## _wp04211

## Source details

**Canonical URL:** [_wp04211](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04211.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04211.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04211.pdf.json)

---

### I. Objective and context
- Objective: Test hypothesis that the interest rate pass-through from policy to market rates plays a lesser role in Romania than in other transition economies in the region.
- Contextual factors considered:
  - Shared factors across transition economies: low degree of monetization, underdeveloped financial markets, capital controls.
  - Romanian-specific features: strong and consistent increase in structural excess liquidity starting in 1997; commercial banks hold substantial deposits over and above reserve requirements at the National Bank of Romania (NBR).
  - Implication explored: policy interest rate may reflect an opportunity cost rather than marginal funding cost, potentially causing asymmetric bank responses to policy rate changes.

### II. Model and econometric approach
- Long-run equilibrium assumed (perfect competition form):
  - im = α + β·ip
    - im = market loan rate; ip = policy rate; α = mark up; β = demand elasticity of market rates with respect to policy rates.
- Interpretation of β:
  - β < 1 : relatively inelastic demand (market power, limited substitutes, limited competition).
  - β ≈ 1 or >1 : more elastic demand (credit not rationed, stronger competition).
- Dynamics and estimation:
  - Error-correction model (ECM) used to capture short-run dynamics around long-run equilibrium:
    - ∆imt = γ1 + γ2·∆imt-1 + γ3·(imt-1 - β·ipt-1 - α) + ηi,t
    - γ3 (c(3) in estimations) indicates speed of adjustment (expected negative); larger absolute value → faster market response.
  - Unit-root testing via augmented Dickey-Fuller (1981); cointegration via Johansen (1988, 1991); Engle-Granger two-step logic justifies ECM if variables are I(1) and cointegrated.

### III. Data and sample
- Countries: Romania, the Czech Republic, Hungary, Poland, the Slovak Republic, Slovenia.
- Sample period: January 1995 - February 2004 (monthly), with exceptions:
  - Slovak Republic: sample limited to 2001-04 owing to monetary regime switch in 1998; years 1999 and 2000 left out.
  - Hungary: some series from January 2000 onwards.
  - Slovenia: policy rate characterized by only a few movements since 1995.
- Series: average short- and long-term loan rates, deposit rates, central bank policy rates; short series for interest rates on new loans available for the Czech Republic, Slovak Republic, and Romania.
- Data limitations:
  - Baltic states excluded for lack of data; Bulgaria excluded due to currency board arrangement.
  - Shortest series still has at least three years of monthly data except newly issued loans in Romania.
  - Some newly issued loan series and Romanian short-term series too short for full unit-root testing.

### IV. Results — outstanding loan rates (long-run and ECM)
- Integration and cointegration:
  - Most policy rates and long- and short-term lending rates are I(1) except:
    - Rom_St_Out (Romanian short-term rate): I(2).
    - SVN_St_Out and SVN_Pol (Slovenian short-term and policy rates): I(0).
  - Johansen cointegration tests generally confirm long-run co-movement where series are I(1); exception: Hungarian short-term market rate and Hungarian policy rate fail cointegration test (likely due to short series).
- Long-run policy-rate coefficients (β estimates):
  - Range across countries: 0.67 to 2.07, with most estimates close to 1.
  - Coefficients below 0.8 observed for: Czech Republic (short- and long-term), Hungary (long-term), Romania (short- and long-term), Slovak Republic (long-term).
  - Poland and Slovenia show relatively elastic demand (coefficients closer to or exceeding 1).
- ECM adaptation coefficient γ3 (c(3)):
  - Estimated range across countries: -0.08 to -0.39; most estimates in range -0.09 to -0.18.
  - Romania: γ3 = -0.14 (short-term) and γ3 = -0.15 (long-term).
  - Statistical significance:
    - For most countries γ3 is negative and significant at the 5 percent uncertainty level.
    - Slovenia: significance only at 8 percent (long-term) or 11 percent (short-term).
    - Romania’s short-term γ3 is significantly larger than -0.08 at the 5 percent level; long-term comparison holds at the 12 percent level.
- Numerical diagnostics quoted:
  - "29.0 and 36.5 for the short- and long-term equation, respectively."
- Conclusion from outstanding loan analysis: The hypothesis that pass-through is low in Romania relative to peers is contradicted by these results.

