## wpiea2019246-print-pdf - Section III describes the data and methodology used; Section IV contains the main results

## Source details

**Canonical URL:** [wpiea2019246-print-pdf - Section III describes the data and methodology used; Section IV contains the main results](https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019246-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2019/wpiea2019246-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2019/wpiea2019246-print-pdf.pdf.json)

---

### Methodology and data — Dominican Republic GaR application
- Data sample: quarterly observations from 1996 through 2018.
- Overall steps:
  - Select explanatory macrofinancial variables and group them into partitions via principal components analysis.
  - Use first principal components of each partition as regressors in quantile regressions.
  - Fit a conditional probability density function of future GDP growth using a skewed generalized t distribution.
  - Apply scenarios to estimate impacts on growth distribution at 4, 8, and 12 quarters ahead.
- Variable grouping (32 quarterly series into five partitions):
  - Domestic Financial Conditions (Domestic FCI): M1, M2, M3, Saving interest rate in USD, Lending interest rate in USD, 30-day interest rate, Reserve requirements rate, Inflation, Policy rate, Commercial interest rate, EMBIG spread, Average lending rate, 2-year interest rate, Personal interest rate, Interbank interest rate, Mortgage interest rate.
  - Domestic Leverage: Private credit-to-GDP, Credit growth, Regulatory capital, NPLs.
  - External Cost of Borrowing: 6-month LIBOR, US Fed Funds rate.
  - External Liquidity: World bond flows, EM bond flows, MOVE index, VIX.
  - External Demand: Terms of Trade, Euro area growth, Net exports-to-GDP, US growth, REER, Oil price.
- Interpretation of partition signs:
  - Positive partition values indicate tighter/contractionary conditions; negative values indicate more accommodative/expansionary conditions.
- Current (end-2018) assessment:
  - On average, the five indicators point to neutral financial conditions.
  - Domestic leverage: neutral in 2018 (consistent with mid-2018 monetary policy tightening that slowed credit growth).
  - Domestic FCI: still accommodative as EMBIG spread and inflation remain low.
  - External indicators: reflect normalization with tightening liquidity (lower bond flows, higher market volatility).

### Quantile regression specification and pdf fitting
- Quantile regression model (notation preserved):
  - Q_{growth h,t+ h, t}^q = β_{0}^q + β_{1}^q X_{1,t} + β_{2}^q X_{2,t} + β_{3}^q X_{3,t} + β_{4}^q X_{4,t} + β_{5}^q X_{5,t} + γ^q growth_{t}^h + e_{t+ h}^q
  - q indicates quantile; h indicates horizon (quarters); growth_t is current GDP growth rate (domestic demand); X_{n,t} are partitioned predictors.
- Regression quantiles and horizons:
  - Quantiles: 0.10, 0.25, 0.50, 0.75, 0.90.
  - Horizons: h = 4, 8, 12 quarters ahead.
- Pdf fitting and risk metrics:
  - Conditional pdf for future GDP growth fitted via skewed generalized t distribution, mode conditional on IMF growth forecasts.
  - Risk metrics: x percent GaR (x used = 5 percent and 10 percent), probability of negative economic growth, baseline vs shock scenario comparisons.

### Main results — short- and medium-term (h=4 and h=8)
- Short-term horizon (h=4): selected coefficients (coefficient (p-value))
  - Domestic Leverage: 0.10: -1.03*** (0.000); 0.25: -0.85*** (0.000); 0.50: -0.76*** (0.000); 0.75: -0.60*** (0.000); 0.90: -0.61*** (0.000); mean: -0.80*** (0.000)
  - External Borrowing Cost: mean: 0.21** (0.099)
  - External Demand: mean: -0.61*** (0.000)
  - External Liquidity: mean: 0.25** (0.050)
  - Domestic FCI: mean: -0.51*** (0.002)
  - Domestic Demand (current growth): mean: -0.59*** (0.000)
  - R^2 by quantile: 0.46, 0.40, 0.33, 0.27, 0.23; overall R^2: 0.52
- Medium-term horizon (h=8): selected coefficients (coefficient (p-value))
  - Domestic Leverage: mean: -0.82*** (0.000)
  - External Borrowing Cost: mean: -0.29* (0.025)
  - External Demand: mean: -0.64*** (0.000)
  - External Liquidity: mean: 0.14 (0.222)
  - Domestic FCI: mean: -0.08 (0.628)
  - Domestic Demand (current growth): mean: -0.40*** (0.002)
  - R^2 by quantile: 0.44, 0.38, 0.37, 0.32, 0.35; overall R^2: 0.56
- Key short- and medium-term patterns:
  - Domestic Leverage: strong, statistically significant negative coefficients across quantiles and horizons, indicating tighter leverage associated with lower future growth, especially in downside quantiles.
  - External Demand: consistently negative and statistically significant across quantiles and horizons.
  - Domestic FCI and current Domestic Demand: negative and often significant at h=4, especially in lower quantiles; Domestic Demand remains important at h=8 in lower quantiles.
  - External Borrowing Cost: weak/mixed at h=4, increasingly negative and sometimes significant at h=8.
  - External Liquidity: generally positive but mostly insignificant; some short-term positive significance at h=4.

