5 percent between 2018 and 2022. - Grace period (either 3 or 5 years for old bonds) and lengthening maturity of pension bonds from 25 years previously to 30–50 years. Changes in pension benefits and parameters: - Reductions in guaranteed DBs for the optados who have yet to retire: their pension capped at 55 percent of the “basic wage” instead of around 68 percent previously. - Cap on the maximum pension for the optados at USD 2,000 per month. - Progressive levy of between 3 and 10 percent depending on the amount on existing pension benefits; proceeds accrue to the SGA. - More stable DC pension benefits over time through changes to calculation methods (individual pension levels would no longer be declining during the retirement phase and would be kept stable in real terms), supported by longevity benefits. - New option for periodic pension benefits for those who contribute between 10 and 25 full years (previously only lump-sum withdrawal was available). Institutional changes: - Actuarial committee to help ensure long-term financial sustainability and enable small increases in the retirement age (up to a maximum of one year every 5 years, starting from 2022). - Improved payroll collection processes and data enhancements. - Risk committee to improve returns and diversification of investments. - Enhanced analysis and reporting, including periodic actuarial studies of long-term sustainability. - Additional options and financial intermediaries to diversify pension fund investments. - Option of anticipated withdrawal from pension accounts (up to 25 percent of balances) for eligible individuals. Government guarantees and resources: - Permanent allocation of 2.5 percent of current revenues (about 0.5 percent of GDP) starting from 2020, with temporarily smaller allocations of 1.7 and 1.8 percent of revenues in 2018 and 2019 respectively, to pay public pension benefits and back-up government liabilities. - Reimbursement from the budget of 3 percentage points of payroll contributions diverted to the SGA to pay defined benefits to account holders upon retirement, together with a return equal to that earned by the conservative fund in the system. - Reimbursement of a portion of the return on longevity benefits to those who take lump-sum withdrawal. - Government guarantee to cover SGA operations should the SGA run deficits. - Continuation of issuance of pension bonds to provide additional resources for legacy publicly guaranteed defined-benefit pensions. Implementation status (early-2018) - By-laws mostly enacted by early-2018, but operationalization depends on follow-up regulations and institution-building. - Central bank tasked to operationalize SGA accounting, anticipated withdrawal modalities, upgraded accounting norms for pension funds, calculation of technical requirements to determine pension rights, and other regulations. - Actuarial and risk committees still needed to be made fully operational. ### Effects of the 2017 reform — fiscal and coverage impacts Overall decomposition: - Two main fiscal effect blocks: 1. Re-distribution of financial flows in the system — generally favorable for fiscal accounts in the short-to-medium run but potentially unfavorable in the longer run because of government guarantees. 2. Underlying parameter changes (cuts in DBs, increases in retirement age, increase in payroll contribution rate) — favorable for fiscal sustainability with effects that are initially small but grow progressively over time. Short-to-medium-term effects on the fiscal deficit: - Diversion of social contributions to the SGA (3 percent of payroll in the first 10 years, or 0.7 percent of GDP annually) to pay public DBs makes the largest immediate impact on reducing the fiscal deficit. - Other fiscal savings initially estimated to be relatively small (less than 0.1 percent of GDP combined, annually): reductions in DBs, levies on pensions, and savings on the minimum pension guarantee. - Interest rate change on pension bonds limited immediate effects; interest rate set at 2.5 percent for 2018 for existing bonds. - Total estimated savings around 0.8 percent of GDP annually in the first few years after the reform. Short-term effects on fiscal financing (below-the-line relief): - Grace period on old pension bonds generates savings of around ½ percent of GDP annually in those years compared with the counterfactual. - Lengthening of bond maturities limits principal payments beyond 2020 relative to the counterfactual. - SGA inflows corresponding to the 2 percent of payroll (0.4 percent of GDP) contribution to longevity benefits could finance certain pension system obligations given longevity spending is backloaded (significant by around 2040). - Overall financing effects favorable in 2018–23 and estimated in the range of ½ to 1 percent of GDP annually, varying by year. Longer-term fiscal effects and actuarial study convergence - Counterfactual: pre-reform baseline where long-term fiscal burden of pensions was expected to decline as DC system fully phased in. - Actuarial studies differ and have methodological weaknesses (difficulty integrating pension flows with consistent macro projections; incomplete accounting of pension-related fiscal liabilities, notably interest on pension debt). - Points of convergence across studies: - The reform would reduce fiscal costs over the next quarter-century mainly because some DBs will be paid by the SGA. - After around 2040, the reform would increase pension system costs because: (i) the SGA’s 3 percent of payroll contribution would be phased out while costs rise; (ii) increase in liabilities from the government guarantee to reimburse flows accruing to the SGA; and (iii) growing liability for longevity benefits of the SGA and potentially the government. - Overall long-term fiscal costs of pension benefits are likely to be contained (e.g., below 2 percent of GDP annually) barring major economic under-performance (e.g., real GDP growth well below the estimated potential rate of 2.2 percent) or ad-hoc decisions to increase benefits. Tabled and graphical estimates referenced in text (selected numeric points preserved from source): - Increase in pension payroll contribution rate: from 13 to 15 percent. - Pension fund commission cuts: 2.2 → 2 percent in 2018 → 1.9 percent from 2020. - Diversion to SGA: 5 percentage points of payroll (3 percentage points temporary; 2 percentage points permanent for longevity benefits). - New nominal interest rate on new pension bond issuances: 6 percent. - Gradual increase in interest rate on old stock of pension bonds: from 2.5 to 4.5 percent between 2018 and 2022. - Grace period on old bonds: either 3 or 5 years. - Lengthening maturity of pension bonds: from 25 years previously to 30–50 years. - Cap on maximum pension for optados: USD 2,000 per month. - Anticipated SGA-related fiscal relief in first 10 years: 3 percent of payroll diversion ≈ 0.7 percent of GDP annually. - Estimated initial total savings from reform: around 0.8 percent of GDP annually in first few years. - Financing relief from grace period: around ½ percent of GDP annually in relevant years. - Longevity contribution inflow: 2 percent of payroll ≈ 0.4 percent of GDP. - Assumed real GDP growth in actuarial calculation context: 2.2 percent over the entire horizon (estimated potential growth). *Source: cr18152-elsalvadorsi (IMF staff content).