## _wp15260 — 1. Literature for the Dominican Republic

## Source details

**Canonical URL:** [_wp15260 — 1. Literature for the Dominican Republic](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15260.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15260.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15260.pdf.json)

---

### Introduction and objectives
- Purpose: Estimate interest rate pass-through for retail rates in the Dominican Republic and test for asymmetries in adjustment to the monetary policy rate.
- Key motivations:
  - Effective monetary policy relies on a well-functioning transmission mechanism.
  - Dominican Republic adopted an inflation targeting regime; rapid and full reflection of policy rate changes in retail rates is important to affect inflation.
- Main contributions:
  - Results for retail rates at different maturities for June 2006–June 2015.
  - Some analysis of fiscal costs (public sector borrowing costs).
  - Tests for asymmetries in adjustment and simulations of symmetric and asymmetric adjustment paths.
- Principal summarized findings:
  - Evidence of complete pass-through to retail rates.
  - Government domestic bond yields increase when monetary policy becomes more contractionary.
  - Faster pass-through to lending rates than to deposit rates.
  - Short-term deposit rates respond faster to policy rate cuts than to hikes.
  - Short-term lending rates respond faster to policy rate hikes than to cuts.
  - Asymmetries theoretically associated with collusive market behaviors; measures to boost competition could help achieve symmetric adjustment.

### Theoretical reasons for stickiness and asymmetry (selected mechanisms)
- Asymmetric information and credit rationing (Stiglitz and Weiss, 1981): upward rigidity of lending rates possible when default risk increases.
- Financial frictions and financial accelerator (Bernanke et al., 1996; Gertler and Karadi, 2011; Gerali et al., 2010): balance-sheet deterioration can induce overpass-through.
- Overpass-through due to asymmetric information (De Bondt, 2005): banks may raise interest rates beyond policy increases to compensate for higher default risk.
- Menu costs (Rotemberg and Saloner, 1987) and switching costs (Lowe and Rohling, 1992): can delay or make asymmetric retail rate changes.
- Risk-sharing, implicit contracts, ownership and macro conditions: state-owned banks and high volatility can alter pass-through speed.
- Oligopolistic / collusive behavior (Hannan and Berger, 1991): predicts deposit rates rigid upwards after policy hikes and lending rates rigid downwards after policy cuts (or vice versa under alternative hypotheses).

### Dominican financial system context (as of June 2015)
- System size and composition:
  - 65 financial entities with assets over 45 percent of GDP and a loan portfolio of 27 percent of GDP.
  - Loan composition: commercial loans 58.7 percent, personal consumption loans 24 percent, mortgage loans 17.3 percent.
  - Financial system breakdown: 17 commercial banks (85.8 percent of system assets), 10 saving and credit institutions (11.1 percent), 19 credit unions (1.9 percent), credit corporations (0.3 percent), development bank (0.9 percent).
- Concentration trends:
  - June 2006: three main banks represented 57.5 percent of deposits and 58.2 percent of loans.
  - June 2015: three main banks represented 67.7 percent of deposits and 69.0 percent of loans.
- Spreads and margins:
  - Spreads are high at all maturities: 6.5 percent on average for three-month spread over 2006-15.
  - Spreads vary significantly when monetary policy rate changes.

### Empirical strategy, data and definitions
- Core model: retail rate at time t = intercept (markup/markdown) + β*(monetary policy rate) + vector of additional factors + error.
- Exogenous covariates:
  - Reserve requirement coefficient for commercial bank deposits in local currency.
  - Non-performing loans (NPLs) as share of total loans.
  - EMBI spread (time-varying risk premium).
  - Volatility Index (VIX).
  - Two regime dummies: inflation targeting period; period when monetary policy rate introduced as benchmark.
- Interpretation of β:
  - β close to zero implies weak transmission.
  - β = 1 implies complete pass-through.
  - β > 1 implies overpass-through.
- Data coverage:
  - Monthly data over June 2006–June 2015.
  - Retail rates: deposit and lending rates at 90, 180, 360 days; weighted average rate; commercial, personal consumption, and mortgage lending rates of commercial banks.
  - Government domestic bond rates: 5, 7, 10 years and 10-year sovereign bonds (shorter samples for bonds: observations start May 2009 for five-year domestic, February 2010 for seven-year domestic, August 2010 for ten-year domestic, April 2010 for ten-year sovereign; all series end June 2015).
- Time series methodology:
  - ADF unit-root tests on levels and first differences.
  - If non-stationary and cointegrated with monetary policy rate: Engle-Granger and ECM.
  - If no cointegration: ADL model and derive long-run pass-through.
  - Tests for asymmetric cointegration using TAR and M-TAR (Enders and Siklos, 2001; Chan, 1993; Enders and Granger, 1998; Caner and Hansen, 1998).

### Baseline unit-root and cointegration results
- ADF tests (levels): cannot reject unit root at five percent for all series except the weighted average deposit rate.
- ADF tests (first differences): null of unit root rejected for all series.
- Engle-Granger cointegration test:
  - Null of no cointegration rejected for all retail rate combinations.
  - Cointegration not found for government bond rates (five-, seven-, ten-year domestic and ten-year external).