### Long-term horizon (h=12) — coefficients and interpretations
- Selected regression coefficients by partition (series of values correspond to quantiles 0.10, 0.25, 0.50, 0.75, 0.90, mean):
  - Domestic Leverage: -0.68*** (0.000); -0.77*** (0.000); -0.78*** (0.000); -0.89*** (0.000); -0.85*** (0.000); -0.75*** (0.000)
  - External Borrowing Cost: -0.64*** (0.000); 0.59*** (0.000); -0.59*** (0.000); -0.50*** (0.000); -0.59*** (0.000); 0.48*** (0.000)
  - External Demand: -0.30*** (0.010); -0.55*** (0.000); -0.80*** (0.000); -0.87*** (0.000); -0.93*** (0.000); -0.76*** (0.000)
  - External Liquidity: 0.14 (0.109); 0.08 (0.325); 0.09 (0.398); 0.00 (0.982); 0.10 (0.517); 0.05 (0.596)
  - Domestic FCI: 0.48*** (0.000); 0.22** (0.067); -0.02 (0.914); -0.17 (0.331); -0.24 (0.203); -0.11 (0.420)
  - Domestic Demand: -0.14 (0.254); -0.25*** (0.031); -0.36*** (0.005); -0.72*** (0.000); -0.59*** (0.000); -0.52*** (0.000)
  - R^2: 0.49, 0.50, 0.49, 0.50, 0.47, 0.71
- Long-term empirical findings:
  - Domestic Leverage: strongest and consistently significant impact at all horizons; credit availability is the most important and persistent driver of domestic growth.
  - External Demand: second most important; strongest at medium- and long-term horizons; transmission always significant at h=8 and h=12.
  - Domestic FCI: significant mainly when growth is below average; sign switches—expansionary at h=4 but can amplify vulnerabilities at h=12.
  - External Borrowing Costs: no short-term impact, significant negative effects at h=12 across distribution; at h=8 significant mainly in high-growth states.
  - External Liquidity: no significant marginal impact at any horizon.
  - Domestic demand autoregressive effect: current growth shock associated with lower future growth, potentially reflecting negative structural shifts.

### Scenario analysis — baseline and shock impacts
- Baseline (end-2018) distribution:
  - 5 (10) percent probability that growth in 2019 could be as low as 4.65 (4.85) percent.
  - Probability of negative growth close to zero.
- Shock methodology:
  - One-standard-deviation negative shocks applied independently to each partition; impacts analyzed at h=4 and h=12.
- Selected shock impacts (modal growth changes and probabilities):
  - Domestic leverage shock:
    - Short run (h=4): modal growth reduced by almost 3 percent (from 7.2 to 4.3 percent); no risk of negative future growth.
    - Long run (h=12): same shock decreases modal growth forecast by 4 percent (from 7.2 to 4.3 percent); leads to a sizable probability of 6.6 percent for growth below zero.
  - Domestic demand negative shock:
    - Decreases modal forecast by 2.4 percent at h=4 and by 3 percent at h=12.
    - In worst 5 percent scenarios, growth could fall to 3.4 percent in the short horizon, with risks of sluggish growth (0.5 percent) in the longer term.
  - Domestic FCI shock:
    - Short term: tighter domestic FCI could reduce modal growth by almost 2 percent (from 7.2 to 5.1 percent) and increase probability of negative growth to 2.5 percent.
    - Long term: mode improves to 5.6 percent; very small impact on other distribution parameters.
    - Implies monetary policy effective in staving off slowdowns; tightening should be done only in response to upward inflationary pressures.
  - External demand shock:
    - Most detrimental among external factors, especially at longer horizons; negative shock can increase probability of negative growth to around 4-5 percent.
  - External borrowing cost shock:
    - Higher external borrowing costs reduce domestic growth prospects mainly at the three-year horizon due to lagged transmission.
  - External liquidity shock:
    - No significant short-term impact given relatively weak global integration and sufficient local market liquidity.