### V. Results — deposit rates, newly issued loans, and panel estimations
- Deposit rates:
  - Long-run equilibrium equations indicate most series are cointegrated.
  - ECM adaptation coefficients for deposit rates are negative and mostly significantly different from 0.
  - Range: -0.13 to -0.60.
  - Romania long-term deposit rate adaptation coefficient: -0.24.
  - Interpretation: deposit rates generally react more quickly to policy rates than loan rates, possibly due to greater deposit market competition or absence of volatile risk premium in deposit rates.
  - Romania does not show an exceptionally slow speed of adaptation for deposit rates.
- Newly issued loans:
  - Data availability limited: time series >2 years only for Czech and Slovak Republics; Romanian series only 16 months (May 2003 to August 2004).
  - Unit-root tests could not be performed for some newly issued loan series; ECM applied with caution for comparability.
  - Long-run coefficients where reliable:
    - Czech and Slovak Republics: between 0.83 and 1.21; close to 1.
    - Czech long-rate coefficient 0.83 is significantly different from 1 at the 5 percent uncertainty level.
    - Czech short-rate and both Slovak coefficients not statistically different from 1.
    - Romania: short-rate policy coefficient 0.24 (significantly different from 1); long-rate policy coefficient 0.62 (not statistically different from 1 at the 5 percent level) — reliability limited by short sample.
  - ECM adaptation γ3 for newly issued loans is much larger (faster pass-through) than for the outstanding portfolio.
  - Conclusion: rates on newly issued loans move quickly toward long-run equilibrium and closely follow policy rates where data permit reliable estimation.
- Panel estimations:
  - Pooling series in a panel regression is inconclusive.
  - Fixed effects panel estimation with a common coefficient on the policy rate confirms the policy rate is a highly significant explanatory variable for market rates (t-statistics reported in source tables).
  - Panel ECM estimation using residuals from the panel regression "does not yield any conclusive results" — likely because "significant changes in monetary policies in the different countries in the sample occurred at very different points in time," making panel residuals heterogeneous.

### VI. Time consistency and sample-split evidence for Romania
- Sample-split breakpoint: the first month in which the policy interest rate was below 40 percent.
- Two subsamples:
  - October 1999-June 2001
  - June 2001-January 2004
- Main findings:
  - Earlier period (1999:10-2001:06):
    - "the policy rate does not significantly influence the market rates."
    - "no cointegration between market and policy rates is found."
  - Later period (2001:07-2004:01):
    - "the policy rate is highly significant for the market rates and the series are cointegrated."
    - Policy-rate coefficients are considerably higher than full-sample estimates; differences statistically significant at the 5 and 10 percent levels for short- and long-term series, respectively.
    - Interpretation: Romanian banking market developed toward a more complete market with increased banking competition and more elastic demand.
- Dynamics:
  - "the interest rate pass-through in Romania has increased over time."
  - Adaptation coefficients in the later subsample indicate a significantly swifter adaptation of short-term dynamics to the long-run equilibrium than in full-sample regressions.
  - Implication: "the NBR could rely on its interest rate policy to a larger extent than in the past."
- Macroeconomic and market-structure interpretation:
  - Improvement partly attributed to successful macroeconomic stabilization in Romania.
  - In a more stable environment, inflationary expectations converge and interest rate policy becomes more effective.
  - Financial deepening and a credit boom are consistent with reduced credit rationing and increased pass-through.