### Long-run pass-through — ECM long-run coefficients (selected, Table 2)
- Deposit rates (long-run monetary policy pass-through coefficients):
  - Three months: 1.098***
  - Six months: 1.115***
  - One year: 1.018***
  - Weighted average deposit: 1.036***
- Lending rates (long-run monetary policy pass-through coefficients):
  - Three months lending: 1.076***
  - Six months lending: 1.175***
  - One year lending: 1.169***
  - Weighted average lending: 1.164***
  - Commercial: 1.186***
  - Personal consumption: 1.014***
  - Mortgage: 0.966***
- Wald tests: cannot reject H0: pass-through = 1 in all reported cases.

### Short-run pass-through and speed of adjustment (ECM short-run equation, selected)
- Short-run impact (within same month) of a one percentage point change in the monetary policy rate:
  - Deposit rates: 0.5 to 0.7 percentage points (examples: three-month deposit 0.668***; six-month deposit 0.641***; one-year deposit 0.474***).
  - Lending rates: 0.5 to 0.8 percentage points (examples: three-month lending 0.528**; weighted average lending 0.781***); six-month lending short-run impact 1.025*** (noted as abnormally high).
- Speed of adjustment (error-correction coefficients, examples):
  - Three-month deposit: -0.417***
  - Six-month deposit: -0.232***
  - One-year deposit: -0.335***
  - Three-month lending: -0.534***
  - Six-month lending: -0.691***
- Summary interpretation:
  - Lending rates correct deviations from long-run equilibrium faster than deposit rates (relative speeds reported between -0.3 and -0.7 for lending; around -0.2 to -0.4 for deposits in examples).

### Explanatory variables — long-run and short-run effects (selected coefficients)
- EMBI spread (long-run): increase associated with higher deposit and lending rates across maturities and lending categories.
  - Examples: three-month deposit 0.407***; six-month deposit 0.329***; one-year deposit 0.22***; three-month lending 0.993***; commercial lending 0.729***.
- Reserve requirement coefficient:
  - Positive effect only on three-month lending (three-month lending: 0.4**) and not robust across maturities.
  - Three-month deposit: -0.092 (example).
- NPL ratio (long-run):
  - Three-month deposit: -0.401***; six-month deposit: -0.393**; weighted average deposit: -0.329**.
  - Three-month lending: 0.358 (example, not always significant).
  - Interpretation: positive change in NPL associated with fall in deposit rates and increases in lending rates in some specifications.
- VIX (examples):
  - Three-month deposit: -0.025* (long-run)
  - Three-month lending: -0.13***
  - Commercial lending: -0.053*
  - Short-run: Diff. VIX impacts mostly small and not robust except raising the change in personal consumption rate by a minor amount.

### ECM model fit and diagnostics (selected)
- Observations: 109 for long-run equations; 108 for short-run equations.
- R-squared (long-run) examples: 0.862 (three months), 0.859 (six months), 0.847 (one year), 0.871 (weighted average).
- R-squared (short-run) examples: 0.394 (three months), 0.391 (six months), 0.296 (one year), 0.429 (weighted average).
- Wald test (H0: pass-through = 1) examples: 1.277 (three months), 1.048 (six months), 0.035 (one year), 0.176 (weighted average).
- Standard errors: Newey-West corrected. ***, **, * indicate significance at 1, 5 and 10 percent, respectively.

### ADL estimation for government bond rates (Table 3) — pass-through to bonds
- Short-run impact of a one percentage point change in the monetary policy rate on domestically-issued bond rates: short-run coefficient range 0.5-0.6 percentage points depending on maturity.
  - Examples: Diff. monetary policy rate: 0.567*** (domestic five years), 0.48*** (seven years), 0.447*** (ten years).
- Long-run coefficient (retrieved via lagged differenced dependent variable): range between 0.8 and 0.9 percentage points.
- Interpretation: Pass-through to domestically-issued government bond rates appears less than complete.
- ADL selected coefficients:
  - Lag diff. monetary policy rate: 0.554*** (five years), 0.759*** (seven years), 0.463** (ten years).
  - Lag diff. dependent variable: -0.493***, -0.325**, -0.210 (for five-, seven-, ten-year examples).
  - Diff. EMBI: 0.945**, 0.680, 0.473.
  - Diff. VIX: -0.074*, -0.019, -0.043**.
- Observations and fit examples:
  - Observations: 544 (external five years), 157 (seven years), 61 (ten years).
  - R-squared examples: 0.398 (external five years), 0.341 (seven years), 0.255 (ten years), 0.682 (domestic ten years).

### Tests for asymmetric cointegration and asymmetric ECM (TAR, M-TAR results)
- TAR and M-TAR tests with threshold set to zero (Table A3):
  - TAR Φ statistics show cointegration for many retail series (examples: three-month deposit TAR Φ = 14.588***; six-month deposit TAR Φ = 15.649***; one-year lending TAR Φ = 20.237***).
  - TAR F statistics often do not reject symmetric cointegration when threshold set to zero (examples: three-month deposit TAR F = 0.032; weighted average lending TAR F = 0.015).
- TAR and M-TAR tests with endogenously determined threshold (Table A4):
  - TAR finds evidence of asymmetric cointegration for: weighted average deposit rate, three-month, six-month, and one-year lending rates, and the mortgage lending rate (TAR F significant in several cases).
  - M-TAR rejects symmetric cointegration for most series (M-TAR F significant in numerous cases).
  - Examples of estimated thresholds (τ): three-month deposit TAR τ = -0.613; three-month deposit M-TAR τ = 0.732; three-month lending TAR τ = -1.292; three-month lending M-TAR τ = -0.402.
- Asymmetric ECM (M-TAR) — speeds of adjustment above τ vs below τ (selected, Table 4):
  - Three-month deposit:
    - Speed above τ: -0.493**
    - Speed below τ: -0.361***
    - Wald test (H0: speeds equal): 6.423**
    - Interpretation: speed above τ is 1.4 times speed below τ; negative deviations more persistent.
  - Three-month lending:
    - Speed above τ: -0.372***
    - Speed below τ: -0.789***
    - Wald test: 7.311***
    - Interpretation: speed below τ is about twice speed above τ; positive deviations more persistent.
  - Higher maturities: the two speed coefficients are typically not significantly different.