* ### 14. Rules and implications for the fiscal backstop. As described above, the reform envisions ### 14. Rules and implications for the fiscal backstop ### Fiscal backstop: rules, scope, and uncertainties - The reform envisions several channels through which the government could provide resources and guarantees to the pension system, but there is substantial uncertainty whether, and in what measure, such channels could be activated. - Sources of uncertainty: - Differences in actuarial projections: private pension funds’ preliminary projections indicate the SGA would not run deficits in the future, while other projections (Melinsky, 2017) suggest a government guarantee is likely to be triggered relatively soon. - Lack of clarity on scope and sequencing of operations in the law/framework. - Key not-fully-settled issues: - Potential deficit of the SGA: actuarial studies differ; Asafondos and SSF studies were being elaborated but full results were not available at the time of the Selected Issues paper. The SSF study uses a granular individual-specific database; most other studies used simplifying assumptions. - Scope of government guarantees: the law references specific public government guarantees but does not fully clarify whether the budget allocation of around ½ percent of GDP would effectively cap spending on all or some guarantees, or whether more resources could be provided through other mechanisms. The law acknowledges the allocation could be exceeded to pay “legacy” defined benefit pensions but does not clarify if it can be exceeded for other purposes. - Sequencing of fiscal backstop operations: Article 224 mentions intended uses of the dedicated allocation (i) payment of “legacy” public pension benefits; (ii) payment of minimum contributory pensions; and (iii) reimbursement of government guarantees related to diverted contributions and lump-sum withdrawals. It is unclear if the order of mentions implies prioritization should the aggregate allocated amount prove constraining. ### Pension adequacy: effects and projections - The 2017 reform aims to address low DC-based pension benefits; expected effects are relatively modest and backloaded. - Main channels to improve adequacy: - Higher interest rate on pension bonds to increase returns on individual accounts over time. - Recalibration of the method of calculating pension benefits to make benefits more stable and slightly higher. - Slightly increasing allocation of total funds accruing to individual accounts from 10.8 to 11.1 percent of payroll. - Preliminary calculations and expected replacement rates: - Replacement rates in the DC system would increase relatively modestly and would, at least for a few initial years, be in the 25-40 percent range. - Observed low replacement rates for the first cohort of unsubsidized DC pensioners that retired in 2017 (e.g., women born in 1962) support modest improvement expectations. - Conclusion: pension adequacy remains an important challenge going forward. ### Coverage of the pension system - The reform by itself is unlikely to significantly raise poor coverage, which is driven by low formal labor market participation. - New options allowing individuals with less than 25 full years of contributions to receive stable pension income and public health care services could incentivize better coverage, but: - Risk that the proportion of such individuals is small because evidence suggests “low-density” contributors favor lump-sum withdrawal of pension account balances. - The option of anticipated withdrawal for current consumption may increase attractiveness of the individual account system but would defeat the purpose of incentivizing saving over consumption. ### Overall assessment of the reform’s fiscal impact - The reform to the DC pension system on balance improves fiscal sustainability in the context of projected population aging. - Key risks prior to reform: - The DC system was becoming untenable because (i) it was already a financing strain on the budget and (ii) future pension levels under pure DC rules were likely too low to be politically acceptable. - How the reform addresses risks: - Lessens risks by addressing funding bottlenecks and most extra costs imposed by ad-hoc grandfathering of early DC cohorts. - Includes measures to increase attractiveness of future DC benefits, though effects will likely be modest. - Large bipartisan support suggests reinforced viability of the DC system for now. - Long-term liabilities: emergence of additional long-term fiscal liabilities is of some concern but they appear relatively contained and backloaded. ### Avenues for further improvement (policy options and recommendations) - Increase the retirement age: - Retirement age unchanged for over 2 decades and is among the lowest in the region. - Current rule: a modest increase by a maximum of 1 year is considered only in 2022, with further envisioned increases capped at 1 year for each future 5-year period. - Recommendation: more ambitious and frontloaded increases could improve future replacement rates in the DC system and limit fiscal contingent liabilities from SGA operations due to longevity benefits. - Improve benefits coverage for the poor and vulnerable: - Changes to law improve coverage only among contributors and do not help the most vulnerable not affiliated to the system. - The DC-based model provides limited redistribution; better distributional outcomes could be enhanced by gradually expanding the non-contributory basic pension (currently $50 per month, a quarter of the contributory minimum pension). - Total cost of this pension to all individuals over 70 is estimated at around 0.8 percent of GDP annually; coverage could be expanded gradually starting from the lowest