## cr18216

## Source details

**Canonical URL:** [cr18216](https://www.imf.org/-/media/files/publications/cr/2018/cr18216.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2018/cr18216.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2018/cr18216.pdf.json)

---

### Major methodologies for estimating potential output
- Hodrick-Prescott (HP) filter applied to GDP series with staff GDP projections for 2018Q1–2019Q4 to reduce end-point bias; standard quarterly λ = 1,600 is used.
- Multivariate filter (MVF) estimating potential GDP from relationships among output, core inflation and unemployment using Bayesian techniques; includes Phillip’s curve and Okun’s law. Calibration method: Vietnam — Regularized Maximum Likelihood.
- Multivariate filter augmented by financial frictions (MVF-FIN) focusing on sustainable output and incorporating the credit gap (BIS methodology) and demeaned asset market price growth; variance ratio constraint (ρ) mirrors HP filter characteristics.
- Human-capital augmented production function: Yt = At Kt^αt (Ht Lt)^(1-αt) with TFP, capital stock, labor force, human capital (years of schooling and return to education) explicitly modeled. Production function extrapolates years of schooling for 2011–2023 using a linear trend due to last reported schooling data as of 2010.

### Key numeric estimates and summary assessment
- HP filter:
  - Potential growth estimated at 6.5 percent in 2017.
  - Output gap at 0.75 percent (text Figure).
- Multivariate filter (MVF):
  - Potential growth estimated at 6.4 percent in 2017.
  - Output gap at -0.1 percent (MVF approach).
- MVF with financial frictions (MVF-FIN):
  - Sustainable output growth estimated at 6.4 percent in 2017.
- Production function approach:
  - Potential GDP growth estimated at 6.5 percent in 2017.
- Overall assessment and synthesis:
  - On balance, potential growth assessed at 6.5 percent in 2017, higher than previous staff estimates of 6.2.
  - Output gap assessed at 0.4 percent in 2017 (concluding synthesis).
- Medium-term projections:
  - Potential growth could average 6.8 percent over the medium-term, reaching 7.3 percent by 2023 (text Table and Figure).
- Total Factor Productivity (TFP) statistics and assumptions:
  - TFP growth averaged 2.1 percent until the Asian Financial Crisis.
  - Between 2000−07, TFP grew a little under 1 percent on average.
  - Since 2013, TFP has recovered and grew at an average of 1.7 percent per year.
  - Assumed TFP growth for projections: average 1.8 percent (similar to the 2014–17 average).
  - Regression analysis supports a wide range of average TFP growth between 1.7 to 2.5 percent.
- Investment and capital:
  - Investment is projected to grow at a rate of 10 percent in average (assumption for projections).
  - In 2017, total investment grew by 10 percent.
  - Physical capital stock growth accelerated to about 8 percent in 2017.
- FDI and trade:
  - In 2017, FDI inflows reached US$17.5 billion, up 40 percent compared to 2014.
  - In 2017, the export-oriented sector was responsible for more than two-thirds of Vietnamese exports and a third of the ASEAN’s tech exports.
- Labor and human capital:
  - Vietnam median age: 26.
  - Participation rates, male and female, have been high at around 80 percent.
  - Human capital growth assumption: 1.02 percent (for projections).
- Other key figures preserved exactly:
  - Potential growth: 6.5 percent
  - Output gap expected to close in 2019
  - HP filter quarterly λ used for GDP: 1,600
  - HP filter for credit-to-GDP trend: lambda = 400,000
  - Capital stock depreciation ratio projected constant at 4.5 percent
  - Return to education assumed constant at 7%
  - Investment projected to grow by 10 percent in 2018–23
  - Government credit growth target for 2018: 17 percent
  - Baseline nominal GDP growth projection for 2018 and 2019: around 10.5 percent
  - Credit growth to close credit gap to zero in 2018: around 14 percent
  - Alternative credit growth scenario considered: 20 percent
  - Other Calibration: Gss 2.2; Uss 6.5

### Drivers of recent and prospective potential growth
- Structural and cyclical drivers identified:
  - A very productive export-oriented sector financed by FDI has boomed, contributing more than two-thirds of exports in 2017 and significant ASEAN tech exports.
  - Urbanization and a shift of employment from agriculture to industry and services boosting productivity.
  - Equitization and reforms in the large SOE sector restricting SOE operations to core areas and increasing private sector participation.
  - Regression analysis links TFP growth to increasing FDI inflows, declining share of employment in agriculture, declining credit to SOEs and increasing domestic private investment.
  - Physical capital accumulation: recent high-quality FDI capital contributing to capital stock growth.
  - Human capital: quantity of schooling incorporated; recent literature emphasizes education “quality” (not directly measured here), with Vietnam noted for strong 2015 PISA scores.

### Model limitations, data caveats and interpretation issues
- HP filter: purely statistical and subject to end-point bias; staff projections appended to mitigate bias.
- MVF: relies on economic relationships that appear weak in Vietnam (Phillip’s curve/link between output gap and inflation is weak; official unemployment series shows little variation and does not respond as expected to aggregate demand shocks).
- MVF-FIN: may underestimate the size of financial imbalances (credit gap and asset price measures affected by HP filtering end-point bias, GDP and credit data quality issues, and omission of structural transformation effects), leading possibly to an overestimation of potential output.
- Production function: limited by availability of Vietnam-specific elasticities and by using years of schooling (quantity) rather than direct measures of education quality; TFP measure may capture other factors such as capacity utilization and average hours worked.
- Data availability/quality notes:
  - Lack of official real estate data affects MVF-FIN (asset market price growth excludes real estate).
  - Credit gap is calculated using a HP filter and therefore is subject to end-point bias.
  - Production function model extrapolates years of schooling for 2011–2023 using a linear trend due to last reported schooling data as of 2010.

