## wp18210

## Source details

**Canonical URL:** [wp18210](https://www.imf.org/-/media/files/publications/wp/2018/wp18210.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2018/wp18210.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2018/wp18210.pdf.json)

---

### I. Introduction and summary findings
- Transparency improves the effectiveness of monetary policy and provides the basis for accountability.
- Dincer and Eichengreen (DE, 2014) measure central bank transparency across 5 broad criteria; inflation-forecast-targeting (IFT) central banks score near the maximum of 15 on the DE index; the Czech National Bank (CNB) scores 14.5 out of 15 on the DE index.
- The CBT-IT index focuses only on inflation-targeting central banks and inspects practices in granular detail to:
  - establish transparency benchmarks reflecting best practices at the most advanced IFT countries;
  - guide improvements on policy objectives, the FPAS, and the policy process.
- Summary finding for the CNB:
  - Clear improvement in transparency over time.
  - Specific areas for further improvement:
    - provide more information about how the CNB is managing the short-run output-inflation tradeoff;
    - explain better the role of the bands;
    - provide more timely documentation of the core projection model;
    - publish a complete set of core macroeconomic variables included in baseline forecasts with confidence bands, risk assessments, and decomposition of forecast revisions;
    - explain the methodology for constructing the forecast confidence bands;
    - publish more detailed contributions by MPC members in the minutes;
    - publish external evaluations of the IFT framework and FPAS at least every 5 years.
  - Several initiatives under consideration by the Board and management.

### II. Principles of Inflation-Forecast Targeting (IFT)
- Core principle:
  - Given a long-run objective for the rate of inflation, the central bank’s own forecast of inflation is an ideal intermediate target (Svensson, 1997).
  - The forecast embodies policymakers’ preferences on the inflation-output tradeoff and the bank’s view of the transmission mechanism summarized in its core forecasting model.
- Defining features (as used in this paper):
  - monetary policy based on a long-run low inflation target and a medium-term forecast path to this target;
  - forecast path for the short-term interest rate is endogenous within the core projection model;
  - forecast is a key input to MPC decisions but not determinative of individual member views;
  - associated forecast paths disclosed soon after policy decisions, with rationale explained in a Monetary Policy Report (MPR) typically quarterly;
  - MPR outlines endogenous forecast of the short-term interest rate qualitatively or numerically;
  - emphasis on risks verbally and with confidence bands for key published variables.
- Transition milestones (illustrated by CNB experience):
  - IT-Lite → Inflation-Forecast Targeting (IFT) → Full-Fledged IFT (publication of conditional numerical forecast path of endogenous interest rate).
- Transmission mechanism:
  - domestic channel: short-term market interest rate;
  - external channel: exchange rate;
  - expectations and the expected Policy Rate Path matter; credibility and management of expectations vital when constrained by the effective lower bound (ELB).

### III. CBT-IT index: structure, coverage, and scoring highlights
- Index coverage:
  - Transparency about Objectives (4 questions: A1-A4).
  - Transparency about the Forecasting and Policy Analysis System (FPAS) (9 questions: B1-B9).
  - Transparency about Policy Process (7 questions: C1-C7).
- Selected scoring rules and frontier expectations (preserving rubrics exactly as provided):
  - A1–A4: full score when inflation is explicitly primary objective, point target preferred, clarity on interaction with financial stability, and use of loss function evaluations.
  - B1–B9: full score elements include making all time series used in forecasts downloadable, publishing core model with code and updated documentation within last 5 years, publishing reaction or loss functions (with coefficients), endogenous-instrument projections for inflation, GDP growth, endogenous interest rate path, output gap, and exchange rate, fan charts and methodology for all key variables, regular publication of forecast revisions and alternative scenarios, and inclusion of financial variables in downloadable format.
  - C1–C7: full score elements include press conferences after every policy meeting (webcast, English), regular forecast updates with Q&A (English), detailed minutes with voting results and attributed contributions within one month, clear roles of staff and policymakers for baseline forecast, annual review of forecasting performance, and external evaluation of policy framework and FPAS within last 5 years.