income categories at a much more modest cost. - Enhance cost-efficiency: - Pension fund fees remain too high and should be reduced given the low risk profile of investments (mainly government bonds). - Anticipated withdrawal of balances for current consumption should be closely monitored and reconsidered if it represents significant administrative or other costs. - Ensure effective backstopping from the budget: - Solvency and liquidity ultimately hinge on the budget. - The envisioned budget allocation (of about ½ percent of GDP annually) alone would be insufficient to cover pension obligations in most years; the broader fiscal guarantee mechanism is yet to be fully clarified. - To ensure credibility, trust, and to minimize costs from disputed claims and uncertainty, pension system accounts should be transparently integrated in fiscal decisions, with full costing of their implications. - The actuarial and risk committees and requirements for actuarial reviews created by the reform offer a promising basis for better transparency and decision-making, contingent on the efficiency of those institutions and processes. ### Timing and urgency - Cross-country experience indicates limited scope for procrastination: parametric reforms (e.g., raising retirement age) should be implemented early as they take time to receive political and social backing and much longer (a decade or more) to yield macro-relevant effects. - Potential risks from underestimating fiscal costs of new longer-term liabilities (government guarantees and longevity benefits) need close monitoring, with a view to promptly rolling out measures to address remaining sustainability risks. *Source: IMF Selected Issues chapter: "Rules and implications for the fiscal backstop."* ### 7. To identify the set of important competitors for El Salvador, we first identified ### 7. To identify the set of important competitors for El Salvador, we first identified ### Methodology for identifying competitors - Products considered: HS-4 digit products whose share in El Salvador’s total exports in 2014 was greater than 1 percent. This yielded 22 products, which together accounted for more than 60 percent of El Salvador exports. - Competitor classification: - A country is a competitor for a given product if its share in world exports of that product was greater than 1 percent and it was among the top fifteen exporters of that product. - An additional criterion: countries with share of world exports of that product between 0.4 percent to 1 percent were included if they were an emerging market (EM) or a low-income developing country (LIDC) to proxy for product quality (examples: Brazil, Guatemala). - Other Central America countries that did not meet the two criteria were also included. - Result: In all, 46 competitors were identified. - Index construction: - Value-Based Index (VBI): for each competitor pool data for all years and average VBI across products and destination markets to estimate average VBI over the entire sample period. - Count-Based Index (CBI): same pooling and averaging procedure applied to the CBI. - Data source noted: HS-6 level data from BACI World Trade Database (authors' calculations). ### Main competitors (global and by income group) - Overall identification: - Many advanced economies (AEs) such as the U.S., Germany and Japan appear as competitors, as do large developing economies like China and India. - AEs (value-based competition): - U.S. is by far the most dominant competitor, followed by Germany, Italy and Japan. - EMs/LIDCs (value-based competition): - China has the highest VBI, followed by Mexico, Guatemala, Brazil and Costa Rica. - Differences across indices: - For AEs, CBI picture remains quite stable relative to VBI, but Japan drops considerably as a competitor in CBI compared to its VBI ranking. - Among EMs/LIDCs, Guatemala drops notably in CBI versus VBI. - Thailand, Indonesia, Poland and Hungary are tougher competitors based on CBI than VBI. - CAPDR members are tougher competitors based on VBI compared to CBI, suggestive of greater specialization among CAPDR countries. ### Top competitors in El Salvador’s two main destination markets (U.S. and CAPDR) - U.S. market: - Among AEs in the U.S., Canada is the top competitor based on both count and value. - In EM/LIDC group for the U.S., Mexico and India eclipse China in terms of count, while China continues to be the top competitor in terms of value. - Guatemala drops significantly as a competitor in the U.S. compared to its global ranking. - Competition in the U.S. is higher based on count relative to the world market for overlapping top products. - CAPDR market: - CAPDR countries are generally much tougher competitors in CAPDR as a destination. - Among AE competitors in CAPDR, the distribution of VBI is more skewed than in the world market, with the U.S.’s dominance in CAPDR being significantly larger. - Top EM/LIDC competitors by VBI in CAPDR include Mexico, China, Guatemala, Panama, Costa Rica, Brazil, Colombia, Honduras, India, Peru, Nicaragua, Thailand, Indonesia, Ecuador, Malaysia, Turkey, Viet Nam, Poland, Cambodia, Pakistan (ordering shown in figures). ### Trends in competition over time (2000–2014) - Aggregate trend: - Competition from advanced economies has declined over time. - Competition from EM/LIDCs has increased over time, both from larger economies like China and smaller economies like Vietnam and Cambodia. - AEs (time series): - Count-based competition remained stable. - Value-based competition declined, driven largely by declines in VBI for U.S., Canada and Japan. - EM/LIDCs (time series): - Competition from China increased, especially in terms of value (more than two-fold increase globally). - In the U.S., China’s VBI increased two-fold; in CAPDR China’s VBI increased more than three-fold. - Vietnam in the U.S. is rapidly increasing CBI—becoming an important competitor by expanding the number of overlapping product lines—while its VBI declined in CAPDR. - In CAPDR, increases in competition observed from Mexico (due to VBI), India and Guatemala (due to CBI); decreases in competition observed from Costa Rica (due to VBI) and Sri Lanka (due to both VBI and CBI). ### Competition in the product space (top product lines) - Scope: - Product-level VBI and CBI for El Salvador: data pooled over all years and averaged across importers and competitors for each HS-4 product line. - Analysis restricted to major export products selected based on export shares (these are likely products where El Salvador has comparative advantage). - Top 