### Policy implications and recommendations
- Domestic credit and macroprudential policy:
  - Private sector credit is the key driver; negative shocks to credit availability would strongly and persistently reduce growth.
  - Macroprudential policies aimed at credit quality should help keep NPLs low and maintain leverage necessary to close the negative credit gap and deepen domestic financial markets.
- Monetary policy:
  - Central bank can stimulate the economy in bad times but should be cautious about negative feedback two years later.
  - Monetary expansion in “bad” times: stimulus at one-year horizon, contractionary impact at three-year horizon — tradeoff between short-run stimulus and long-run vulnerabilities.
  - Decisions to tighten should be carefully analyzed and made only in response to upward inflationary pressures.
- External vigilance:
  - Authorities should monitor low growth prospects in the US and the euro area, worsening terms of trade, and competitiveness.
  - External borrowing costs matter mainly at long horizons; external liquidity has limited marginal effect.
- Complementary policies:
  - Create fiscal space, accumulate adequate foreign reserves, and build liquidity and capital buffers in the banking system in “good” times to cushion potential downturns.

*Source: wpiea2019246-print-pdf - Section III describes the data and methodology used; Section IV contains the main results*

### Section III describes the data and methodology used; Section IV contains the main results;

### wpiea2019246-print-pdf - Section III describes the data and methodology used; Section IV contains the main results

### II. Literature review — informational content of financial-sector data and GaR
- Asset prices and pooled information improve GDP growth forecasts (Stock and Watson (2003)).
- Financial statistical factors contain information about macroeconomic variables, most clearly output (Erdem and Tsatsaronis (2013)).
- Financial variables (credit growth, stock prices, house prices, bond yields) have considerable predictive power for macro variables at short-term horizons (Chen and Ranciere (2016) for 62 countries, 1980–2013).
- Financial leverage is an important source of vulnerability (Kiyotaki and Moore (1997)).
- Rapid deleveraging can depress aggregate demand (Eggertsson and Krugman (2012)).
- Growth-at-Risk (GaR) framework:
  - Maps financial and economic conditions into the probability distribution of future GDP growth.
  - Does not establish causal links; designed to estimate severity of risks to growth via quantile regression and scenario analysis.
  - Finds nonlinear relationships between financial conditions and conditional distribution of GDP growth; left tail of growth distribution positively correlated with slack in financial conditions (Adrian, Boyarchenko and Giannone (2019)).
  - Highlights monetary policy’s role in tradeoffs between present and future objectives through financial conditions (Adrian and Liang (2016)).
- Country-specific GaR applications emphasize context dependence:
  - Albania: trading-partner conditions largest short-run impact on growth (IMF 2019a).
  - Panama: external conditions (financial and demand) key short-run drivers (Beaton (2019)).
  - Portugal: domestic financial conditions (interest rates, leverage, credit growth) most powerful signal of growth risks (Kouet-Vickot (2019)).
  - Singapore: external financial conditions most explanatory for future GDP growth (IMF 2019b).