### Projection assumptions and scenario inputs (selected)
- Investment projected to grow at 10 percent in average.
- Labor force growth assumed to decline following population growth trends.
- Human capital growth assumed to remain constant at 1.02 percent.
- TFP assumed to grow by an average 1.8 percent for baseline projections; regressions support a range between 1.7 and 2.5 percent.
- Capacity utilization assumed unchanged; actual changes captured in TFP.
- Depreciation ratio projected constant at 4.5 percent.
- Return to education assumed constant at 7% over the period.

### Vietnam: credit growth, asset markets, and early warning indicators
- Credit-to-GDP developments:
  - Credit-to-GDP ratios rose from around 30 percent in the early 2000s to around 130 percent in 2017.
  - HP filter trend (lambda = 400,000) shows strong financial deepening; sharp increases after 2009–2010, correction around 2011 real estate bust, and strong increase since 2012.
- Early warning indicators (EWI) results:
  - Credit gap was positive but less than 10 percent in 2017, versus more than 20 percent during the 2011 crisis.
  - Credit growth gap reached almost the 2011 crisis level in mid-2016 but declined in 2017 due to higher nominal GDP growth and lower credit growth.
- Scenarios and policy implications:
  - Government’s credit growth target for 2018 is 17 percent.
  - Baseline scenario: nominal GDP growth around 10.5 percent in 2018 and 2019 → credit gap and credit growth gap increase but remain around 6–7 percent and 7–9 percentage points in 2019.
  - If credit growth accelerates to 20 percent in 2018 and 2019 → credit gap and credit growth gap would swiftly rise to approach previous crisis levels, posing financial stability risks.
  - To close the credit gap to zero, credit growth should remain around 14 percent in 2018.
  - Long-run recommendation: phase out credit growth targets and adopt a monetary policy framework based on inflation targeting and open market operations so credit growth is market-determined.
- Benchmarking credit-to-GDP:
  - Benchmark credit-to-GDP (Cbench) assumed function of per capita GDP; deviations from fitted values interpreted as excessive credit supply.
  - Benchmarking indicates recent credit-to-GDP increase may have been too rapid; ratios are outside the 1.5-SD band after recovery from 2011 crisis.
  - Historical evidence: large positive deviations from benchmark often led to banking crises and sharp credit corrections, though substantial deviations can persist without correction in some cases.

### Stock market performance, real estate valuation, and policy implications
- Stock market performance (year-on-year, US dollar basis):
  - more than 15 percent in 2016
  - more than 40 percent in 2017
- Market capitalization:
  - increased by more than 70 percent in 2017
  - rose to around 70 percent of GDP at the end of 2017 from around 40 percent in 2016
  - HOSE’s share of total market capitalization in Vietnam was around 70 percent at end-2017
  - As of January 2018, share of Top 5 and 10 firms in HOSE and NHX are 36 percent and 52 percent; among these top 5 and 10 firms, 4 and 7 firms, respectively are SOEs
- Valuation and risks:
  - Rapid increases in PER, PBR, and dividend yield in 2017 signal elevated investor expectations; a correction is possible if expectations are not realized.
  - Valuation levels are not substantially higher than those in other countries, but the magnitude and speed of 2017 changes are notable.
- Real estate misalignment assessment methods:
  - Price-to-Rent ratios and regression-based affordability model using per capita GDP and cross-country panel for regional comparators.
- Real estate findings for Vietnam:
  - Since 2015, HCMC residential prices continued to increase, but price-to-rent ratios have not significantly increased and are converging to the 2015 level due to rent increases.
  - Regression-based assessment indicates Vietnam’s property price appears not to substantially deviate from the estimated trend in recent years, but the model has limited predictive power for Vietnam due to data quality issues.
  - Drivers supporting housing demand: strong income growth of a growing middle class, steady urbanization, and limited government mortgage policies via state-owned banks.
- Policy recommendations:
  - The central bank’s credit growth target should be decreased to avoid excessive risk taking by banks.
  - Continued SOE and other growth-supporting reforms would help better align stock market valuations with fundamentals.
  - The State Bank of Vietnam (SBV) should develop a macroprudential policy framework including loan-to-value (LTV) and debt service-to-income (DSTI) requirements to address potential real estate exuberance.
  - More granular data (real estate, bank and corporate balance sheets) are needed to assess spillovers and risks.

### Fiscal rules, debt-anchor calibration, and operational expenditure rules
- Current statutory limit:
  - Vietnam follows a debt rule with a statutory limit of 65 percent of GDP on public and publicly guaranteed debt (PPG).
- Fiscal context:
  - Estimated deficits: 4.5 percent of GDP in 2017 and 4.6 percent in 2018.
  - PPG debt increased rapidly between 2011 and 2016; average deficits rose from 2 percent (2000–2011) to 6.2 percent (2012–2016).
- Demographic fiscal pressures:
  - Age-related spending needs estimated to increase by 8 percent of GDP by 2050, increasing pension outlays by 6.7 percent of GDP.
  - Current pension scheme projected to run deficits as of 2030 absent reforms.
- Debt-anchor calibration steps and results:
  - Step 1: set maximum debt limit at 80 percent of GDP (70 percent MAC DSA limit plus 10 percent buffer for public guarantees); subtracting estimated 8 percent of GDP age-related spending reduces this to 72 percent of GDP.
  - Step 2: perform stochastic simulations identifying joint distribution of GDP growth, interest rates, and exchange rate based on past means and variances.
  - Step 3: simulations starting in 2017 suggest a debt anchor of 66.5 percent could be sufficient to keep debt below the maximum debt limit of 80 percent with 90 percent certainty.
  - Step 4: account for fiscal risks and long-term aging costs — a lower debt anchor may be warranted.
- Scenario analysis — Debt Anchor (% of GDP):
  - 1) 5 percent realization of contingent liabilities: 58.8
  - 2) Real GDP growth slowdown to 4 percent of GDP: 52.3
  - 3) Scenario 2 + 2.5 percent realization of contingent liabilities: 50.1
  - 4) Scenario 2 + real depreciation of 4 percent per year: 49.9
  - 5) Age related spending (maximum debt limit of 72 percent of): 53.8
- Recommendation on debt target:
  - Presence of fiscal risks and contingent liabilities requires a lower debt limit than the current statutory ceiling.
  - Recommendation: authorities should aim at a target of 55 percent of GDP to account for fiscal risks and aging costs that could materialize in the medium to long term; the current 65 percent of GDP ceiling should be considered an upper bound.