### Simulation: one percentage point monetary policy tightening
- Policy experiment: one percentage point tightening in monetary policy rate starting from equilibrium; simulate adjustment over following year.
- Symmetric adjustment outcomes (summarized from Figure 3):
  - Completed changes in spreads between 0.0 and 0.2 percentage points across maturities.
  - On impact, increase in policy rate generates a negative spread of about 0.1 percentage points for three-month rates; spread approaches zero after one month due to faster lending-rate adjustment.
  - Under symmetric adjustment, a monetary policy easing implies a reduction in banks’ profitability by the same amount during the first month.
- Loan-category timing:
  - Personal consumption and commercial loans incorporate policy changes fastest; personal consumption has lower pass-through but faster completion.
  - Mortgage loans show a small short-run effect but similar speed of convergence to long-run equilibrium as commercial loans.
- Asymmetric adjustment path for three-month rates (by construction short- and long-run impacts identical; different speeds imply):
  - Deposits: deviations above τ (policy rate cuts) corrected faster than deviations below τ (policy rate hikes).
  - Loans: deviations below τ (policy rate hikes) corrected faster than deviations above τ (policy rate cuts).

### Conclusions and policy implications
- Data and scope: monthly sample June 2006–June 2015; deposit and lending rates at different maturities, loan-category rates, and government bond rates.
- Main empirical conclusions:
  - Pass-through to deposit and lending rates is complete (long-run coefficients around or above 1.0).
  - Government domestic bond rates respond to monetary policy changes but pass-through to bonds is less than complete (long-run coefficients between 0.8 and 0.9).
  - Pass-through to lending rates generally faster than to deposit rates.
  - Robust evidence of asymmetric adjustment in short-term retail rates:
    - Short-term deposit rates respond faster to policy rate cuts than to hikes.
    - Short-term lending rates respond faster to policy rate hikes than to cuts.
- Policy implication:
  - Collusive market hypothesis (bank bargaining power) can explain observed asymmetries—lending rates show downward rigidity to policy cuts and upward flexibility to hikes.
  - Measures to reduce bank concentration and boost competition could enhance monetary policy transmission by reducing asymmetries in retail rate adjustments.

### Appendix — selected unit-root and cointegration test values (Tables A1–A4)
- ADF (first differences): monetary policy rate -8.510***, three-month lending first differences -11.970***, six-month lending first differences -15.009***, weighted average deposit first differences -5.214***, ten-year government external bond first differences -6.451*** (examples).
- Engle-Granger cointegration test (Table A2) examples:
  - Three-month deposit rate: -5.489***
  - Six-month lending rate: -8.774***
  - Mortage lending rate: -7.324***
  - Five-year government domestic bond: -3.640 (no cointegration rejection)
- TAR/M-TAR with threshold set to zero (Table A3) example statistics:
  - Three-month deposit rate TAR Φ = 14.588*** ; TAR F = 0.032 ; M-TAR Φ = 14.648*** ; M-TAR F = 0.124
  - One-year lending rate TAR Φ = 20.237*** ; TAR F = 0.141 ; M-TAR Φ = 3.605 ; M-TAR F = 0.480
- TAR/M-TAR with endogenously determined threshold (Table A4) example entries:
  - Three-month deposit TAR Φ = 15.165*** ; TAR F = 0.929 ; TAR τ = -0.613 ; M-TAR Φ = 16.390*** ; M-TAR F = 2.835*** ; M-TAR τ = 0.732
  - One-year lending TAR Φ = 21.470*** ; TAR F = 1.908*** ; TAR τ = 1.188 ; M-TAR Φ = 23.321*** ; M-TAR F = 4.562*** ; M-TAR τ = 0.827

*Source: _wp15260 - 1. Literature for the Dominican Republic (PDF).*

### 1. Literature for the Dominican Republic ...............................................................................

### _wp15260 - 1. Literature for the Dominican Republic ...............................................................................

### Major sections listed
- 1. Literature for the Dominican Republic ................................................................................10
- 2. ECM Estimation for Retail Rates ........................................................................................18
- 3. ADL Model Estimation for Government Bonds ..................................................................19
- 4. Asymmetric M-TAR ECM Estimation for Retail Rates ......................................................21

### Appendices and tests
- A1. Unit Root, ADF Test .........................................................................................................24
- A2. Cointegration, Engle-Granger Test ................................................................................24
- A3. Cointegration, TAR and M-TAR Tests with Threshold Set to Zero ................................25
- A4. Cointegration, TAR and M-TAR Tests with Endogenously Determined Threshold .......25

### Figures listed
- Figure 1. Monetary Policy and Retail Rates..........................................................................................8
- Figure 2. Banks’ Margins ......................................................................................................................9
- Figure 3. Timing of One-Percentage Point Change in the Monetary Policy Rate ..............................22

*Source: _wp15260 - 1. Literature for the Dominican Republic (PDF).*

### References .............................................................................................................

### _wp15260 - References .............................................................................................................