### III. Methodology and data — Dominican Republic GaR application
- Overall steps applied:
  - Select explanatory macrofinancial variables and group them into partitions via principal components analysis.
  - Use partitioned indicators in quantile regression to forecast output growth.
  - Fit a probability density function of future GDP growth using a skewed generalized t distribution to quantify downside risks.
  - Apply scenarios to estimate impact of hypothetical shocks on growth distribution at 4, 8, and 12 quarters ahead.
- Data sample: quarterly observations from 1996 through 2018.
- Variable grouping:
  - 32 quarterly economic series (1996–2018) grouped into five partitions: Domestic Financial Conditions (Domestic FCI), Domestic Leverage, External Cost of Borrowing, External Liquidity, External Demand.
  - First principal components of each partition used as regressors in quantile regressions.
- Table 1 partitions and selected indicators (as presented):
  - Domestic Financial Conditions: M1, M2, M3, Saving interest rate in USD, Lending interest rate in USD, 30-day interest rate, Reserve requirements rate, Inflation, Policy rate, Commercial interest rate, EMBIG spread, Average lending rate, 2-year interest rate, Personal interest rate, Interbank interest rate, Mortgage interest rate.
  - Domestic Leverage: Private credit-to-GDP, Credit growth, Regulatory capital, NPLs.
  - External Cost of Borrowing: 6-month LIBOR, US Fed Funds rate.
  - External Liquidity: World bond flows, EM bond flows, MOVE index, VIX.
  - External Demand: Terms of Trade, Euro area growth, Net exports-to-GDP, US growth, REER, Oil price.
- Interpretation of partition signs:
  - Positive partition values indicate tighter/contractionary conditions; negative values indicate more accommodative/expansionary conditions.
- Partition-specific economic interpretations:
  1. Domestic FCI captures price of risk and monetary policy via interest rates and policy indicators; higher values (higher interest rates, spreads, inflation, reserve requirements, policy rate) indicate tighter stance and contractionary impact; higher monetary aggregates and savings rates indicate looser conditions and expansionary impact.
  2. Domestic Leverage reflects credit conditions and financial stability; higher values indicate lower availability of credit (lower credit growth, higher NPLs, higher regulatory capital), contractionary for growth.
  3. External Cost of Borrowing reflects foreign market credit cost; increases (higher LIBOR or US Fed Funds rate) indicate tighter global conditions and potential contractionary transmission to domestic growth if transmission is strong.
  4. External Liquidity captures global market volatility and bond flows; increases (declining bond flows, higher VIX/MOVE) indicate external liquidity crunch, potential transmission to lower domestic growth if domestic liquidity insufficient or banking system dependent on external financing.
  5. External Demand captures external market pressures on trade balance; increases (higher oil price, REER appreciation, decline in trading-partner real incomes, lower net exports-to-GDP, worse terms of trade) indicate contractionary pressures on domestic growth if external demand matters.
- Dynamics and historical behavior (1996–2018):
  - Domestic FCI: broadly accommodative after GFC; tight during 2003 banking crisis and episodic tightening in 2008 and around 2011–12; easing in 2017 to stimulate slowing economy and subsequent gradual tightening.
  - Domestic Leverage: spike (severe tightening) during 2003 crisis (credit-to-GDP fell dramatically), followed by recovery and episodes of temporary credit tightness.
  - External Cost of Borrowing: severe tightness in late 1990s emerging-market crises and GFC; prolonged loose international credit thereafter, with gradual tightening in recent years.
  - External Liquidity: most volatile; dried up significantly in 2008–09, expanded in following decade, tightened again in 2018.
  - External Demand: dominated by commodity cycles; contractionary effect during 2006–14 driven by high oil prices (country is net fuel importer); partly offset by exchange rate depreciation post-2003 and trade-partner real output growth.
- Current (2018) assessment from indices:
  - On average, the five indicators point to neutral financial conditions.
  - Domestic leverage (credit) indicates neutral conditions in 2018, consistent with mid-2018 monetary policy tightening that slowed credit growth.
  - Domestic FCI suggests still accommodative financial conditions as EMBIG spread and inflation remain low.
  - External indicators (cost of borrowing, liquidity, demand) reflect normalization from earlier accommodative stance, with tightening liquidity capturing lower bond flows and higher market volatility.

### III.B. Quantile regression specification and pdf fitting
- Quantile regression specification (as given):
  - Q_{growth h,t+ h, t}^q = β_{0}^q + β_{1}^q X_{1,t} + β_{2}^q X_{2,t} + β_{3}^q X_{3,t} + β_{4}^q X_{4,t} + β_{5}^q X_{5,t} + γ^q growth_{t}^h + e_{t+ h}^q
  - q indicates quantile; h indicates horizon (quarters); growth_t is current GDP growth rate (domestic demand); X_{n,t} are partitioned predictors.
- Regression fitted on quantiles: 0.10, 0.25, 0.50, 0.75, 0.90 and horizons h = 4, 8, 12 quarters ahead.
- Purpose:
  - Estimate mapping between macrofinancial variables and future growth at different horizons and distribution points.
  - Capture nonlinear and state-dependent marginal impacts (coefficients vary across quantiles and horizons).
  - Use quantile regression results to fit conditional pdf for future GDP growth via skewed generalized t distribution.
- Risk metrics derived from estimated pdf:
  - x percent GaR: lowest growth rate observable with x percent probability (x used = 5 percent and 10 percent).
  - Probability of negative economic growth.
  - Comparison between baseline and shock scenarios quantifies risks to growth.