### Expenditure rules (ERs): cross-country context, design and calibration for Vietnam
- Cross-country experience:
  - As of 2015: 46 countries had ERs, 14 of which were in emerging markets (EMs).
  - ER characteristics include limits on total, primary, or current spending; nominal or real ceilings; time horizon typically three to five years.
- Choosing an ER for Vietnam:
  - Recommendation: an ER specified in terms of nominal growth rates would be appropriate for Vietnam.
  - Design considerations:
    - Expenditure growth could be calibrated on nominal GDP growth initially, but should be relatively independent of the GDP series during implementation.
    - ERs should be re-calibrated every 3 to 5 years.
    - ER should be formulated on overall public expenditure (general government level) and defined in nominal terms for transparency and ease of monitoring and enforcement.
- Calibration and illustrative scenarios:
  - A 10 to 11 percent growth rate ceiling on a nominal expenditure rule would be consistent with a medium-term debt target of 55 percent of GDP.
  - Methodology notes: average expenditure growth for the first 5 years (2018–2023) under a given fiscal balance, assuming a stable revenue ratio at 23 percent of GDP and initial debt level at end-2017 of 58.5 percent of GDP; illustrative scenario assumes long-term nominal interest rate of 7 percent and long-term nominal GDP growth rate of 10 percent.
  - Table 3 key rows (preserved exactly):
    - Illustrative Scenario — 10% — 4.4% — 10.7% — 54%
    - Consolidation Scenario 1/ — 10.5% — Declining to 3 % by 2023 — 10.3% — 52.8%
    - Stronger growth — 12% — 5.1% — 11.2% — 54%
    - Weaker growth — 7% — 3.0% — 9.6% — 54%
  - Notes: illustrative scenario close to consolidation scenario suggested in the 2018 Article IV; consolidation scenario incorporates multiplier, tax reform in 2019, PIM improvements, and structural reforms with net real GDP growth benefit.
- Sensitivity and prudence:
  - Sensitivity to nominal GDP growth implies deficit targets and expenditure growth rates can be relaxed or tightened depending on growth performance.
  - Given growth slowdown risk, increasing aging costs and contingent liabilities, a prudent fiscal rule framework should set an expenditure growth ceiling of 10–11 percent per year (tantamount to a limit to the overall balance of 3–4 percent of GDP).

### Calibration table (Prior -> Posterior) — model parameters
- lambda 0.250 -> 0.498
- beta 0.250 -> 0.252
- phi 0.600 -> 0.618
- theta 0.100 -> 0.161
- std_RES_LGDP_BAR 0.500 -> 0.500
- std_RES_G 0.500 -> 0.500
- std_RES_Y 1.000 -> 0.999
- std_RES_PIE 1.000 -> 1.365
- std_RES_UNR_GAP 0.800 -> 0.798
- std_RES_UNR_BAR 0.300 -> 0.300
- std_RES_G_UNR_BAR 0.300 -> 0.298
- tau1 0.100 -> 0.094
- tau2 0.100 -> 0.136
- tau3 0.100 -> 0.101
- tau4 0.100 -> 0.101
- Other Calibration: Gss 2.2; Uss 6.5

### Conclusions and policy implications (synthesis)
- The four methodologies provide a range of estimates for Vietnam’s potential output in 2017, clustering around 6.4–6.5 percent.
- On balance, staff assess potential growth at 6.5 percent in 2017; output gap estimates vary by method (HP: 0.75 percent; MVF: -0.1 percent), consolidated assessment: 0.4 percent in 2017.
- Under baseline assumptions the model projects potential growth could average 6.8 percent over the medium-term and reach 7.3 percent by 2023, conditional on continued reforms, FDI inflows, and productivity improvements.
- Key policy recommendations:
  - Continue SOE reforms and privatization to support productivity and align equity valuations with fundamentals.
  - Reduce reliance on credit growth targets; move toward inflation-targeting and open market operations.
  - Develop macroprudential tools (LTV, DSTI) and strengthen data collection (real estate prices, sectoral credit, capacity utilization, informal sector labor) to improve monitoring and policy responses.
  - Adopt a prudent fiscal framework anchored by a debt objective and complemented by a nominal expenditure growth rule calibrated at around 10–11 percent to achieve a medium-term debt target near 55 percent of GDP.

*IMF staff analysis and estimates as presented in the provided chapter content.*

### References _________________________________________________________________________ 15

### References — cr18216

### Major methodologies for estimating potential output
- Hodrick-Prescott (HP) filter applied to GDP series with staff GDP projections for 2018Q1–2019Q4 to reduce end-point bias.
- Multivariate filter (MVF) estimating potential GDP from relationships among output, core inflation and unemployment using Bayesian techniques; includes Phillip’s curve and Okun’s law.
- Multivariate filter augmented by financial frictions (MVF-FIN) focusing on sustainable output and incorporating the credit gap (BIS methodology) and demeaned asset market price growth.
- Human-capital augmented production function: Yt = At Kt^αt (Ht Lt)^(1-αt) with TFP, capital stock, labor force, human capital (years of schooling and return to education) explicitly modeled. Appendix I contains model equations, calibration and assumptions.