10 products (by export share) together account for 45 percent of merchandise exports in 2014. - Key product findings (VBI highlights): - Highest levels of VBI: Cane/beet sugar, Electrical capacitors, Underpants of men’s/boys, Coffee. - Cane/beet sugar and Coffee also exhibit high levels of CBI. - T-shirts and vests, and Jerseys and pullovers (combined share in exports value is around 20 percent) exhibit high CBI but among the lowest VBI. - Medicaments exhibits a pattern similar to Electrical capacitors (relatively higher VBI). - Destination differences: - U.S. market: composition of top products is heavily skewed towards apparel; competition is higher in terms of count than in the world market for overlapping products. - World market: sees higher levels of value-based competition than the U.S. for top overlapping products. - CAPDR market: top products include much less apparel; competition is lower for both count and value compared to the world market. - Exports to CAPDR show greater diversification compared to exports to the U.S., but competition is much higher in the U.S. especially based on count. ### Evolution of competition by product over time (selected observations) - World market trends (CBI and VBI): - CBI: decline observed for Electrical capacitors; increases for Toilet paper, tissues and napkins; Tights, stockings, socks, and Underpants of men’s/boys (knitted or crocheted). - VBI: mostly stable, except: - Significant increase for Underpants of men’s/boys, knitted/crocheted in 2005 followed by a comparable decline by 2010. - Large increase for Toilet paper, tissues and napkins and for Cane/beet sugar. - U.S. market product trends: - CBI: substantial increases for Underpants of men’s/boys and Suits and ensembles of women/girls; Cane/beet sugar saw a big drop in count-based competition. - VBI: most top products saw increases in value-based competition, with Tights, socks and hosiery showing large increases; Electrical capacitors saw a sharp reduction in VBI. - CAPDR region product trends: - CBI: noticeable increases for Waters (including mineral and aerated waters) and Medicaments; declines in Bread, cakes, biscuits and Fabrics (knitted/crocheted). - VBI: general decline for most products, with the sharpest decline for Jerseys, pullovers, cardigans. ### Key statistics and counts preserved from analysis - Thresholds and counts: - Export share threshold to select products: greater than 1 percent. - Number of products selected: 22 products. - Combined share of those 22 products: more than 60 percent of El Salvador exports. - Competitor inclusion lower threshold for EM/LIDC proxy: share between 0.4 percent to 1 percent. - Total competitors identified: 46 competitors. - Time period referenced in indices and figures: averages and time series typically reported for 2000-14 (average, 2000-14) and time paths 2000–2014. - Reported magnitudes of change over time: - China: more than two-fold increase in VBI globally; two-fold increase in the U.S.; more than three-fold increase in CAPDR. - Descriptive shares: top 10 products account for 45 percent of merchandise exports in 2014; T-shirts and vests and Jerseys and pullovers combined share in exports value is around 20 percent. _International Monetary Fund. Excerpts and figures based on HS-6 level data from BACI World Trade Database and authors' calculations (average and time series 2000–2014)._ ### 14. Given the large share of textiles in the exports of El Salvador (about 46 percent), we ### cr18152-elsalvadorsi - 14. Given the large share of textiles in the exports of El Salvador (about 46 percent) ### Textiles — competition and dynamics - Textiles account for about 46 percent of El Salvador’s exports. - Regional concentration: Asian countries have emerged as top competitors. - VBI changes (2000–2014): - China’s VBI doubled over the 14-year period. - Vietnam’s VBI tripled. - Cambodia’s VBI increased by almost a factor 4. - U.S. VBI halved over the sample period (but remains the most dominant competitor after China). - South Korea’s VBI in 2014 was about one-third of the 2000 level. - CBI changes (2000–2014): - Almost all top competitors exhibit an increase in CBI. - Cambodia’s CBI increased by a factor of 3, followed by China and Vietnam. - Levels of competition (2014): - China and U.S. compete with El Salvador in more than 96 percent of El Salvador’s export products within the textile industry. - Textiles definition: HS2 chapters from 50 to 63. ### Electrical capacitors — market share and competitor shifts - Electrical Capacitors accounted for about 5 percent of total merchandise exports in 2014. - Competitor VBI shifts (2000→2014): - Japan and U.S. were top competitors in 2000; by 2014 China and Germany had taken their positions. - China increased competition very sharply compared to other major competitors. - Czech Republic, South Korea, and Malaysia also saw big increases in their VBIs. - Competitor CBI shifts (2000→2014): - U.S. and Japan saw modest declines in CBI. - France, UK, South Korea, and Germany experienced larger drops in CBI. - China and Mexico saw significant rises in CBI; Malaysia’s CBI remained steady. - Based on CBI in 2014: U.S., China, Japan, Mexico, and Czech Republic were the most dominant competitors. ### Cane/Beet sugar — competitor evolution - Sugar accounts for 3.3 percent of total exports. - VBI changes (2000→2014): - Brazil experienced the largest decline in VBI, followed by the U.S. - Other major competitors saw VBI increases, led by Colombia and France, followed by U.K. and Germany. - Within Central America, Guatemala increased its degree of competition with El Salvador. - Between 2000 and 2014 Brazil was replaced by Colombia as the most dominant competitor (by VBI). - Count-based competition: - High for most top competitors; U.S. and Brazil ranked as top competitors by count. - Between 2000 and 2014 most top competitors exhibit a decline in count-based competition, except Italy. ### Share in world textiles exports — tabulated values (as presented) - Table 1. Share in World Exports of Textiles (In percent) — entries as in source: - 2000 0.5 5.4 18.6 3.0 4.5 6.0 2.7 4.5 1.4 2.2 0.6 0.8 0.3 - 2005 0.4 4.3 26.8 3.7 2.6 6.0 4.0 4.3 1.8 1.9 1.1 0.7 0.6 - 2010 0.3 3.5 34.6 5.0 2.1 4.7 3.9 3.9 3.1 2.1 2.4 0.7 0.7 - 2014 0.3 2.3 33.3 5.0 2.0 4.6 4.2 3.6 4.4 2.2 3.8 0.8 1.3 - (Table formatting preserved as presented in source.) ### Policy priorities and diagnostics - Key observed trend (2000→2014): competition from large exporters of 2000s (U.S., Germany, Japan) declined; China largely