### Introduction
- Purpose: Estimate interest rate pass-through for retail rates in the Dominican Republic and test for asymmetries in adjustment to the monetary policy rate.
- Key motivations:
  - Effective monetary policy relies on a well-functioning transmission mechanism (Mishkin, 1995; Becker et al., 2012).
  - Dominican Republic adopted an inflation targeting regime; rapid and full reflection of policy rate changes in retail rates is important to affect inflation.
- Main contributions:
  - Provide results for retail rates at different maturities for a recent period, including some analysis of fiscal costs (public sector borrowing costs).
  - Test for asymmetries in adjustment and simulate symmetric and asymmetric adjustment of retail rates to changes in the monetary policy rate.
- Principal findings (summarized in the paper):
  - Evidence of complete pass-through to retail rates.
  - Government domestic bond yields increase when monetary policy becomes more contractionary.
  - Faster pass-through to lending rates than to deposit rates.
  - Short-term deposit rates respond faster to policy rate cuts than to hikes.
  - Short-term lending rates respond faster to policy rate hikes than to cuts.
  - Asymmetries in speed of adjustment theoretically associated with collusive market behaviors; measures to boost competition could help achieve symmetric adjustment.

### Theoretical contributions (reasons for stickiness and asymmetry)
- Asymmetric information and credit rationing (Stiglitz and Weiss, 1981):
  - Banks may not raise lending rates despite higher funding costs if loan default risk increases, leading to upward rigidity and asymmetric adjustment.
- Financial frictions and financial accelerator (Bernanke et al., 1996; Gertler and Karadi, 2011; Gerali et al., 2010):
  - Balance sheet deterioration in response to policy hikes can tighten financing and induce overpass-through.
- Overpass-through due to asymmetric information (De Bondt, 2005):
  - Banks may raise interest rates beyond the policy rate increase to compensate for higher default risk.
- Menu costs (Rotemberg and Saloner, 1987):
  - Firms change prices only when benefits exceed costs; small/temporary policy changes may be delayed.
- Switching costs (Lowe and Rohling, 1992):
  - High customer switching costs can induce retail rate stickiness and asymmetric adjustment.
- Risk-sharing, implicit contracts, and long-term relationships (Fried and Howitt, 1980; Berger and Udell, 1992):
  - Risk-sharing agreements and implicit contracts contribute to stickier retail rates.
- Ownership and macro conditions:
  - State-owned banks may adjust rates with delay due to political aims.
  - High volatility reduces information content of policy signals; pass-through faster during high inflation and high growth periods (Égert et al., 2007; Égert and Mac Donald, 2009).
- Oligopolistic / collusive behavior and asymmetric hypotheses (Hannan and Berger, 1991):
  - Collusive behavior: deposit rates rigid upwards after policy hikes; lending rates downward-rigid after policy cuts.
  - Adverse customer reaction: deposit rates rigid downwards when policy falls; lending rates rigid upwards when policy rises.

### Characteristics of the Dominican financial system and empirical evidence
- Institutional and policy timeline:
  - Monetary and Financial Law of 2002 established price stability as the main central bank mandate.
  - From 2004: moved from exchange rate targeting to monetary targeting; introduced overnight deposit window and Lombard facility creating a corridor; overnight rate served as monetary stance signal.
  - January 2012: BCRD adopted inflation targeting regime.
  - February 2013: monetary policy rate introduced as benchmark (middle of corridor).
- Financial system size and composition (as of June 2015):
  - System composed of 65 financial entities with assets over 45 percent of GDP and a loan portfolio of 27 percent of GDP.
  - Loan composition: commercial loans 58.7 percent, personal consumption loans 24 percent, mortgage loans 17.3 percent.
  - Financial system breakdown: 17 commercial banks (85.8 percent of system assets), 10 saving and credit institutions (11.1 percent), 19 credit unions (1.9 percent), credit corporations (0.3 percent), development bank (0.9 percent).
- Concentration and trends:
  - June 2006: three main banks represented 57.5 percent of deposits and 58.2 percent of loans.
  - June 2015: three main banks represented 67.7 percent of deposits and 69.0 percent of loans.
- Spreads and margins:
  - Spreads are high at all maturities (6.5 percent on average for three-month spread over 2006-15) and vary significantly when monetary policy rate changes.
  - Implication: deposit and lending rates may react differently or asymmetrically to policy rate changes.
- Summary of prior empirical literature for the Dominican Republic (selected results presented in the paper):
  - González Pantaleón (2010): ECM; Loans: 0.96-1.44; Deposits: 0.91-1.23; explanatory variables include interbank rate, external interest rate, EMBI spread, reserve requirement coefficient, Lombard window rate.
  - Rivas (2011): Asymmetric ECM; Loans: 0.90-1.42; Deposits: 1.00; interbank rate.
  - Medina, Carrión, and Frantischek (2011): Panel S-GMM (2004–10) for 40 countries; Loans (DR): 0.65; explanatory variables include monetary policy rate, financial dollarization, exchange rate flexibility, size of banking sector, banking concentration.
  - Andujar (2012): ECM; Loans: 0.94; Deposits: 0.73-0.77; interbank rate.
  - Aristy Escuder (2014): ADL model; Loans: 2.66; Deposits: 1.65; monetary policy rate (overnight).
  - BCRD (2015): Loans: 1.06-1.63; Deposits: 0.85-0.92; reported speeds of adjustment in months: generally between one and two months for all interest rates except six-month lending rate adjusts in about four months.
- Heterogeneity in past results:
  - Most literature finds overpass-through to lending rates, with significant variation.
  - Deposit rate estimates range from incomplete to overpass-through.