### Key numeric estimates and statistics (preserved exactly)
- HP filter:
  - Potential growth estimated at 6.5 percent in 2017.
  - Output gap at 0.75 percent (text Figure).
- Multivariate filter (MVF):
  - Potential growth estimated at 6.4 percent in 2017.
  - Output gap at -0.1 percent (MVF approach).
- MVF with financial frictions (MVF-FIN):
  - Sustainable output growth estimated at 6.4 percent in 2017.
- Production function approach:
  - Potential GDP growth estimated at 6.5 percent in 2017.
- Overall assessment:
  - On balance, potential growth assessed at 6.5 percent in 2017, higher than previous staff estimates of 6.2.
  - Output gap assessed at 0.4 percent in 2017 (concluding synthesis).
- Projections and medium-term path:
  - Potential growth could average 6.8 percent over the medium-term, reaching 7.3 percent by 2023 (text Table and Figure).
- Total Factor Productivity (TFP):
  - TFP growth averaged 2.1 percent until the Asian Financial Crisis.
  - Between 2000−07, TFP grew a little under 1 percent on average (text: "a little under 1 percent").
  - Since 2013, TFP has recovered and grew at an average of 1.7 percent per year.
  - Assumed TFP growth for projections: average 1.8 percent (similar to the 2014–17 average).
  - Regression analysis supports a wide range of average TFP growth between 1.7 to 2.5 percent.
- Investment and capital:
  - Investment is projected to grow at a rate of 10 percent in average (assumption for projections).
  - In 2017, total investment grew by 10 percent.
  - Physical capital stock growth accelerated to about 8 percent in 2017.
- FDI and trade:
  - In 2017, FDI inflows reached US$17.5 billion, up 40 percent compared to 2014.
  - In 2017, the export-oriented sector was responsible for more than two-thirds of Vietnamese exports and a third of the ASEAN’s tech exports (qualitative, preserved from text).
- Labor and human capital:
  - Vietnam median age: 26.
  - Participation rates, male and female, have been high at around 80 percent.
  - Human capital growth assumption: 1.02 percent (for projections).
- Other model and data notes:
  - Credit gap is calculated using a HP filter and therefore is subject to end-point bias.
  - AVAILABILITY/QUALITY: Lack of official real estate data affects MVF-FIN (asset market price growth excludes real estate).
  - The production function model extrapolates years of schooling for 2011–2023 using a linear trend due to last reported schooling data as of 2010.

### Drivers of recent and prospective potential growth
- Structural and cyclical drivers identified in the text:
  - A very productive export-oriented sector financed by FDI has boomed, contributing more than two-thirds of exports in 2017 and significant ASEAN tech exports.
  - Urbanization and a shift of employment from agriculture to industry and services boosting productivity.
  - Equitization and reforms in the large SOE sector restricting SOE operations to core areas and increasing private sector participation.
  - Regression analysis links TFP growth to increasing FDI inflows, declining share of employment in agriculture, declining credit to SOEs and increasing domestic private investment.
  - Physical capital accumulation: recent high-quality FDI capital contributing to capital stock growth.
  - Human capital: quantity of schooling incorporated; recent literature emphasizes education “quality” (not directly measured here), with Vietnam noted for strong 2015 PISA scores.

### Model limitations, data caveats and interpretation issues
- HP filter: purely statistical and subject to end-point bias; staff projections appended to mitigate bias.
- MVF: relies on economic relationships that appear weak in Vietnam (Phillip’s curve/link between output gap and inflation is weak; official unemployment series shows little variation and does not respond as expected to aggregate demand shocks).
- MVF-FIN: may underestimate the size of financial imbalances (credit gap and asset price measures affected by HP filtering end-point bias, GDP and credit data quality issues, and omission of structural transformation effects), leading possibly to an overestimation of potential output.
- Production function: limited by availability of Vietnam-specific elasticities and by using years of schooling (quantity) rather than direct measures of education quality; TFP measure may capture other factors such as capacity utilization and average hours worked.

### Projection assumptions and scenario inputs (selected)
- Investment projected to grow at 10 percent in average.
- Labor force growth assumed to decline following population growth trends.
- Human capital growth assumed to remain constant at 1.02 percent.
- TFP assumed to grow by an average 1.8 percent for baseline projections; regressions support a range between 1.7 and 2.5 percent.

### Conclusions and synthesis
- The four methodologies provide a range of estimates for Vietnam’s potential output in 2017, clustering around 6.4–6.5 percent.
- On balance, staff assess potential growth at 6.5 percent in 2017 (higher than the 2014 estimate of 6.2 percent).
- Output gap estimates vary by method (HP: 0.75 percent; MVF: -0.1 percent), with the consolidated assessment placing the output gap at 0.4 percent in 2017.
- Under baseline assumptions the model projects potential growth could average 6.8 percent over the medium-term and reach 7.3 percent by 2023, conditional on continued reforms, FDI inflows, and productivity improvements.