replaced them by 2014. - China’s world market share in textiles peaked in 2010 and saw a small decline thereafter; China’s share increased most rapidly during 2000–10. - Between 2010 and 2014: - Most developed economies’ shares declined. - Shares of many developing economies increased; El Salvador did not increase its share (market share declined 2000–2010 and remained stagnant 2010–2014). - Countries filling space left by China include Vietnam and Cambodia. - Vietnam’s competition increased in the U.S. market; Cambodia’s competition increased in the CAPDR market (mostly due to increase in CBI). - Two issues to address for export growth: - (a) improving competitiveness of exporting firms; - (b) diversifying the export base. - Determinants of firm competitiveness (listed in source): - (i) access to a well-trained workforce (exports tend to be of higher quality); - (ii) access to finance (exporting involves high fixed/sunk costs); - (iii) ability to invest in research, technology, and innovation (to improve competitiveness and product quality over time); - (iv) access to strong infrastructure and an entrepreneur-enabling regulatory environment. - El Salvador’s institutional and business environment diagnostics: - Ranks poorly relative to other Central and Latin American countries in starting a business, dealing with construction permits, access to electricity, and protection of investors (World Bank Doing Business). - Global Competitiveness rankings show El Salvador lags in almost every dimension compared to CAPDR and LA-5; gaps widest in institutions, human capital, innovation, and labor market efficiency. - El Salvador’s Doing Business Index ranking jumped 22 notches in 2018, from 95 to 73. - El Salvador ranks 140 in terms of regulations on starting a business. - Very low new business entry density compared to LA-5 and other CAPDR economies. - Policy priorities recommended (paragraph 20): - Improve access to education and training to upgrade skills and keep up with technological change. - Improve infrastructure and the institutional environment to support private investment and reduce internal trade costs. - Improve financial inclusion. - Promote greater gender equality to support entrepreneurial activity and activity in more sectors. - Invest in research, technology, and innovation to improve product quality. - Note on informality: - High level of informality in El Salvador makes improving competitiveness and diversification more challenging. ### Industrial policy stance - Conventional view: avoid industrial policy because of high informational demand, poor implementation, and potential capture due to rent-seeking. - Recent literature: conditional support for industrial policy under conditions: - Policies should avoid "picking winners"; implement at sector level. - Support sectors with large positive spillovers only insofar as they are competitive. - Support should be spread across many firms and not be detrimental to entry of new firms. ### Annex — products, competitors, and export composition - Top exports (HS-4) and share in SLV exports, percent (excerpt from Table AI.1): - Knit T-Shirts 13.0 - Knit Sweaters 6.5 - Knit Socks Hosiery 4.4 - Knit Men’s Undergarments 3.6 - Light Rubberized Knitted Fabric 2.0 - Women’s Undergarments 1.6 - Knit Men’s Suits 1.6 - Knit Women’s Suits 1.5 - Non-Knit Women’s Suits 1.0 - Non-Knit Men’s Shirts 1.0 - Raw Sugar 3.3 - Flavored Water 2.0 - Baked Goods 1.6 - Processed Fish 1.4 - Electrical Capacitors 4.9 - Insulated Wire 1.3 - Plastic Lids 3.2 - Toilet Paper 2.2 - Paper Containers 1.3 - Coffee 2.6 - Packaged Medicaments 2.1 - Refined Petroleum 1.6 - Competitor inclusion criteria (Table AI.2): - Competitor if share in world exports of a given product > 1% AND among the top 15 exporters of that product — list includes Austria, Bangladesh, Belgium-Luxembourg, Cambodia, Canada, China, Hong Kong SAR, Croatia, Czech Republic, France, Germany, Honduras, India, Indonesia, Italy, Japan, Malaysia, Mexico, Morocco, Netherlands, Pakistan, Philippines, Poland, Republic of Korea, Spain, Sri Lanka, Switzerland, Thailand, Turkey, United Kingdom, USA, Vietnam. - Competitor if share in world exports of a given product 0.4% to 1% AND EM or LIC — list includes Brazil, Colombia, Costa Rica, Ecuador, Guatemala, Hungary, Nicaragua, Peru, Portugal, Serbia, South African Customs Union, United Arab Emirates, Other Central American Countries, Belize, Panama. ### Impact of emigration — methodology and key data points - Emigration entails costs (reduced labor supply, especially skilled workers) and benefits (remittances). - Analysis sample: 31 countries in Central America, South America and the Caribbean. - Emigration rate definition: hi_e,t = Mi_e,t / (Ri_e,t + Mi_e,t) (as presented in source formulation). - Theoretical framework: based on Mishra (2007a) — labor demand and supply model; emigration causes emigration loss via wage increases and factor-income distribution effects. - Emigration loss expressions (as presented in source): - Emigration loss: L = (e/2) * m^2 * SL (equation numbering preserved in source). - High-skilled emigration loss: (expression presented in source with variables S_S, e_S, m_S). - Augmented emigration loss: expression includes S_U and γ and captures negative externalities on marginal product of remaining skilled/unskilled labor. - Education expenditure loss for high-skilled emigrants: - Annual expenditure on education of high-skilled migrants (as a ratio of GDP) = c_s * M_s / Y (expression presented as in source). - c_s measured by dividing government expenditure (percent of GDP) on primary, secondary, and tertiary education by enrollment in public institutions at respective levels (UNESCO UIS data referenced). - Practical approximation: education expenditures averaged over 1990–2005, 1995–2005, and 2000–2010 to approximate costs for migrants educated prior to emigration; public education expenditure data available for only 19 countries under this method. - Data and assumptions highlights: - For some countries (notably Caribbean) overall emigration rate is around 40 percent and emigration rate of high-skilled labor around 80 percent (averages cited). - El Salvador exhibits higher emigration rates compared to most Central American neighbors. - Personal remittances (percent GDP) received by El Salvador are among the highest in the region. - Graphical summaries (as reported): - Skilled Labor Emigration Rate (Average of 2000, 2005 and 2010) — ranked series includes Guyana down to Chile. - Personal Remittances, % GDP (Average over 2000 to 2010) — Haiti and El Salvador among highest. - Overall Emigration Rate (Average of 2000, 2005 and 2010) — charted across sample (El Salvador included). *Source: IMF staff report content provided in the supplied document.