### VII. Conclusions, policy implications, and suggested extensions
- Core conclusion:
  - Empirical evidence does not support the hypothesis that interest rate pass-through in Romania is weaker than in other transition economies in the sample.
  - "The estimates of interest rate pass-through from policy interest rates to rates on the outstanding volume of loans and deposits in Romania are in line with coefficient estimates for other transition economies in the region."
  - Even if the NBR’s execution implies an opportunity cost (rather than marginal cost) for banks, "commercial banks react to these opportunity costs in much the same way as they would to ‘real’ marginal costs."
- Key quantitative takeaways:
  - Long-run policy rate coefficients mostly close to 1; estimates range 0.67 to 2.07 across countries.
  - ECM adaptation coefficients for outstanding loan rates range -0.08 to -0.39; Romania: -0.14 and -0.15.
  - Deposit rate adaptation coefficients range -0.13 to -0.60; Romania long-term: -0.24.
  - Newly issued loan coefficients for Czech and Slovak Republics between 0.83 and 1.21; Romania’s newly issued loan series too short for fully reliable inference but long-rate coefficient 0.62 not statistically different from 1 at the 5 percent level.
- Policy recommendation:
  - Romania "could potentially follow some of its peers by introducing an inflation targeting regime, provided that the monetary authorities build up the necessary forecasting and communications capacity."
- Evidence on newly issued loans:
  - "Market rates for new loans react to policy rate changes quite fast and in most cases almost one-to-one."
  - Caveat: Romanian newly issued loan results should be treated with caution because the series span too short a period for firm conclusions.
- Suggested extension:
  - Natural extension: analyze the complete interest rate channel from the policy rate to the consumer price index, possibly following a VAR methodology; this goes beyond the scope of the paper.
- Methodological caveats:
  - Some series are too short to perform unit-root tests; results for those series should be treated with caution.
  - Appendix notes: "all series are I(1) at the 5 percent uncertainty level, except Rom_St_Out, which is I(2) and SVN_St_Out and SVN_Pol, which are I(0)."

*Source: _wp04211 - References..............................................................................................................................11*

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

### _wp04211 - References..............................................................................................................................11

### I. INTRODUCTION
- Objective: Test hypothesis that the interest rate pass-through from policy to market rates plays a lesser role in Romania than in other transition economies in the region.
- Contextual factors considered:
  - Shared factors: low degree of monetization, underdeveloped financial markets, capital controls.
  - Additional Romanian features: strong and consistent increase in structural excess liquidity starting in 1997 (Anthoni, Udea, and Braun, 2003); commercial banks hold substantial deposits over and above reserve requirements at the National Bank of Romania (NBR).
  - Implication: policy interest rate may reflect an opportunity cost rather than marginal funding cost, potentially causing asymmetric bank responses to policy rate changes.
- Primary finding preview: After estimating pass-through coefficients for several Central European economies (CEEs), pass-through in Romania is in line with that in other countries in the region.
- Paper organization: Sections II–VII cover model, data, results on outstanding loans, deposit and newly issued loan rates, time consistency for Romania, and conclusions.

### II. THE MODEL
- Long-run relationship assumed under perfect competition:
  - im = α + β·ip  (equation (1))
  - im = market loan rate; ip = policy rate; α = mark up; β = demand elasticity of market rates wrt policy rates.
- Interpretation of β:
  - β < 1 indicates relatively inelastic demand (market power, limited substitutes, limited competition).
  - β ≈ 1 or >1 indicates more elastic demand (credit not rationed, stronger competition, possible risk-adjusted pricing causing >1).
- Timing and dynamics:
  - Equation (1) interpreted as long-run equilibrium; market rates adjust gradually to policy rate changes.
  - Error-correction framework employed to capture short-run dynamics around long-run equilibrium.
- ECM specification used:
  - ∆imt = γ1 + γ2·∆imt-1 + γ3·(imt-1 - β·ipt-1 - α) + ηi,t (equation (2))
  - γ3 (c(3) in estimations) indicates speed of adjustment (expected negative); higher absolute value signals faster market response and more effective interest rate channel.
- Econometric approach:
  - Unit root testing via augmented Dickey-Fuller (1981).
  - Cointegration testing via Johansen (1988, 1991).
  - Engle-Granger two-step logic justifies ECM if variables are I(1) and cointegrated.