*IMF staff estimates and analyses as presented in the source content.*

### 6.5 percent over the medium-term, and

### cr18216 - 6.5 percent over the medium-term, and

### Potential growth and drivers
- Potential growth is estimated at 6.5 percent over the medium-term, and the output gap is expected to close in 2019.
- Drivers cited:
  - Relatively high investment levels.
  - A large and well-educated labor force moving towards higher value-added industries.
  - Reforms that reduced the size of the SOE sector and boosted private sector participation, enhancing productivity.
  - Booming FDI sector that can enhance capital quality and facilitate technology and expertise transfer to the domestic sector.
- Caveats and extensions:
  - Further work will explicitly incorporate structural transformation due to labor reallocation and better account for the quality of human capital (education quality).
  - Improvements in data quality (real estate prices, quarterly GDP, unemployment rate and labor force in the informal sector, capacity utilization) could enhance the analysis.

### Potential growth estimates (2001–20)
- Charted series: Estimates, Average, Growth (In percent).
- Sources: Authorities' Data and IMF Staff estimates.

### Models, equations, and calibration (Appendix I)
- I. HP filter
  - Standard quarterly λ = 1,600 is used.
  - GDP series: actual data for 2000Q1–2017Q4 and staff GDP projections for 2018Q1–2019Q4 to minimize end-point bias.

- II. Multivariate Filter — core equations and interpretation
  - Output: potential output level evolves according to potential growth and shocks; potential growth converges to steady-state path according to parameter θ; output gap subject to demand shocks.
  - Phillips Curve: core inflation (π) related to lagged inflation, output gap, and shocks.
  - Okun’s Law: unemployment gap dynamics linked to output gap and shocks.
  - Unemployment (NAIRU) is time varying, subject to shocks and variations in trend allowing persistent deviations from steady-state NAIRU.

- Calibration method
  - Vietnam: Regularized Maximum Likelihood.

- III. Phillips Curve (Box 1) — empirical notes
  - In regression (1), statistical significance of coefficients on output gap and exchange rate is weak, at 0.2.
  - Estimated coefficient of output gap may be affected by quarterly GDP quality undermining HP filter calculation.
  - Exchange rate maintained within a tight range to the U.S. Dollar during most of the period, limiting variation to explain inflation.
  - Adding change in reserves and lagged credit growth increases adjusted R2 to 0.74 for the period 2008–12; reserve intervention impacted inflation in that period.
  - One caveat: small number of observations may undermine robustness of regression (2).

- Calibration table (Prior -> Posterior)
  - lambda 0.250 -> 0.498
  - beta 0.250 -> 0.252
  - phi 0.600 -> 0.618
  - theta 0.100 -> 0.161
  - std_RES_LGDP_BAR 0.500 -> 0.500
  - std_RES_G 0.500 -> 0.500
  - std_RES_Y 1.000 -> 0.999
  - std_RES_PIE 1.000 -> 1.365
  - std_RES_UNR_GAP 0.800 -> 0.798
  - std_RES_UNR_BAR 0.300 -> 0.300
  - std_RES_G_UNR_BAR 0.300 -> 0.298
  - tau1 0.100 -> 0.094
  - tau2 0.100 -> 0.136
  - tau3 0.100 -> 0.101
  - tau4 0.100 -> 0.101
  - Other Calibration: Gss 2.2; Uss 6.5

- IV. Multivariate Filter with Financial Frictions
  - Sustainable GDP estimated by decomposing observed GDP into sustainable output and business cycle components.
  - Variance ratio constraint (ρ) mirrors HP filter characteristics.
  - Observable variables x_t included to help identify output gap: headline inflation, stock market price growth, credit gap.
  - Model parameters estimated via maximum likelihood.

### Production function approach and projection assumptions (Potential Output Estimates)
- Labor statistics
  - Participation rate projected to remain constant during projection period.
  - Labor force growth projected to continue to decline at the rate of population growth (United Nation’s population projections).
  - Unemployment rate assumed unchanged given large informal/self-employment share and lack of reliable data.

- Human capital
  - Schooling for 2011–23 extrapolated from a linear trend.
  - Return to education assumed constant at 7% over the period.

- Physical capital
  - Capital stock data available from Penn World Table until 2014.
  - For 2015–23, new stock calculated as K_t = (1 - δ_{t-1}) K_{t-1} + I_t.
  - Depreciation ratio projected constant at 4.5 percent.
  - Investment data based on IMF baseline projections for gross fixed capital formation (2015–17 provided by Vietnamese authorities).
  - Investment projected to grow by 10 percent in 2018–23, in line with recent average growth rates.
  - Capacity utilization assumed unchanged; actual changes captured in TFP.

### Vietnam: Credit growth and asset market valuations — main findings
- Context and data constraints
  - Vietnam experienced strong credit growth for several years and more recently strong asset price growth.
  - Analysis constrained by data weaknesses: lack of accurate sectoral credit growth (real estate, preferred sectors), lack of granular bank and corporate balance sheet data, and absence of official real estate price statistics; housing price indices from private firms are volatile and sample sizes small.

- A. Credit Growth — stylized facts and indicators
  - Credit-to-GDP ratios rose from around 30 percent in the early 2000s to around 130 percent in 2017.
  - HP filter trend (lambda = 400,000) shows clear upward trend indicating financial deepening; sharp increases after 2009–2010, correction around 2011 real estate bust, and strong increase since 2012 linked to government high annual credit growth targets.

- Early warning indicators (EWI) used
  - Credit gap: deviation of credit-to-GDP from HP trend.
  - Credit growth gap: year-on-year percentage point changes in credit-to-GDP ratios.
  - Literature notes: credit gap widely used and predictive; credit growth gap may be more informative for emerging economies.

- EWI results for Vietnam
  - Credit gap was positive but less than 10 percent in 2017, versus more than 20 percent during the 2011 crisis.
  - Credit growth gap reached almost the 2011 crisis level in mid-2016 but declined in 2017 due to higher nominal GDP growth and lower credit growth.
  - Aggregate credit data cannot capture sectoral risks (e.g., high consumer loan growth) or risks from directed credit to preferred sectors at lower interest rates.