* ### 12. Values of parameters to compute loss. The variation in emigration loss across countries ### 12. Values of parameters to compute loss. The variation in emigration loss across countries ### Parameter values used to compute emigration loss - Elasticity of factor price of labor e: 0.3 or 0.4 (based on Mishra (2007b)). - Elasticity of factor price of skilled labor eS: 0.3 or 0.4 (assumed same as e). - Elasticity of marginal product of skilled and unskilled labor γ: 0.05 or 0.1 (based on Mishra (2007a)). - Share of labor in national income SL: 0.69 (based on Guerriero (2012)). - Share of skilled labor in national income SS: 0.39 (assuming skilled labor constitutes the top 20 percent earners; income share of top 20 percent earners is 0.56, an average across 12 LAC countries taken from WDI). ### C. Results — Emigration loss (summary findings) - Simple emigration loss (equation (1)): - Dominica and Grenada show the highest losses. - For 29 out of 31 countries, remittances exceed the simple emigration loss. - Selected country values (Loss, remittances, all figures in percent of GDP): - Antigua and Barbuda: Loss (e=0.3) 1.07; Loss (e=0.4) 1.43; Remittances (Average 2000-10) 1.89 - Dominica: Loss (e=0.3) 2.32; Loss (e=0.4) 3.09; Remittances 4.82 - Grenada: Loss (e=0.3) 2.24; Loss (e=0.4) 2.99; Remittances 4.43 - Guyana: Loss (e=0.3) 2.19; Loss (e=0.4) 2.92; Remittances 13.25 - El Salvador: Loss (e=0.3) 0.55; Loss (e=0.4) 0.73; Remittances 16.04 - Average (sample): Loss (e=0.3) 0.59; Loss (e=0.4) 0.79; Remittances 5.16 - High-skilled emigration loss (equation (2)): - Emigration loss is higher when only high-skilled emigration is considered because skilled emigration rates are higher. - Guyana and Barbados show the highest high-skilled losses. - Remittances exceed the high-skilled loss: - In 26 countries when eS = 0.3. - In 24 countries when eS = 0.4. - For El Salvador: - High-skilled loss (eS=0.3) 0.68; (eS=0.4) 0.90; Remittances 16.04. - The loss is higher than for Nicaragua, Dominican Republic and Honduras but lower than for Guatemala. - Selected country values (Loss, remittances, all figures in percent of GDP): - Antigua and Barbuda: Loss (eS=0.3) 4.62; Loss (eS=0.4) 6.16; Remittances 1.89 - Barbados: Loss (eS=0.3) 5.10; Loss (eS=0.4) 6.80; Remittances 3.04 - Guyana: Loss (eS=0.3) 5.78; Loss (eS=0.4) 7.71; Remittances 13.25 - Average (sample): Loss (eS=0.3) 1.59; Loss (eS=0.4) 2.12; Remittances 5.16 - Augmented emigration loss including "external effects" (equation (3)): - For only 16 to 17 countries (depending on γ = 0.05 or 0.1), remittances exceed the augmented emigration loss. - Drivers of net gain or loss vary: - Antigua & Barbuda, Barbados, Dominica, and Trinidad & Tobago: loss driven by very high rates (80 percent or more) of emigration of skilled labor and low remittances. - Guyana and Haiti: comparable emigration rates but remittances are substantially large (13 to 21 percent of GDP) and act as an offset. - El Salvador, Guatemala and Honduras: similar augmented loss (lower than Caribbean countries); remittances (8 to 16 percent of GDP) act as an offset. - Colombia, Costa Rica, Mexico, Paraguay and Peru: lower emigration rates; remittances are only marginally higher than the augmented loss. - Augmented emigration loss plus public education expenditure (sunk cost for high-skilled migrants): - For 19 countries with education expenditure data, remittances exceed the sum of augmented loss and education expenditure in 11 to 12 countries (depending on γ). - Countries with high tertiary education expenditure (more than 5 percent of GDP) include Antigua and Barbuda, Barbados, Guyana and Jamaica. - Remittances offset education spending for Guyana and Jamaica but not for Antigua and Barbuda and Barbados. - For 12 countries with missing education data, remittances exceed augmented loss by 6 percent of GDP or more in Dominican Republic, Haiti and Honduras — likely not offset even if education expenditure were included. - For El Salvador: - Education Expenditure* (Average 2000-2010) 0.52; Loss (augmented, γ unspecified in table) 2.37; Loss (augmented, alternative γ) 4.01; Remittances 16.04 - Remittances outweigh the sum of emigration loss and education expenditure by the widest margin in the sample. - Selected rows from Table 3 (values are in percent of GDP): - Antigua and Barbuda: Education Expenditure 3.10; Loss (augmented) 8.65; Loss (augmented, alt) 11.43; Remittances 1.89 - Guyana: Education Expenditure 3.81; Loss (augmented) 10.22; Loss (augmented, alt) 13.00; Remittances 13.25 - Haiti: Education Expenditure -7.74; Loss (augmented) 10.47; Remittances 21.38 - Average (sample, where available): Education Expenditure 1.01; Loss (augmented) 3.50; Loss (augmented, alt) 5.03; Remittances 5.16 ### D. Conclusion and caveats - Conclusion: - Remittances outweigh the sum of augmented emigration loss and education expenditure for nearly half of the 31 countries in the sample. - Most of these countries are from Central America; countries for which remittances do not outweigh the sum are mostly Caribbean countries. - Caveats (limitations explicitly mentioned): - Other costs and benefits of emigration are not included in the framework (e.g., impact on trade, investment and marginal productivity of capital). - Hard-to-measure factors such as demography, civil unrest, and natural disasters are not captured. - The value of elasticity of factor price of skilled labor may differ from that of overall labor; country-level elasticities would better capture country characteristics but require detailed survey data. - Methodological/data notes: - Emigration loss in Table 1 is calculated using equation (1). - High-skilled emigration loss in Table 2 is calculated using equation (2). - Augmented emigration loss in Table 3 is calculated using equation (3). - Education expenditure per high-skilled migrants (% GDP) is the average of tertiary education expenditure, % GDP/ enrollment in tertiary public institutions (2000-2010), and averages for primary and secondary years noted in the source. - Bahamas excluded where remittances data is missing. *Sources: World Bank and author's calculations.