### III. THE DATA
- Countries included: Romania, the Czech Republic, Hungary, Poland, the Slovak Republic, Slovenia.
- Sample period: January 1995 - February 2004 (monthly data), with exceptions:
  - Slovak Republic sample limited to 2001-04 period owing to monetary regime switch in 1998; years 1999 and 2000 left out of Slovak series.
  - Hungary: data from January 2000 onwards for some series.
  - Slovenia: policy rate characterized by only a few movements since 1995.
- Data series: average short- and long-term loan rates, deposit rates, central bank policy rates; short series of interest rates on new loans available for the Czech Republic, Slovak Republic, and Romania.
- Data limitations noted: Baltic states excluded for lack of data; Bulgaria excluded due to currency board arrangement; shortest series still has at least three years of monthly data except for newly issued loans in Romania.

### IV. RESULTS ON OUTSTANDING LOAN RATES: EQUILIBRIUM EQUATION AND BASIC ECM
- Unit root and cointegration tests:
  - All policy rates and long- and short-term lending rates are I(1) except:
    - Romanian short-term rate: I(2).
    - Slovenian short-term and policy rates: I(0).
  - Johansen cointegration tests generally confirm long-run co-movement between policy and market rates for series that are I(1). One exception: Hungarian short-term market rate and Hungarian policy rate fail cointegration test (likely due to short series).
- Long-run coefficient estimates (policy rate explanatory power):
  - Estimated coefficients vary between 0.67 and 2.07, with most estimates close to 1.
  - Coefficient estimates below 0.8 for: Czech Republic (short- and long-term), Hungary (long-term), Romania (short- and long-term), Slovak Republic (long-term).
  - Poland and Slovenia show relatively elastic demand (coefficients closer to or exceeding 1).
- ECM adaptation coefficient γ3 (c(3)):
  - Expected negative; indicates speed of pass-through.
  - Estimated range across countries: -0.08 to -0.39; most estimates in range -0.09 to -0.18.
  - Romania estimates: -0.14 (short-term) and -0.15 (long-term).
  - Statistical findings:
    - For most countries γ3 is negative and significant at the 5 percent uncertainty level.
    - Slovenia: significance only at 8 percent (long-term) or 11 percent (short-term).
    - Romania’s coefficients are not substantially different from other countries; the short-term Romanian γ3 is significantly larger than -0.08 at the 5 percent level; long-term comparison holds at the 12 percent level.
- Conclusion from outstanding loan results: Hypothesis that pass-through is low in Romania relative to peers is contradicted.

### V. RESULTS ON DEPOSIT, NEWLY ISSUED LOAN RATES, AND PANEL ESTIMATIONS
- Deposit rates:
  - Long-run equilibrium equations indicate most series are cointegrated.
  - ECM adaptation coefficients for deposit rates: negative; most significantly different from 0.
  - Range of adaptation coefficients: -0.13 to -0.60.
  - Romania long-term deposit rate adaptation coefficient: -0.24.
  - Interpretation: deposit rates generally react more quickly to policy rates than loan rates, possibly due to greater deposit market competition or absence of volatile risk premium in deposit rates.
  - Statistical testing: Romania does not exhibit an exceptionally slow speed of adaptation for deposit rates.
- Newly issued loans:
  - Data availability limited: time series >2 years only for Czech and Slovak Republics; Romanian series only 16 months (May 2003 to August 2004).
  - Limitations: Romanian newly issued loan series too short for unit root tests; ECM applied for comparability despite caveats.
  - Long-run coefficients for newly issued loans:
    - Czech and Slovak Republics: estimates between 0.83 and 1.21; close to 1.
    - Czech long-rate coefficient 0.83 is significantly different from 1 at the 5 percent uncertainty level.
    - Czech short-rate and both Slovak coefficients not statistically different from 1.
    - Romania: short-rate policy coefficient 0.24 (significantly different from 1); long-rate policy coefficient 0.62 (not statistically different from 1 at the 5 percent level)—but reliability lower due to short sample.
  - ECM adaptation γ3 for newly issued loans is much larger (faster pass-through) than for outstanding loan portfolio, consistent with expectation.
  - Conclusion: rates on newly issued loans move quickly toward long-run equilibrium and closely follow policy rates where data permit reliable estimation.
- Panel estimations:
  - Pooling series in a panel regression yields inconclusive results.
  - Fixed effects panel estimation with a common coefficient on the policy rate confirms policy rate as a highly significant explanatory variable for market rates (t-statistics reported in source tables).