### Empirical strategy
- Core model: retail rate at time t modeled as a function of an intercept (markup/markdown), monetary policy rate, vector of additional factors, and stochastic error term.
- Vector of exogenous covariates includes:
  - Reserve requirement coefficient for commercial bank deposits in local currency.
  - Non-performing loans (NPLs) as a share of total loans.
  - EMBI spread (time-varying risk premium).
  - Volatility Index (VIX) (external market volatility expectation for next 30 days).
  - Two regime dummies: inflation targeting period; period when monetary policy rate introduced as benchmark.
- Long-run pass-through parameter of interest: β, where β close to zero implies weak transmission and β = 1 implies complete pass-through; β > 1 implies overpass-through.
- Data and coverage:
  - Monthly data over June 2006-June 2015.
  - Retail rates analyzed: deposit and lending rates at 90, 180, 360 days; weighted average rate; commercial, personal consumption, and mortgage lending rates of commercial banks.
  - Government domestic bond rates analyzed at 5, 7, 10 years and 10-year sovereign bonds in the sovereign market (secondary market used; shorter sample with some missing observations: observations start in May 2009 for five-year domestic bonds, February 2010 for seven-year domestic bonds, August 2010 for ten-year domestic bonds, and April 2010 for ten-year sovereign bonds; all series end in June 2015; results for these instruments interpreted with caution).
- Time series methodology:
  - Test series for unit roots using Augmented Dickey Fuller (ADF) on levels and first differences.
  - If non-stationary and cointegrated with the monetary policy rate, use Engle-Granger test and estimate error correction model (ECM).
  - If no cointegration, estimate autoregressive distributed lag (ADL) model and derive long-run pass-through.
  - Test for asymmetric cointegration using TAR and M-TAR (Enders and Siklos, 2001; Chan, 1993; Enders and Granger, 1998; Caner and Hansen, 1998), including tests (F and t-max) and estimation of asymmetric speed of adjustment parameters.

### Results — Baseline
- Unit root and cointegration tests:
  - ADF on levels (constant and trend): cannot reject unit root at five percent for all series except the weighted average deposit rate.
  - After first-differencing: null of unit root safely rejected for all series.
  - Engle-Granger cointegration test: null of no cointegration rejected for all retail rate combinations except government bond rates, indicating long-run relationship between policy rate and retail rates but not government bond rates.
- Long-run pass-through (ECM estimates; Table 2 summarized):
  - Deposit rates: long-run coefficient ranges between 1.0 and 1.1 (complete pass-through).
  - Lending rates: long-run coefficient ranges between 1.1 and 1.2 (complete pass-through).
  - Wald tests: cannot reject the null hypothesis that long-run pass-through coefficient equals one in all cases.
- Explanatory variables (significance and direction):
  - EMBI spread: increase is unequivocally reflected in higher deposit and lending rates across maturities and lending categories.
  - Reserve requirement coefficient: positive effect only on three-month lending rates (consistent with liquidity withdrawal).
  - NPL ratio: increase associated with lower deposit rates (possibly reflecting banks’ need to find liquidity); not significant for one-year deposit regressions.
  - VIX: some evidence of lower lending rates with heightened advanced-economies volatility, but effect relatively small and not robust across maturities.
- Short-run pass-through (from ECM, within the same month):
  - Deposit rates: short-run effect of a one percentage point increase in the change of the monetary policy rate on deposit rate change ranges between 0.5 and 0.7 percentage points (i.e., 50 to 70 percent transmitted within same month).
  - Lending rates: short-run impact ranges between 0.5 and 0.8 percentage points, except the six-month lending rate where short-run impact is 1.0 percentage points (noted as abnormally high and to be treated with caution).
  - The only maturity where short-run impact on lending rate exceeds that on deposit rate is the three-month maturity, implying a short-run decrease in the spread.
- Speed of adjustment (error-correction coefficient):
  - Lending rates: relative speed of adjustment ranges between -0.3 and -0.7 (deviations from long-run equilibrium corrected more quickly).
  - Deposits: relative speed of adjustment observed at -0.2 (the source text truncates the full reported range).

*Source: _wp15260 - References .............................................................................................................*

### 0.4 for deposit rates.

### _wp15260 - 0.4 for deposit rates.

### Explanatory variables — effects on retail and bond rates
- Changes in the EMBI spread are associated with increases in deposit rates and most lending rates.
- Changes in the reserve requirement increase deposit and lending rates, but these effects:
  - are not robust across maturities, and
  - are not significant for lending categories.
- A positive change in the NPL ratio is associated with:
  - a fall in deposit rates, and
  - increases in lending rates.
  - The finding is not robust for all rates but underscores banks’ need to find liquidity when more loans become non-performing.
- Changes in the VIX index are significant only in raising the change in the personal consumption rate and by a minor amount.