* ### 9. An econometric assessment shows that the recent credit growth is not excessive and is ### 9. An econometric assessment shows that the recent credit growth is not excessive and is ### Econometric approach and definitions - Long-term trend of the credit-to-GDP ratio is estimated using an HP filter. - The “deviation” is defined as the difference between actual credit-to-GDP and its HP-filter trend. - A credit boom is identified when the deviation exceeds a threshold set equal to 1.5 standard errors of the deviation level, which is intended to cover 90 percent of episodes of credit cycle expansion. - The analysis is also conducted using credit-to-potential GDP (long-term trend of credit-to-potential GDP) as an alternative specification. ### Main findings - Under both specifications (credit-to-GDP and credit-to-potential GDP), the deviations from trend are well below the threshold. - Conclusion: recent credit growth is moderate and reflects financial deepening in line with macroeconomic fundamentals. ### Key statistics preserved from the source - Estimated threshold (El Salvador) = 1.43 percent. - Basel III micro-prudential rule threshold = 2 percent. ### Figure note - Figure 2 presents time series charts for: - Credit to Private Sector (Percent of GDP): Actual and Trend (2002Q1–2017Q1). - Credit to Private Sector (Deviation from trend): quarterly deviations (2002Q1–2017Q1). - Credit to Private Sector (Percent of Potential GDP): Actual and Trend (2002Q1–2017Q1). - Credit to Private Sector (deviation from trend): quarterly deviations (2002Q1–2017Q1). - Sources for figures: Central Bank and Fund staff calculations. ### Cautions and methodological limitations - The HP filter should be used with caution when time series are not sufficiently long. - Seidler and Gersl (2012) note the estimated trend could depend on the starting point of the time series. - Drehmann and Tsatsaronis (2014) show the dependence problem worsens as the length of available time series shrinks. - Borio and Lowe (2002) suggest the methodology is not advisable for time series of length shorter than 10 years. - The HP-filter methodology could fail to account for positive credit expansion driven by financial deepening. - A sustained period of high growth in credit-to-GDP can translate into a faster trend growth estimate, which could undermine the adoption of micro-prudential measures such as countercyclical capital buffers. *Source: Central Bank and Fund staff calculations as presented in the IMF Selected Issues chapter excerpt.* ### 13. Barriers to business entry hamper competition, trade and investment. In the airline ### 13. Barriers to business entry hamper competition, trade and investment. ### Barriers to entry — Findings - Airline sector: Incumbent airlines can prevent new entry at the “public consultation” phase in the approval process for air traffic rights. - Electricity sector: Conditions for third party access to the transmission grid are not regulated. - Professional services: Lawyers are not allowed to offer their services through public and private limited liability companies. - Food sector: Registration of food companies is bureaucratic and lengthy; there is still no digital registration of products to shorten and streamline the process. - Road transport sector: Restrictions to foreign firm participation (e.g. only vehicles registered by companies with at least 51 percent of local capital are allowed to operate). - Rice importers: Entry of new rice importers is limited by predetermined quotas; a new importer needs to have a record of four years before obtaining higher quotas. - Industrial associations: Can limit entry, fix prices or quantities produced, and set quotas for imports in markets such as rice, sugar, maize, beans, and coffee; frequently act as exclusive importers at lower cost and sell at high prices in the domestic market; decide on access to import quotas for new members. ### Judicial system — Findings - Anti-Corruption Unit within the Prosecutor General (“Fiscalia”) lacks personnel capacity and technical expertise and relies mostly on other institutions and areas of the Prosecutor General office. - Prosecutor General distributes corruption cases to the Anti-Corruption Unit in a discretionary way without taking into account the Unit’s mandate. - Prosecution and sanctioning of many cases are being delayed indefinitely. - Backlog of pending antitrust cases revealed by the Competition Authority; the Prosecutor General delays review or restarts review from scratch, indicating inefficiencies and redundancies. - Court of Accounts (“Corte de Cuentas”) lacks autonomy and independence, has no effective control over public funds, and has suffered internal corruption (several employees investigated for letting cases expire and for unnecessary spending on goods and services). ### Policy recommendations (permits, regulatory reform, and competition) - Permits and registration: - Permits and registration processes should follow clear procedures and responses be expedited. - Ministry of Health, Ministry of Agriculture, Ministry of the Environment, and water and sewage authorities need to work on best practices to give legal certainty in granting permits. - Pass laws for administrative procedures to provide a foundation for regulatory simplification and simplification in customs procedures. - Regulatory Agency (OMR) should review and prioritize simplification of procedures without undue delay. - Note: One step forward is the likely approval of legislation on administrative procedures which is in discussion in Congress. - Customs union: - Move forward with reforms required for the creation of the customs union with Guatemala and Honduras. - Harmonize legal framework and VAT refunds systems and have a plan to reduce import duties to allow more competition in sugar and agricultural products. - Regulatory governance: - Avoid government meddling in tripartite representation at key Commissions and boards of regulatory agencies. - Strengthen institutionality of election of Board members of SIGET. - Auctions should not be manipulated and concessions should be granted to the best offer. - Minimum wage and competitiveness: - Given dollarization, no further increases in the minimum wages should be passed. - More predictability in decisions on minimum wages is needed from the Ministry of Labor. - Representation of the private sector at the National Council on the Minimum Wage should be improved. - Strengthening judicial and anti-corruption institutions: - Dedicate more resources to the Prosecutor’s General Office. - Achieve better coordination between agencies fighting corruption (Prosecutor, Court of Accounts, and Ethics Tribunal). - Give civil society