### IV. Main results — interpretation of quantile regressions and term structure of coefficients
- General: Term structure of quantile regression coefficients at different distribution points reveals macrofinancial linkages and shock transmission mechanisms in the Dominican Republic. Table 1 presents regression estimates for each quantile-horizon pair (h = 4, 8, 12). The text below reports results for h=4 and h=8 as provided.

- Short-term horizon (h=4) — selected quantile regression estimates (coefficient (p-value)):
  - Domestic Leverage:
    - 0.10: -1.03*** (0.000)
    - 0.25: -0.85*** (0.000)
    - 0.50: -0.76*** (0.000)
    - 0.75: -0.60*** (0.000)
    - 0.90: -0.61*** (0.000)
    - mean: -0.80*** (0.000)
  - External Borrowing Cost:
    - 0.10: -0.06 (0.770)
    - 0.25: 0.18 (0.303)
    - 0.50: 0.20 (0.216)
    - 0.75: 0.03 (0.843)
    - 0.90: 0.05 (0.774)
    - mean: 0.21** (0.099)
  - External Demand:
    - 0.10: -0.84*** (0.000)
    - 0.25: -0.69*** (0.000)
    - 0.50: -0.45*** (0.006)
    - 0.75: -0.20 (0.109)
    - 0.90: -0.34 (0.035)
    - mean: -0.61*** (0.000)
  - External Liquidity:
    - 0.10: 0.20 (0.179)
    - 0.25: 0.37 (0.014)
    - 0.50: 0.15 (0.354)
    - 0.75: 0.12 (0.357)
    - 0.90: 0.31 (0.021)
    - mean: 0.25** (0.050)
  - Domestic FCI:
    - 0.10: -0.42** (0.061)
    - 0.25: -0.63*** (0.005)
    - 0.50: -0.29 (0.154)
    - 0.75: -0.06 (0.682)
    - 0.90: -0.18 (0.289)
    - mean: -0.51*** (0.002)
  - Domestic Demand (current growth):
    - 0.10: -0.45*** (0.002)
    - 0.25: -0.73*** (0.000)
    - 0.50: -0.52*** (0.001)
    - 0.75: -0.22** (0.068)
    - 0.90: -0.15 (0.241)
    - mean: -0.59*** (0.000)
  - R^2 by quantile: 0.46, 0.40, 0.33, 0.27, 0.23; overall R^2: 0.52

- Medium-term horizon (h=8) — selected quantile regression estimates (coefficient (p-value)):
  - Domestic Leverage:
    - 0.10: -0.87*** (0.000)
    - 0.25: -0.87*** (0.000)
    - 0.50: -0.91*** (0.000)
    - 0.75: -0.65*** (0.000)
    - 0.90: -0.57*** (0.000)
    - mean: -0.82*** (0.000)
  - External Borrowing Cost:
    - 0.10: -0.21 (0.294)
    - 0.25: -0.18 (0.379)
    - 0.50: -0.22 (0.184)
    - 0.75: -0.33* (0.023)
    - 0.90: -0.65*** (0.001)
    - mean: -0.29* (0.025)
  - External Demand:
    - 0.10: -0.71*** (0.000)
    - 0.25: -0.64*** (0.001)
    - 0.50: -0.68*** (0.000)
    - 0.75: -0.47*** (0.006)
    - 0.90: -0.48*** (0.006)
    - mean: -0.64*** (0.000)
  - External Liquidity:
    - 0.10: 0.03 (0.840)
    - 0.25: 0.01 (0.924)
    - 0.50: 0.21 (0.137)
    - 0.75: 0.09 (0.637)
    - 0.90: 0.17 (0.507)
    - mean: 0.14 (0.222)
  - Domestic FCI:
    - 0.10: -0.17 (0.408)
    - 0.25: -0.13 (0.565)
    - 0.50: -0.23 (0.263)
    - 0.75: 0.02 (0.931)
    - 0.90: 0.18 (0.404)
    - mean: -0.08 (0.628)
  - Domestic Demand (current growth):
    - 0.10: -0.80*** (0.000)
    - 0.25: -0.63*** (0.001)
    - 0.50: -0.38* (0.019)
    - 0.75: -0.13 (0.481)
    - 0.90: 0.05 (0.819)
    - mean: -0.40*** (0.002)
  - R^2 by quantile: 0.44, 0.38, 0.37, 0.32, 0.35; overall R^2: 0.56