### ECM estimation for retail rates — key coefficients and diagnostics (Table 2)
- Long-run monetary policy pass-through coefficients (Monetary policy rate) by series:
  - Three months: 1.098***
  - Six months: 1.115***
  - One year: 1.018***
  - Weighted average deposit: 1.036***
  - Three months lending: 1.076***
  - Six months lending: 1.175***
  - One year lending: 1.169***
  - Weighted average lending: 1.164***
  - Commercial: 1.186***
  - Personal consumption: 1.014***
  - Mortgage: 0.966***
- Reserve requirement coefficient examples:
  - Three-month deposit: -0.092
  - Three-month lending: 0.4**
  - Mortgage: 0.202
- NPLs (long-run):
  - Three-month deposit: -0.401***
  - Six-month deposit: -0.393**
  - Weighted average deposit: -0.329**
  - Three-month lending: 0.358
  - Personal consumption: -0.212
- EMBI (long-run) examples:
  - Three-month deposit: 0.407***
  - Six-month deposit: 0.329***
  - One-year deposit: 0.22***
  - Three-month lending: 0.993***
  - Commercial lending: 0.729***
- VIX (long-run) examples:
  - Three-month deposit: -0.025*
  - Three-month lending: -0.13***
  - Commercial lending: -0.053*
- Short-run dynamics (short-run equation, selected coefficients):
  - Speed of adjustment (error-correction term) examples:
    - Three-month deposit: -0.417***
    - Six-month deposit: -0.232***
    - One-year deposit: -0.335***
    - Three-month lending: -0.534***
    - Six-month lending: -0.691***
  - Diff. monetary policy rate (short-run impact) examples:
    - Three-month deposit: 0.668***
    - Six-month deposit: 0.641***
    - One-year deposit: 0.474***
    - Three-month lending: 0.528**
    - Six-month lending: 1.025***
    - Weighted average lending: 0.781***
  - Diff. EMBI (short-run) examples:
    - Three-month deposit: 0.278***
    - Six-month deposit: 0.234**
    - One-year deposit: 0.199**
    - Three-month lending: 0.663***
- Model fit and diagnostics (examples from Table 2):
  - Observations: 109 for long-run equations; 108 for short-run equations.
  - R-squared (long-run) examples: 0.862 (three months), 0.859 (six months), 0.847 (one year), 0.871 (weighted average).
  - R-squared (short-run) examples: 0.394 (three months), 0.391 (six months), 0.296 (one year), 0.429 (weighted average).
  - Wald test (H0: pass-through = 1) examples: 1.277 (three months), 1.048 (six months), 0.035 (one year), 0.176 (weighted average).  
- Notes: Standard errors are Newey-West corrected. ***, **, * indicate significance at 1, 5 and 10 percent, respectively.

### ADL estimation for government bond rates (Table 3) — pass-through to bonds
- Short-run impact of a one percentage point change in the monetary policy rate on domestically-issued bond rates:
  - Short-run coefficient range: 0.5-0.6 percentage points depending on maturity.
- Long-run coefficient retrieved via lagged differenced dependent variable:
  - Long-run coefficient range: between 0.8 and 0.9 percentage points.
- Interpretation: Pass-through to domestically-issued government bond rates appears less than complete.
- Selected ADL coefficients (Domestic five-, seven-, ten-year):
  - Diff. monetary policy rate: 0.567*** (five years), 0.48*** (seven years), 0.447*** (ten years)
  - Lag diff. monetary policy rate: 0.554***, 0.759***, 0.463**
  - Lag diff. dependent variable: -0.493***, -0.325**, -0.210
  - Diff. EMBI: 0.945**, 0.680, 0.473
  - Diff. VIX: -0.074*, -0.019, -0.043**
- Observations and fit:
  - Observations: 544 (external five years), 157 (seven years), 61 (ten years), Domestic ten years  ? (table shows mixed counts)
  - R-squared examples: 0.398 (external five years), 0.341 (seven years), 0.255 (ten years), 0.682 (domestic ten years).

### Asymmetries — tests and asymmetric ECM (TAR, M-TAR, Table 4)
- TAR and M-TAR tests:
  - With threshold set to zero: cannot reject null of symmetric cointegration for TAR; similar non-rejection for M-TAR.
  - With endogenously determined threshold:
    - TAR finds evidence of asymmetric cointegration for: weighted average deposit rate, three-month, six-month, and one-year lending rates, and the mortgage lending rate.
    - M-TAR rejects symmetric cointegration for all series except the six-month lending rate.
- Asymmetric ECM (M-TAR) — speed of adjustment above τ vs below τ (selected results, Table 4):
  - Three-month deposit:
    - Speed above τ: -0.493**
    - Speed below τ: -0.361***
    - Wald test (H0: speeds equal): 6.423**
    - Interpretation: For three-month deposit, speed of adjustment for positive deviations is 1.4 times that for negative deviations; negative deviations are more persistent.
  - Three-month lending:
    - Speed above τ: -0.372***
    - Speed below τ: -0.789***
    - Wald test: 7.311***
    - Interpretation: For three-month lending, speed of adjustment for negative deviations is two times that for positive deviations; positive deviations are more persistent.
  - For higher maturities: the two speed coefficients are not significantly different from each other.
- Short-run asymmetric coefficients (Diff. monetary policy rate examples):
  - Three-month deposit: 0.668*** (same as symmetric short-run impact)
  - Three-month lending: 0.553**
- Notes: Standard errors are Newey-West corrected; ***, **, * indicate significance at 1, 5 and 10 percent, respectively.