### IV. Application to the Czech National Bank — evolution and milestones
- Overall trajectory:
  - CBT-IT index: increased from 1 in 1998-2001 to 11.75 out of 20 in 2017.
- Key historical milestones:
  - 1998: IT-lite introduced after currency crisis and floating exchange rate adoption.
  - 2002: Law amended; price stability primary objective; press conferences after each policy decision.
  - Mid-2002: Implementation of a new FPAS (development over a 3-year time frame).
  - 2003: FPAS book published (Coats, Laxton and Rose, 2003).
  - Spring 2004: Point inflation target at 3 percent announced; effective beginning of 2006.
  - March 8, 2007: New inflation target of 2 percent announced, effective from January 2010, with a “tolerance band” of one percentage point on either side.
  - Mid-2008: QPM-Gap replaced by a DSGE model (QPM-g3).
  - 2008: Endogenous interest rate forecast published; historical and key forecast databases released.
  - 2009: Alternative scenarios and loss function values no longer published; exchange rate forecast published.
  - 2013–2014: Exchange rate forecast interrupted during exchange rate commitment; CNB became formally responsible for macroprudential policy by 2014.
  - February (year unspecified in excerpt): CNB resumed publication of the exchange rate forecast of koruna to euro (expected to increase Category B score for 2018).
- Index trajectory notes:
  - Aggregate CBT-IT index and component scores reported in figures (referenced).
  - The index equaled 1 out of 20 between 1998 and (text continues beyond provided excerpt).