- Key patterns and interpretations (as implied by coefficients):
  - Domestic Leverage:
    - Strong, statistically significant negative coefficients across quantiles and horizons (h=4 and h=8), indicating tighter domestic leverage conditions are associated with lower future growth, with particularly large negative effects in lower quantiles (downside).
  - External Demand:
    - Consistently negative and statistically significant across quantiles and horizons, indicating weaker external demand is associated with lower future growth.
  - Domestic FCI and Domestic Demand (current growth):
    - At h=4, Domestic FCI and current Domestic Demand display negative and often statistically significant coefficients, particularly in lower quantiles, indicating state-dependent transmission to short-term downside risk.
    - At h=8, Domestic Demand remains negatively associated with future growth, particularly in lower quantiles.
  - External Borrowing Cost:
    - Effects are weaker and more mixed at h=4 (mostly insignficant) but show increasingly negative and sometimes significant effects at higher quantiles for h=8 (e.g., 0.90: -0.65*** (0.001)), resulting in a negative mean coefficient at h=8 (-0.29* (0.025)).
  - External Liquidity:
    - Coefficients generally positive but mostly statistically insignificant; some positive and significant values at h=4 (0.25 (0.014) at 0.25 quantile; mean 0.25** (0.050)) suggesting periods when greater external liquidity is associated with higher short-term growth outcomes.

*Source: wpiea2019246-print-pdf - Section III describes the data and methodology used; Section IV contains the main results*

### 3.     Long-term horizon (h=12)

### 3.     Long-term horizon (h=12)

### Regression coefficients (selected partitions)
- Domestic Leverage: -0.68*** (0.000); -0.77*** (0.000); -0.78*** (0.000); -0.89*** (0.000); -0.85*** (0.000); -0.75*** (0.000)
- External Borr. Cost: -0.64*** (0.000); 0.59*** (0.000); -0.59*** (0.000); -0.50*** (0.000); -0.59*** (0.000); 0.48*** (0.000)
- External Demand: -0.30*** (0.010); -0.55*** (0.000); -0.80*** (0.000); -0.87*** (0.000); -0.93*** (0.000); -0.76*** (0.000)
- External Liquidity: 0.14 (0.109); 0.08 (0.325); 0.09 (0.398); 0.00 (0.982); 0.10 (0.517); 0.05 (0.596)
- Domestic FCI: 0.48*** (0.000); 0.22** (0.067); -0.02 (0.914); -0.17 (0.331); -0.24 (0.203); -0.11 (0.420)
- Domestic Demand: -0.14 (0.254); -0.25*** (0.031); -0.36*** (0.005); -0.72*** (0.000); -0.59*** (0.000); -0.52*** (0.000)
- R^2: 0.49 0.50 0.49 0.50 0.47 0.71

Notes:
- First number = regression coefficient; number in parenthesis = p-value.
- Asterisks *, **, and *** indicate statistical significance at 5, 10, and 1 percent levels, respectively.