### VI. TIME CONSISTENCY FOR ROMANIA (SUMMARY OF NOTE)
- The paper examines time consistency of Romanian estimates (Section VI referred to in organization), including robustness checks and sample-split estimations (e.g., Table 7: Romania: Estimation Results for Different Samples – Loan Rates). Results summarized above indicate Romanian pass-through estimates are broadly comparable to peers, though some Romanian series (short-term loan rates; newly issued loans) suffer from limited sample length or different integration properties (I(2) for Romanian short-term rate).

### VII. CONCLUSION (SUMMARY)
- Overall conclusion: The empirical evidence across outstanding loan rates, deposit rates, and newly issued loan rates does not support the hypothesis that interest rate pass-through in Romania is weaker than in other transition economies in the sample.
- Key quantitative takeaways:
  - Long-run policy rate coefficients mostly close to 1; estimates range 0.67 to 2.07 across countries.
  - ECM adaptation coefficients (γ3) for outstanding loan rates range -0.08 to -0.39, with Romanian values -0.14 and -0.15.
  - Deposit rate adaptation coefficients range -0.13 to -0.60, with Romania long-term at -0.24.
  - Newly issued loan coefficients for Czech and Slovak Republics between 0.83 and 1.21; Romania’s newly issued loan series too short for fully reliable inference but long-rate coefficient 0.62 not statistically different from 1 at the 5 percent level.
- Recommended further research: analyze contribution of various factors to the effectiveness of the policy interest rate and estimate transmission from policy rates to inflation.

*Source: _wp04211 - References..............................................................................................................................11*

### 29.0 and 36.5 for the short- and long-term equation, respectively. Further estimation in an

### _wp04211 - 29.0 and 36.5 for the short- and long-term equation, respectively. Further estimation in an

### ECM estimation and cross-country panel issues
- Panel ECM estimation, using residuals from the panel regression for the long-run equilibrium equation, "does not yield any conclusive results."
- Presumed cause: "significant changes in monetary policies in the different countries in the sample occurred at very different points in time."
- Consequence: "the residuals of the long-run equilibrium relation look very different when this relationship is estimated in a panel than when estimated for the countries individually."
- Numeric values preserved from the source:
  - "29.0 and 36.5 for the short- and long-term equation, respectively."

### Results: time consistency (Romania)
- Sample split criterion: the break point is "the first month in which the policy interest rate was below 40 percent."
- Two subsamples:
  - "October 1999-June 2001"
  - "June 2001-January 2004"
- Main findings comparing earlier vs. later period:
  - Earlier period (1999:10-2001:06):
    - "the policy rate does not significantly influence the market rates."
    - "no cointegration between market and policy rates is found," preventing well-founded ECM interpretation.
  - Later period (2001:07-2004:01):
    - "the policy rate is highly significant for the market rates and the series are cointegrated."
    - "the coefficients for the policy rates are considerably higher than the estimates for the full sample," with this difference "statistically significant at the 5 and 10 percent level for the short- and long term rate series, respectively."
    - "These higher estimates indicate that the Romanian banking market has developed towards a more complete market while banking competition increased."
- Interest rate pass-through and dynamics:
  - "the interest rate pass-through in Romania has increased over time."
  - "The estimates of the adaptation coefficients in the ECMs indicate a significantly swifter adaptation of short-term dynamics to the long-run equilibrium than in the regressions for the full sample."
  - Implication: "at least this first part of the interest rate channel of monetary policy has become more effective, indicating that the NBR could rely on its interest rate policy to a larger extent than in the past."
- Macroeconomic context and market structure:
  - Improvement is "surely at least in part due to the successful macroeconomic stabilization which has taken hold in Romania over the last few years."
  - In a more stable macroeconomic environment "inflationary expectations and hence interest rate expectations of different market participants converge, yielding interest rate policy more effective."
  - "The analysis for different time periods also suggests that the Romanian banking market was in general further developed in the later years, and became more competitive, with less market power for individual banks."
  - Evidence: "The policy rate coefficients in the long-run equations are much higher in the later time period, which indicates more elastic demand."
  - Interpretation: "more elastic demand points to a limitation of the market power of the commercial banks, i.e., a more competitive environment."
  - Additional signal: "credit is in general rationed to a much less extent in the later period," consistent with "the financial deepening and credit boom observed in Romania during the last few years."