- Scenarios and policy implications
  - Government’s credit growth target for 2018 is 17 percent.
  - Baseline scenario: nominal GDP growth around 10.5 percent in 2018 and 2019 → credit gap and credit growth gap increase but remain around 6–7 percent and 7–9 percentage points in 2019.
  - If credit growth accelerates to 20 percent in 2018 and 2019 → credit gap and credit growth gap would swiftly rise to approach previous crisis levels, posing financial stability risks.
  - To close the credit gap to zero, credit growth should remain around 14 percent in 2018.
  - Long-run recommendation: phase out credit growth targets and adopt a monetary policy framework based on inflation targeting and open market operations so credit growth is market-determined.

- Benchmarking credit-to-GDP
  - Benchmark credit-to-GDP (Cbench) assumed function of per capita GDP; deviations from fitted values interpreted as excessive credit supply.
  - Simple regression framework used; per capita GDP explains cross-country variation well.
  - Benchmarking indicates recent credit-to-GDP increase may have been too rapid; ratios are outside the 1.5-SD band after recovery from 2011 crisis.
  - Historical evidence: large positive deviations from benchmark often led to banking crises and sharp credit corrections, though substantial deviations can persist without correction in some cases.

### Key statistics and thresholds preserved exactly
- Potential growth: 6.5 percent
- Output gap expected to close in 2019
- HP filter quarterly λ used for GDP: 1,600
- HP filter for credit-to-GDP trend: lambda = 400,000
- Capital stock depreciation ratio projected constant at 4.5 percent
- Return to education assumed constant at 7%
- Investment projected to grow by 10 percent in 2018–23
- Government credit growth target for 2018: 17 percent
- Baseline nominal GDP growth projection for 2018 and 2019: around 10.5 percent
- Credit growth to close credit gap to zero in 2018: around 14 percent
- Alternative credit growth scenario considered: 20 percent
- Calibration priors and posteriors listed exactly (see Calibration table above)
- Other Calibration: Gss 2.2; Uss 6.5

*Source: IMF staff analysis and estimates as presented in the provided chapter content.*

### 9.      The stock market performance in

### 9.      The stock market performance in

### Stock market performance
- Year-on-year growth of stock prices on a US dollar basis:
  - more than 15 percent in 2016
  - more than 40 percent in 2017
- Market capitalization:
  - increased significantly but steadily in volume and valuation until 2016
  - increased by more than 70 percent in 2017
  - rose to around 70 percent of GDP at the end of 2017 from around 40 percent in 2016
- Market composition:
  - Market capitalization reflects the sum of the Ho Chi Minh City Stock Exchange (HOSE), Hanoi Stock Exchange (HNX), and the Unlisted Public Company Market (UPCoM)
  - At the end of 2017, HOSE’s share of total market capitalization in Vietnam was around 70 percent
  - As of January 2018, the share of Top 5 and 10 firms in terms of market capitalization in HOSE and NHX, respectively, are 36 percent and 52 percent; among these top 5 and 10 firms, 4 and 7 firms, respectively are SOEs

### Valuation measures and investor expectations
- Rapid increases in valuation measures in 2017:
  - Substantial changes occurred in price-to-earnings ratios (PER), price-to-book ratios (PBR), and dividend yield
  - Large and rapid changes in 2017 imply investors’ expectations for Vietnam dramatically changed during 2017
- Risk implication:
  - Elevated expectations of future growth appear to be driving high stock prices
  - If elevated expectations for solid economic growth are not realized, high stock prices could experience correction
- Relative level:
  - Valuation levels are not substantially higher than those in other countries, but the magnitude and speed of 2017 changes are notable

### SOE privatization, equity market channels, and potential spillovers
- Importance of continued SOE reforms and privatization:
  - Continued progress in privatizing and reforming state-owned enterprises (SOE) will be critical to support high economic growth and to contribute directly to the performance of listed SOEs, which still dominate stock markets in Vietnam
- Channels through which stock market corrections could affect the real economy:
  - declines in IPOs
  - slowdown in SOE equitization
  - deterioration in bank and corporate balance sheets
- Data gap:
  - More granular data are needed to further assess the impact through these channels

### Real estate prices — methods for misalignment assessment
- Two approaches used:
  1. Price-to-Rent ratios (house prices divided by rent indexes) — rent represents the fundamental value; ratios should be stationary (Poterba 1984)
  2. Regression-based assessment — conventional model for house price valuation using affordability defined as house prices divided by per capita GDP; estimation conducted with (unbalanced) cross-country panel data for Indonesia, South Korea, Malaysia, Philippines, Singapore, Thailand, China and Vietnam; deviations from fitted values are interpreted as house price misalignment
- Note on regression implementation:
  - The estimated equation gives fitted values for the growth rate of house prices rather than levels; base level set to historical average level of real house prices in data and fitted levels calculated by extending house prices using estimated growth rate

### Real estate prices — findings for Vietnam
- Price-to-rent ratios and HCMC residential prices:
  - Since 2015, residential property price index in HCMC has continued to increase
  - Price-to-rent ratios have not significantly increased as in the previous crisis period in 2011 and are converging to the 2015 level due to increases in rent
  - Suggests increases in house prices are justified by economic fundamentals
- Drivers supporting housing demand:
  - strong income growth of a growing middle class
  - steady rate of urbanization
  - limited government policy to provide mortgages at low interest rates to low income households through state-owned banks may have helped support property prices
- Regression-based assessment:
  - Indicates Vietnam’s property price appears not to substantially deviate from the estimated trend in recent years
  - Based on forecasts for per capita income and working age population, the model suggests real house prices will be almost flat or increase only moderately for the next several years as houses become more affordable
- Caution on interpretation:
  - For Asian countries other than Vietnam, the model accounts for most variation in house prices
  - For Vietnam, large deviations of actual historical values from fitted values imply the model does not seem to have strong predictive powers
  - Low quality of real estate data in Vietnam may explain large deviations; more reliable house price data are necessary for more precise assessment