### Key empirical findings (term- and state-dependence)
- Domestic leverage partition:
  - Has the strongest and always significant impact on growth in the Dominican Republic at all three horizons.
  - Negative coefficient indicates higher availability of private sector credit is associated with higher economic growth in the following year; a credit crunch would have a strong and prolonged negative impact on domestic growth.
  - Private sector credit is the most important and persistent driver of domestic growth at all states and horizons.
- External demand:
  - Second most important factor for growth; strongest impact at medium- and long-term horizons.
  - Favorable external demand yields strong positive impact; decline in external demand yields negative impact.
  - Transmission channel always significant at h=8 and h=12; at h=4 significant only when growth is below the sample average (q=0.1, 0.25, and 0.5).
- Domestic financial conditions (Domestic FCI):
  - Significant impact only when growth is below average.
  - Shows a sign switch: expansionary in the short run (h=4) but can amplify vulnerabilities in the long run (h=12).
  - Important for q=0.1 and q=0.25 percentiles and for the mean at h=4 and h=12.
  - Reflects monetary policy potency through the interest rate-investment channel given limited deep stock markets.
- External borrowing costs and external liquidity:
  - External borrowing costs: no short-term impact but significant and negative at longer horizons (h=12) across all points of growth distribution; at h=8 significant only in times of higher-than-average growth; in the near term impact positive but significant only for the mean.
  - External liquidity: no significant marginal impact on domestic economic growth at any horizon.
- Domestic demand autoregressive component:
  - A shock indicating current growth rate leads to lower future growth; such shocks may reflect negative structural shifts (natural disaster, population aging, resource constraints).

### Scenario analysis — baseline and shock impacts
- Distribution fitting:
  - Quantile regressions used to fit a probability density for GDP growth using a skewed generalized T distribution; mode conditional on IMF growth forecasts.
  - Baseline (end-2018): 5 (10) percent probability that growth in 2019 could be as low as 4.65 (4.85) percent; probability of negative growth close to zero.
- Shock methodology:
  - Scenario tests compare one-standard-deviation negative shocks to each partition, independently, and analyze shifts in growth distributions at h=4 and h=12.
- Domestic leverage shock:
  - Short run: negative shock reduces modal growth by almost 3 percent (from 7.2 to 4.3 percent) without risk of negative future growth.
  - Long run: same shock decreases modal growth forecast by 4 percent (from 7.2 to 4.3 percent), leading to decline in activity with a sizable probability of 6.6 percent (probability of growth below zero).
- Domestic demand negative shock:
  - Could decrease modal forecast by 2.4 percent at h=4 and by 3 percent at h=12.
  - In the worst 5 percent scenarios, growth could fall to 3.4 percent in the short horizon, with risks of sluggish growth (0.5 percent) in the longer term.
- Domestic FCI shock:
  - Significant only in bad times; reverses impact from negative at h=4 to positive at h=12.
  - Short term: tighter domestic FCI could reduce modal growth by almost 2 percent (from 7.2 to 5.1 percent) and increase probability of negative growth to 2.5 percent.
  - Long term: mode improves to 5.6 percent; very small impact on other distribution parameters.
  - Implies monetary policy effective in staving off slowdowns; tightening should be done only in response to upward inflationary pressures.
- External demand shock:
  - Deterioration in external demand would be most detrimental among external factors, especially at longer horizons.
  - Negative shock can shift growth distribution strongly left, increasing probability of negative growth to around 4-5 percent.
- External borrowing cost shock:
  - Higher external borrowing costs reduce domestic growth prospects only at a three-year horizon due to lagged transmission.
- External liquidity:
  - Does not have a significant short-term impact given relatively weak global integration of the Dominican financial sector and sufficient local market liquidity.

### Policy implications and recommendations
- Domestic credit and macroprudential policy:
  - Private sector credit is the key driver; a negative shock to credit availability would strongly and persistently reduce growth.
  - Macroprudential policies aimed at credit quality should help keep NPLs low and maintain leverage necessary to close the negative credit gap and deepen domestic financial markets.
- Monetary policy:
  - Central bank has significant power to stimulate the economy in bad times but should be cautious of negative feedback two years later.
  - Monetary expansion in “bad” times: stimulus at one-year horizon, contractionary impact at three-year horizon — tradeoff between short-run stimulus and long-run vulnerabilities.
  - Decisions to tighten should be carefully analyzed and made only in response to upward inflationary pressures.
- External vigilance:
  - Authorities should be cautious of low growth prospects in the US and the euro area, worsening terms of trade, and competitiveness.
  - External borrowing costs matter mainly at long horizons; external liquidity has limited marginal effect.
- Complementary policies:
  - While monetary policy is important, other instruments matter: creating fiscal space, accumulating adequate foreign reserves, and building liquidity and capital buffers in the banking system in “good” times to cushion potential downturns.

*Source: wpiea2019246-print-pdf - 3.     Long-term horizon (h=12)*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019246-print-pdf.pdf_