opportunity to be consulted on corruption cases. - Ensure the July 2018 election of four members of the Constitutional Court is transparent. - Strengthen the Anti-Corruption Unit with higher personnel capacity, technical expertise, and adequate funds to ensure proper functioning and guarantee stability. - Introduce a mechanism to prioritize review of corruption cases so prosecution and sanctioning are swift. - Court of Accounts reforms: - i) Election of independent judges by a qualified majority (now elected with a simple majority). - ii) Revise regulatory framework to create 2 separate entities: one with audit functions and another with jurisdictional functions; introduce job requirements to attest for competence, honesty and independence. - iii) Produce and publish comprehensive annual reports. - Antitrust enforcement: - Prosecutor General should resolve the backlog of pending antitrust court cases. - If the Competition Authority has found evidence of anti-competitive behavior, the Prosecutor should follow and internalize that analysis rather than starting review from scratch. - Ruling and sanctioning on strong evidence should not take too long. ### State of play — Human capital and migration (findings and statistics) - Education and skills: - Mean years of schooling: 7 years in El Salvador versus 8.4 years in Latin America. - Educational attainment skewed towards primary and lower secondary education for both women and men. - Post-secondary education attainment is higher for women than for men. - El Salvador ranked 49 out of 53 countries in the 2007 TIMS for math/science tests. - The bulk of spending in education is at the primary level; sizeable coverage gap in secondary and tertiary education. - Pupil-teacher ratios in secondary education are much higher compared to regional peers. - Innovation: - El Salvador ranks poorly in spending on R&D, tertiary enrollment rates, number of patent applications, FDI inflows, ease of protecting investors, knowledge-intensive employment, and creative services exports (Global Innovation Index). - Migration: - Approximately 1.4 million immigrants from El Salvador resided in the United States in 2015. - Age-sex pyramid of foreign-born from El Salvador: diamond shape with largest numbers in ages 20 to 54; relatively small numbers under age 20 and over age 54. - Foreign-born population from El Salvador has more men overall than women. - Anecdotal evidence indicates migrants are highly skilled. - Labor force participation and gender gap: - Women labor participation rates are particularly low; El Salvador’s gender gap is around 30 percent (more than double the U.S. gender gap). - Nearly four-fifths of men in the region participate in the labor force, versus about two-thirds in advanced economies. - Key labor market statistics: - Unemployment: 7 percent in 2016. - Informal employment share: 68.2 percent (ILO, 2014). - Informality (IDB System): 72 percent. - Informal employment in urban areas (2016, household surveys): 42.6 percent (2006: 48.7 percent). - Informality is higher for women than for men; more prevalent in commerce followed by manufacturing. - 64 percent of poor are part of the informal sector. - Share of employees not covered by social security: as high as 60 percent. - 44 percent of the economically active population is not covered by the minimum wage legislation. - Corporate tax rate: 35 percent (noted as appearing too high given low investment and high informality). - Private sector employees earning below minimum wage: increases noted since 2008. - Since 2008, unremunerated family workers and domestic workers have grown the most, as well as workers in large and micro enterprises. - Based on cited sources, formal employment has been stagnant since 2000 while informality remains high. - Education budget and wages: - The wage bill represents 68 percent of the education budget. - The wage bill is 82 percent higher in 2014 compared to 2007. ### Policy recommendations — Human capital, female participation, and informality - Human capital: - Focus additional future education spending on higher levels of education rather than primary levels to account for gradual ageing. - Reduce number of primary teachers in favor of secondary education teachers. - Define teaching standards to guide teacher development, articulate continuous training with a strategy for teacher professionalization. - Establish more demanding criteria to select and retain the most talented teachers. - Review salary structure and establish incentives for professional development. - Enhance R&D/technological diffusion by strengthening institutions, human capital and research, and achieving higher business and market sophistication and competition in product and labor markets. - Female labor force participation and migration: - Policies to foster higher women labor force participation and to absorb returning migrants productively would raise employment growth. - Illustrative calculation: if Latin American countries raised female labor force participation to the average of the Nordic countries (which is 61 percent), their GDP per capita could be up to 10 percent higher, depending on the country and existing level of female participation. - Suggested incentives: free or subsidized childcare programs; increased children’s hours in school to provide additional time for mothers to work. - Implement the “El Salvador Seguro” plan and supplement it with adequate resources and grant funding to reduce crime and reverse migration. - Reducing informality: - Increase the cost of avoiding regulations, including through penalties for tax evasion within formal firms and penalties for informal activity. - Strengthen enforcement to detect informal workers/firms. - Create incentives for firms to become formal by reducing tax, financial, and regulatory constraints, enhancing access to credit at a reasonable cost, and reducing costs of registration (noting permits can cost more than bribes). - Consider reducing the corporate tax rate of 35 percent significantly while limiting incentives and loopholes to broaden the tax base. - Revamp benefits of belonging to the formal workforce and create more flexible social security schemes for segments of the workforce (with lower contribution rates but also lower benefits). - Enhance coverage and quality of education to reduce mismatches between university-level graduates and market needs. *EL SALVADOR — INTERNATIONAL MONETARY FUND (excerpts).* --- _Source: https://www.imf.org/-/media/files/publications/cr/2018/cr18152-elsalvadorsi.pdf_