### Simulation of a one percentage point monetary policy tightening (Section C)
- Policy experiment:
  - A policy tightening of one percentage point in the monetary policy rate, starting from equilibrium, and simulating adjustment over the following year.
- Symmetric adjustment outcomes (Figure 3 summary):
  - Completed changes in spreads between 0.0 and 0.2 percentage points across maturities.
  - On impact, increase in monetary policy rate generates a negative spread of about 0.1 percentage points only for the three-month rates; spread approaches zero after one month due to faster lending-rate adjustment.
  - Because the simulated adjustment is symmetric, any monetary policy easing implies a reduction in banks’ profitability by the same amount during the first month.
- Loan-category timing:
  - Personal consumption and commercial loans are the fastest in incorporating monetary policy changes; personal consumption has lower pass-through so adjustment completes earlier despite lower speed.
  - Mortgage loans show a small short-run effect but similar speed of convergence to long-run equilibrium as commercial loans.
- Asymmetric adjustment path for three-month rates:
  - By construction short- and long-run impacts are identical regardless of deviation sign; different speeds imply:
    - Deposits: deviations above τ (policy rate cuts) are corrected faster than deviations below τ (policy rate hikes).
    - Loans: deviations below τ (policy rate hikes) are corrected faster than deviations above τ (policy rate cuts).

### Conclusions (Section V)
- Data and scope:
  - Ten-year monthly sample from 2006 to 2015 for the Dominican Republic.
  - Analysis includes deposit and lending rates at different maturities, loan category rates, and government bond rates.
- Main findings:
  - The pass-through to deposit and lending rates is complete, confirming effectiveness of the monetary policy transmission mechanism.
  - Government domestic bond rates react to monetary policy changes.
  - Pass-through to lending rates is generally faster than to deposit rates.
  - Evidence of asymmetric adjustment:
    - Short-term deposit rates respond faster to monetary policy rate cuts than hikes.
    - Short-term lending rates respond faster to monetary policy rate hikes than cuts.
- Policy implication:
  - The collusive market hypothesis (bargaining power of banks) can explain asymmetric adjustment — lending rates show downward rigidity to policy rate cuts and upward flexibility to hikes.
  - Measures to reduce bank concentration and boost competition could enhance the effectiveness of monetary policy transmission by reducing asymmetries in retail rate adjustments.

*Source: Authors' calculations.*

### Appendix II. Unit Root and Cointegration Test Results

### Appendix II. Unit Root and Cointegration Test Results

### Table A1. Unit Root, ADF Test — Levels and First differences
Notes: The null hypothesis is that the series has a unit root. The lagged differences are included in the specifications to obtain white noise residuals. The Schwartz Information Criterion is used to select the optimal lag length. ***, **, * next to a number indicate statistical significance at 1, 5 and 10 percent, respectively.

- Monetary policy rate: Levels -1.490 -2.266 -2.399 ; First differences -8.510*** -8.534*** -8.543***
- Three-month deposit rate: Levels -1.285 -2.910* -2.999 ; First differences -7.112*** -7.104*** -7.074***
- Six-month deposit rate: Levels -1.203 -3.212** -3.328* ; First differences -5.567*** -5.558*** -5.541***
- One-year deposit rate: Levels -1.331 -3.114** -3.275* ; First differences -6.546*** -6.540*** -6.526***
- Weighted average deposit rate: Levels -1.355 -3.537** -3.696** ; First differences -5.214*** -5.207*** -5.192***
- Three-month lending rate: Levels -1.056 -2.516 -2.712 ; First differences -11.970*** -11.938*** -11.907***
- Six-month lending rate: Levels -1.079 -2.242 -2.466 ; First differences -15.009*** -14.971*** -14.907***
- One-year lending rate: Levels -1.119 -2.442 -2.579 ; First differences -12.273*** -12.238*** -12.170***
- Weighted average lending rate: Levels -0.990 -2.314 -2.494 ; First differences -7.407*** -7.398*** -7.360***
- Commercial lending rate: Levels -0.968 -2.231 -2.369 ; First differences -7.576*** -7.561*** -7.531***
- Personal consumption lending rate: Levels -0.833 -2.468 -2.560 ; First differences -7.804*** -7.792*** -7.746***
- Mortgage lending rate: Levels -1.175 -2.286 -2.790 ; First differences -6.050*** -6.063*** -6.014***
- Five-year government domestic bond: Levels -1.962** -3.113** -3.414* ; First differences -12.655*** -13.009*** -12.787***
- Seven-year government domestic bond: Levels 2.002** -2.182 -2.577 ; First differences -6.808*** -7.053*** -6.966***
- Ten-year government domestic bond: Levels 1.458 -0.806 -2.000 ; First differences -7.556*** -7.696*** -7.643***
- Ten-year government external bond: Levels 1.416 -1.559 -2.024 ; First differences -6.451*** -6.534*** -6.497***

Source: Authors' calculations.

### Table A2. Cointegration, Engle-Granger Test
Notes: The null hypothesis is that the series do not have a cointegration relationship. The Schwartz Information Criterion is used to select the optimal lag length. The critical values are from MacKinnon (1991). ***, **, * next to a number indicate statistical significance at 1, 5 and 10 percent, respectively.

- Three-month deposit rate: -5.489***
- Six-month deposit rate: -5.531***
- One-year deposit rate: -4.928**
- Weighted average deposit rate: -6.185**
- Three-month lending rate: -6.187***
- Six-month lending rate: -8.774***
- One-year lending rate: -6.806***
- Weighted average lending rate: -6.741***
- Commercial lending rate: -6.373***
- Personal consumption lending rate: -6.158***
- Mortage lending rate: -7.324***
- Five-year government domestic bond: -3.640
- Seven-year government domestic bond: -3.224
- Ten-year government domestic bond: -3.094
- Ten-year government external bond: -3.040

Source: Authors' calculations.