### Conclusions and policy recommendations
- Overall assessment:
  - Asset price and credit growth appear to be stronger than warranted by fundamentals
- Policy recommendations:
  - The central bank’s credit growth target should be decreased to avoid excessive risk taking by banks; raising the target would not be warranted
  - Continued SOE and other growth-supporting reforms would help better align the rapid rise in stock market valuations with fundamentals
  - The State Bank of Vietnam (SBV) should develop a macroprudential policy framework including loan-to-value (LTV) and debt service-to-income (DSTI) requirements to deal with future possibilities of excessive exuberance in the real estate market

### Fiscal rules and debt-anchor calibration (selected findings)
- Current statutory limit:
  - Vietnam follows a debt rule with a statutory limit of 65 percent of GDP on public and publicly guaranteed debt (PPG)
- Fiscal context:
  - Authorities committed to adhering to the debt rule; PPG has remained below the statutory limit in recent years
  - Estimated deficits: 4.5 percent of GDP in 2017 and 4.6 percent in 2018
  - PPG debt increased rapidly between 2011 and 2016; average deficits rose from 2 percent (2000–2011) to 6.2 percent (2012–2016)
- Demographic fiscal pressures:
  - Age-related spending needs estimated to increase by 8 percent of GDP by 2050, increasing pension outlays by 6.7 percent of GDP
  - Current pension scheme projected to run deficits as of 2030 absent reforms
- Calibration steps and results:
  - Step 1: set maximum debt limit at 80 percent of GDP (70 percent MAC DSA limit plus 10 percent buffer for public guarantees); subtracting estimated 8 percent of GDP age-related spending reduces this to 72 percent of GDP
  - Step 2: perform stochastic simulations identifying joint distribution of GDP growth, interest rates, and exchange rate based on past means and variances
  - Step 3: estimate debt anchor in baseline scenario — simulations starting in 2017 suggest a debt anchor of 66.5 percent could be sufficient to keep debt below the maximum debt limit of 80 percent with 90 percent certainty
  - Step 4: account for fiscal risks and long-term aging costs — a lower debt anchor may be warranted
- Scenario analysis — Debt Anchor (% of GDP):
  - 1) 5 percent realization of contingent liabilities: 58.8
  - 2) Real GDP growth slowdown to 4 percent of GDP: 52.3
  - 3) Scenario 2 + 2.5 percent realization of contingent liabilities: 50.1
  - 4) Scenario 2 + real depreciation of 4 percent per year: 49.9
  - 5) Age related spending (maximum debt limit of 72 percent of): 53.8
- Suggested operational rule:
  - The paper suggests an expenditure rule as an additional operational target, concluding that a first best option would be a nominal expenditure growth rule of 10–11 percent

*Source: IMF staff calculations and text from the provided content unit.*

### 6.      Each of these scenarios demonstrates that the presence of fiscal risks and contingent

### cr18216 - 6.      Each of these scenarios demonstrates that the presence of fiscal risks and contingent

### Fiscal risks and debt ceilings
- Presence of fiscal risks and contingent liabilities requires a lower debt limit than the current statutory ceiling.
- Scenarios 3 and 4 indicate that a combined growth and contingent liability shock, or growth and exchange rate shock, would necessitate a debt target of below 55 percent of GDP in the medium-term.
- Recommendation: authorities should aim at a target of 55 percent of GDP to account for fiscal risks and aging costs that could materialize in the medium to long term; the current 65 percent of GDP ceiling should be considered an upper bound.

### Augmenting debt ceilings with expenditure rules (ERs)
- A debt rule provides a useful anchor for medium term fiscal policy but should be augmented with operational guidance for short-term policy.
- Operational rules target aggregates under direct government control (public expenditure, revenue, budget balance).
- Vietnam’s current framework: multiple thresholds and floors determine the current balance, overall balance, tax revenue, total spending (upper limit), capital spending (lower threshold), and others; some specified annually, some medium-term.
- Drawbacks of multiple targets:
  - If one target underperforms, others will (e.g., revenue shortfall → cuts in spending, often via capital spending → miss capital spending target).
  - Limits flexibility to respond to crises.
  - Complicates communication, especially if several targets are missed to compensate for one missed target.
- Prefer replacing multiple targets with one or two shorter-term operational rules under government control with a close and predictable link to debt dynamics.
- Suggested operational rule in this paper: a ceiling on the nominal growth rate of expenditure (an expenditure rule).
- Appendix I contains principles of operational rules and pros/cons of types of operational rules.

### Cross-country experience on expenditure rules
- As of 2015:
  - 46 countries had ERs, 14 of which were in emerging markets (EMs).
- ER characteristics:
  - Limits on total, primary, or current spending.
  - Limits apply to nominal or real expenditure.
  - Typically set in absolute terms (levels) or growth rates, occasionally in percent of GDP.
  - Time horizon of three to five years.
- In most of the 14 EMs, ER is accompanied by a debt rule (Botswana, Brazil, Bulgaria, Croatia, Ecuador, Georgia, Namibia, Peru, Poland, Romania).
- Half of nominal and real growth rate ceilings are specified in terms of either potential or nominal GDP growth; others specified in absolute growth rates.
- Of the 14 rules, 9 cover overall expenditures; 5 cover current and/or primary expenditure.
- EMs with ERs specified in nominal growth rates: Bulgaria, Colombia, Poland, Romania (all but Colombia also have a debt rule).