### Conclusions and policy implications
- Overall counterpoint to prior claims:
  - "Claims that the particular features of Romania’s monetary policy regime result in a lower effectiveness of its interest rate instrument are contradicted by the results of this study."
- Summary findings:
  - "The estimates of interest rate pass-through from policy interest rates to rates on the outstanding volume of loans and deposits in Romania are in line with coefficient estimates for other transition economies in the region."
  - Although NBR execution leads commercial banks to face "an opportunity cost of funding rather than the marginal cost of funding," "the results for the transmission process are similar."
  - Conclusion: "commercial banks react to these opportunity costs in much the same way as they would to ‘real’ marginal costs."
  - Romania's transmission "is similar to that in most other economies in the region," supporting potential for policy shift.
- Policy recommendation and conditionality:
  - Romania "could potentially follow some of its peers by introducing an inflation targeting regime, provided that the monetary authorities build up the necessary forecasting and communications capacity."
- Evidence from newly issued loans:
  - "Results for data series on newly issued loans suggest that market rates for new loans react to policy rate changes quite fast and in most cases almost one-to-one."
  - Caveat: "the results for Romania, however, should be treated with caution, as the time series span too short a period for firm conclusions."
  - Specific contrasts:
    - "the result for Romanian loans with maturity of more than a year is consistent with the thesis that the transmission mechanism in Romania is similar to that in its peers."
    - "the rate on newly issued short-term loans in Romania seems to be lower than one, which is in contrast to the results for the Czech and Slovak Republics that were obtained from longer time series."
- Reiteration on pass-through and competitiveness:
  - "the interest rate pass-through from policy to market rates has become more pronounced over time. Hence, the NBR’s interest rate instrument has become more effective."
  - This supports "the feasibility of the gradual shift towards inflation targeting that the Romanian authorities are undertaking."
  - "The analysis for different time periods also indicates that the Romanian banking market became more competitive over time, a fact consistent with the financial deepening observed in Romania during the last few years."

### Suggested extension (research agenda)
- Natural extension: analyze the complete interest rate channel from the policy rate to the consumer price index.
- Suggested methodology: "such an analysis could follow a VAR methodology" as used in referenced studies, but "goes beyond the scope of this paper."

### Selected methodological and data caveats (from appendix notes)
- Unit-root and cointegration testing notes:
  - Some series are "too short to perform unit root tests" (notably Romanian series for certain tables).
  - Appendix tables use "the standard Johansen Cointegration Test."
  - The appendix notes: "all series are I(1) at the 5 percent uncertainty level, except Rom_St_Out, which is I(2) and SVN_St_Out and SVN_Pol, which are I(0)." (Table 1 note)
  - For some newly issued loan series: "These test results should be treated with caution, as no unit root tests could be performed on the series." (Table 5 and Table 6 notes)
- Model specifications cited in appendix:
  - Long-run specification form: "Country_Rate,t = c(1) + c(2) * Country_Policy_Rate,t"
  - ECM specification form: "D(Country_Rate),t = c(1) + c(2) * D(Country_Rate),t-1 + c(3) * L-T-Eq_Resid"

*Source: IMF working paper content provided in the supplied PDF excerpt.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04211.pdf_