### Table A3. Cointegration, TAR and M-TAR Tests with Threshold Set to Zero
Notes: The null hypothesis of the test statistic Φ is that the series do not have a cointegration relationship, and the null hypothesis of the test statistic F is that the series have symmetric cointegration relationships. The critical values are generated via Monte Carlo simulations. ***, **, * next to a number indicate statistical significance at 1, 5 and 10 percent, respectively.

- Three-month deposit rate: TAR Φ = 14.588*** ; TAR F = 0.032 ; M-TAR Φ = 14.648*** ; M-TAR F = 0.124
- Six-month deposit rate: TAR Φ = 15.649*** ; TAR F = 0.062 ; M-TAR Φ = 16.393*** ; M-TAR F = 1.202
- One-year deposit rate: TAR Φ = 3.406 ; TAR F = 0.111 ; M-TAR Φ = 3.606 ; M-TAR F = 0.480
- Weighted average deposit rate: TAR Φ = 18.225*** ; TAR F = 0.617 ; M-TAR Φ = 18.723*** ; M-TAR F = 1.355
- Three-month lending rate: TAR Φ = 5.116* ; TAR F = 0.481 ; M-TAR Φ = 6.975** ; M-TAR F = 3.823
- Six-month lending rate: TAR Φ = 8.884** ; TAR F = 0.996 ; M-TAR Φ = 16.393*** ; M-TAR F = 1.202
- One-year lending rate: TAR Φ = 20.237*** ; TAR F = 0.141 ; M-TAR Φ = 3.605 ; M-TAR F = 0.480
- Weighted average lending rate: TAR Φ = 17.890*** ; TAR F = 0.015 ; M-TAR Φ = 18.055*** ; M-TAR F = 0.260
- Commercial lending rate: TAR Φ = 19.294*** ; TAR F = 0.048 ; M-TAR Φ = 19.418*** ; M-TAR F = 0.228
- Personal consumption lending rate: TAR Φ = 2.211 ; TAR F = 0.018 ; M-TAR Φ = 2.276 ; M-TAR F = 0.143
- Mortage lending rate: TAR Φ = 17.346*** ; TAR F = 1.393* ; M-TAR Φ = 18.007*** ; M-TAR F = 2.392

Source: Authors' calculations.

### Table A4. Cointegration, TAR and M-TAR Tests with Endogenously Determined Threshold
Notes: The null hypothesis of the test statistic Φ is that the series do not have a cointegration relationship, and the null hypothesis of the test statistic F is that the series have symmetric cointegration relationships. τ is the estimated threshold. The critical values are generated via Monte Carlo simulations. ***, **, * next to a number indicate statistical significance at 1, 5 and 10 percent, respectively.

- Three-month deposit rate:
  - TAR Φ = 15.165*** ; TAR F = 0.929 ; TAR τ = -0.613
  - M-TAR Φ = 16.390*** ; M-TAR F = 2.835*** ; M-TAR τ = 0.732
- Six-month deposit rate:
  - TAR Φ = 15.848*** ; TAR F = 0.367 ; TAR τ = -0.342
  - M-TAR Φ = 17.084*** ; M-TAR F = 2.260*** ; M-TAR τ = -0.316
- One-year deposit rate:
  - TAR Φ = 3.649 ; TAR F = 0.543 ; TAR τ = -0.727
  - M-TAR Φ = 5.338* ; M-TAR F = 3.694*** ; M-TAR τ = 0.089
- Weighted average deposit rate:
  - TAR Φ = 19.058*** ; TAR F = 1.851*** ; TAR τ = -0.525
  - M-TAR Φ = 19.848*** ; M-TAR F = 3.022*** ; M-TAR τ = 0.341
- Three-month lending rate:
  - TAR Φ = 6.488** ; TAR F = 2.948*** ; TAR τ = -1.292
  - M-TAR Φ = 7.932** ; M-TAR F = 5.543*** ; M-TAR τ = -0.402
- Six-month lending rate:
  - TAR Φ = 9.334*** ; TAR F = 1.752*** ; TAR τ = 0.505
  - M-TAR Φ = 8.806** ; M-TAR F = 0.865 ; M-TAR τ = 1.417
- One-year lending rate:
  - TAR Φ = 21.470*** ; TAR F = 1.908*** ; TAR τ = 1.188
  - M-TAR Φ = 23.321*** ; M-TAR F = 4.562*** ; M-TAR τ = 0.827
- Weighted average lending rate:
  - TAR Φ = 18.243*** ; TAR F = 0.538 ; TAR τ = 0.853
  - M-TAR Φ = 19.547*** ; M-TAR F = 2.469*** ; M-TAR τ = -0.577
- Commercial lending rate:
  - TAR Φ = 19.916*** ; TAR F = 0.951 ; TAR τ = -0.889
  - M-TAR Φ = 20.300*** ; M-TAR F = 1.508** ; M-TAR τ = -0.73
- Personal consumption lending rate:
  - TAR Φ = 2.295 ; TAR F = 0.178 ; TAR τ = -1.039
  - M-TAR Φ = 4.136 ; M-TAR F = 3.682*** ; M-TAR τ = 0.082
- Mortage lending rate:
  - TAR Φ = 18.416*** ; TAR F = 3.012*** ; TAR τ = -0.583
  - M-TAR Φ = 19.092*** ; M-TAR F = 4.033*** ; M-TAR τ = -0.024

Source: Authors' calculations.

*Source: Authors' calculations.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15260.pdf_