- Table 2 stylized counts (as presented):
  - Public Expenditure Specified as ... Number of Countries
  - Ratio to GDP (30-40%) 4
  - Ratio to Revenue 3
  - Nominal Growth Ceiling 4
  - Real Growth Ceiling 4
  - Growth Limited by Potential GDP Growth 3
  - Growth Limited by Total GDP Growth 1
  - Others 1 3
- Source cited: IMF FAD Fiscal Rules Database.

### Selecting the 'right' expenditure rule for Vietnam
- Recommendation: an ER specified in terms of nominal growth rates would be appropriate for Vietnam.
- Design considerations:
  - Expenditure growth could be calibrated on the basis of nominal GDP growth initially, but should be relatively independent of the GDP series during implementation.
  - ERs should be re-calibrated every 3 to 5 years to account for underlying economic changes.
- Advantages of growth-rate ERs vs ratio-to-GDP:
  - ERs defined in growth rates can support macroeconomic stabilization and improve compliance; ERs set as a ratio to GDP tend to be procyclical.
  - Compliance with ERs specified as nominal ceilings or nominal growth rate ceilings is significantly higher than those specified as a ratio to GDP or in real growth terms (Cordes et al., 2015).
  - Government has less incentive to comply with ratio-to-GDP rules since accountability for non-compliance is reduced.
- Additional design recommendations:
  - ER should be formulated on overall public expenditure (general government level) to avoid creative accounting.
  - ER should be defined in nominal terms for transparency and ease of monitoring and enforcement.
  - Nominal targets incorporate inflation developments and have better stabilization properties.
  - Expenditure targets in real terms can reduce stabilization effect and can foster strategic manipulation of deflators.

### Calibrating expenditure rules and illustrative scenarios
- A 10 to 11 percent growth rate ceiling on a nominal expenditure rule would be consistent with a medium-term debt target of 55 percent of GDP.
- Methodology notes for Table 3:
  - Expenditure growth rates presented are average expenditure growth for the first 5 years (2018–2023) under a given fiscal balance and assuming a stable revenue ratio at 23 percent of GDP.
  - Initial debt level set at end-2017 level of 58.5 percent of GDP.
  - Illustrative scenario assumes a long-term nominal interest rate of 7 percent, and a long-term nominal GDP growth rate of 10 percent.
- Table 3: Options for Expenditure Rules (key rows preserved exactly)
  - Scenario — Nominal GDP Growth — Overall Deficit — Nominal Expenditure Growth Rule — Debt Level in 2023
  - Illustrative Scenario — 10% — 4.4% — 10.7% — 54%
  - Consolidation Scenario 1/ — 10.5% — Declining to 3 % by 2023 — 10.3% — 52.8%
  - Stronger growth — 12% — 5.1% — 11.2% — 54%
  - Weaker growth — 7% — 3.0% — 9.6% — 54%
- Notes on the consolidation scenario:
  - Mirrors 2018 Article IV Staff Report.
  - Incorporates negative GDP growth impact of fiscal consolidation with a multiplier of 0.3.
  - Assumes implementation of a tax policy reform in 2019, and a positive impact of public investment efficiency gains following PIM improvements to begin in 2020.
  - Assumes structural reforms to improve public spending efficiency and address bank recapitalization needs; together with higher public investment expected to increase real GDP growth by 0.4 percent over the baseline by 2023.
- Illustration: In the illustrative scenario, Vietnam’s debt would reach 54 percent of GDP by 2023 if the average deficit was 4.4 percent of GDP and expenditure growth around 10.7 percent on average over the next 5 years.
- The illustrative scenario is close to the consolidation scenario suggested in the 2018 Article IV, where the deficit is declining to 3 percent by 2023, with an average expenditure growth of 10.3 percent annually until 2023.

### Sensitivity analysis and prudent rule recommendation
- Sensitivity to nominal GDP growth implies deficit targets and expenditure growth rates can be relaxed or tightened depending on growth performance.
- Given the possibility of growth slowdown, increasing aging costs and materialization of contingent liabilities, a prudent fiscal rule framework should set an expenditure growth ceiling of 10–11 percent per year (tantamount to a limit to the overall balance of 3–4 percent of GDP).

### Appendix I — Principles of fiscal rules (summary)
- Fiscal rules system should be anchored by a debt objective to preserve fiscal sustainability.
- Debt anchor:
  - Directly linked to fiscal sustainability; informs medium-term expectations.
  - Not meant to provide short-term guidance.
  - Public debt is persistent and affected by many developments besides changes in the overall budget balance.
- Short-term guidance should be via operational rules under direct government control and with a close and predictable link to debt dynamics.
- Framework design principles:
  - Parsimonious set of rules: combine a debt rule with one or a few operational rules (expenditure growth limit or budget balance).
  - Rules should assist the debt anchor in ensuring sustainability; should ensure stabilization (avoid procyclicality); be simple, easily understood, easy to monitor/enforce; provide clear operational guidance; be resilient (in place for sustained period).
- Trade-offs and multiple rules:
  - Fiscal policy has multiple objectives; one operational rule per key objective may be preferable.
  - Example: combine a ceiling on total expenditure and a ceiling on current expenditure to protect capital expenditure while ensuring sustainability.
- Calibration and correction mechanisms:
  - Relationships between debt anchor and operational rules should be transparent and grounded in economic analysis.
  - Debt ceiling should be set first; operational rules calibrated from the debt ceiling to ensure consistency.
  - Correction mechanisms after breaches can avoid drifting away from the anchor; must balance credibility and avoiding abrupt economically inappropriate corrections.

*Source: IMF Staff Estimates.*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2018/cr18216.pdf_