### V. Detailed findings and recommendations by CBT-IT categories (preserving numeric values)
- Aggregate and component scores:
  - CNB total score: 11.75 out of 20.
  - Category A (Transparency about Objectives): increased from 0.5 points in 1998 to 2.5 points by 2014 and remained at 2.5 through 2017; perfect score target = 4 points.
  - Category B (Transparency about FPAS): rose from 0 points at IT inception to 1.6 points after mid-2002 implementation of FPAS, to 4.8 points following developments up to 2008–2009, and 3.5 points out of 9 in 2017; perfect score target = 9.
  - Category C (Transparency about Policy Process): increased from 0.5 points (1998-2001) to 5.75 points in 2017; perfect score target = 7.
- Category A — key findings and recommendations:
  - A1. Formal statement of objectives: Score: 1/1. Statutory language cited from Central Bank Act No. 6/1993.
  - A2. Clarity of inflation target: Score: 0.5/1.
    - CNB announced “annual consumer price index growth of 2 percent, effective from January 2010” and will “strive to ensure that actual inflation does not differ from the target by more than one percentage point on either side.”
    - Recommendation: change language to explicitly announce a well-defined point target and clearly explain role of bands.
  - A3. Interaction of financial stability and inflation objectives: Score: 1/1.
    - CNB publishes list of macroprudential tools and is formally responsible for macroprudential policy.
  - A4. Use of loss function evaluation: Score: 0/1.
    - Recommendation: include regular loss function evaluations in future monetary policy reports.
- Category B — key findings and recommendations:
  - B1. Availability of basic economic data: Score: 1/1. All series used in Inflation Report published in downloadable Excel format.
  - B2. Publication/documentation of core quarterly projection model: Score: 0/1.
    - A 2009 paper exists but is not referenced in the Inflation Report.
    - Recommendation: publish model code and web front-end; update documentation within last 5 years.
    - CNB plans to introduce extended forecasting model, publish documentation, and keep it up to date in 2018.
  - B3. Transparency of reaction/loss functions: Score: 0/1.
    - Recommendation: publish coefficients and judgmental adjustments.
    - CNB comment: reaction function published in general form without coefficients; coefficients to be provided in extended documentation.
  - B4. Publication of consistent endogenous-instrument quarterly projections: Score: 0.8/1.
    - CNB published inflation, GDP growth, endogenous interest rate path, and output gap projections; exchange rate not published in 2017.
    - CNB resumed exchange rate projection publication in February 2018.
  - B5. Publication of forecast densities (fan charts): Score: 0.6/1.
    - Fan charts published for inflation, GDP growth, and endogenous interest rate; not for exchange rate and output gap in 2017.
    - CNB resumed exchange rate fan chart publication in February 2018; no plan to publish output gap densities.
  - B6. Methodology for constructing forecast densities: Score: 0/1.
    - Methodology not explained in each Inflation Report; brief notes used.
  - B7. Assessment of forecast revisions: Score: 0.6/1.
    - “Comparison with the previous forecast” section included in each 2017 Inflation Report; no revisions published for exchange rate or output gap.
    - Exchange rate assessment reintroduced with resumed publication in February 2018.
  - B8. Publication of alternative scenarios: Score: 0/1.
    - Alternative scenarios not regularly published in first 3 Inflation Reports in 2017; occasional sensitivity scenarios included.
  - B9. Publication of financial variables and forecasts: Score: 0.5/1.
    - Historical data published for: government bond yield curve, consumer lending rates, mortgage rates, property prices, and credit standards (loan officer surveys).
- Category C — key findings and recommendations:
  - C1. Press statements immediately after policy decisions: Score: 1/1. Published in Czech and English.
  - C2. Post-decision press conference and English availability: Score: 1/1. Press conferences webcast; presentations in Czech and English.
  - C3. Presentation of regular forecast updates with Q&A; English availability: Score: 1/1. Regular meetings with analysts; materials and videos available.
  - C4. Publication of minutes with voting attribution: Score: 0.75/1.
    - Detailed minutes with voting results provided; individual member contributions not recorded.
    - Recommendation: consider fully-attributed minutes (Riksbank example); CNB plans to discuss switching to fully-attributed minutes in first half of 2018.
  - C5. Clarity of roles of staff and policymakers in baseline forecast: Score: 1/1. Ownership and schedule clearly defined.
  - C6. Annual review of forecasting performance: Score: 1/1. Reviewed in every Inflation Report and Annual Report.
  - C7. External evaluation of policy framework and FPAS: Score: 0/1.
    - Finding: no external or independent evaluation in the last 5 years.
    - CNB may consider asking the IMF to do an external evaluation in the future after switching to the planned extended core model.

### VI. Procedural and operational transparency rubrics (selected items and scoring thresholds)
- Procedural Transparency questions and scoring included explicit policy rule presence (0/1), comprehensive account of policy deliberations within reasonable time (0/1), and disclosure of how each decision was reached (0/1).
- Policy Transparency scoring included prompt announcement of adjustments (0/1), explanation provision when decisions are announced (0, 1/2, 1), and disclosure of explicit policy inclination after every meeting (0/1).
- Operational Transparency scoring included regular evaluation of whether operating targets achieved (0, 1/2, 1), regular provision of information on macroeconomic disturbances (0, 1/2, 1), and regular evaluation of policy outcome in light of objectives (0, 1/2, 1).

### VII. Key numeric values and dates (preserved exactly)
- CBT-IT index: increased from 1 (1998-2001) to 11.75 out of 20 in 2017.
- Category A: 0.5 points (1998) → 2.5 points (2014–2017); perfect score target = 4 points.
- Category B: 0 points (initial) → 1.6 points (post-2002) → 4.8 points (post-2008) → 3.5 points out of 9 in 2017; perfect score target = 9.
- Category C: 0.5 points (1998-2001) → 5.75 points in 2017; perfect score target = 7.
- Inflation target chronology:
  - Point inflation target at 3 percent announced spring 2004; effective beginning of 2006.
  - New inflation target: “annual consumer price index growth of 2 percent, effective from January 2010.” CNB will “strive to ensure that actual inflation does not differ from the target by more than one percentage point on either side.”
- Exchange rate commitment communicated as 27 CZK/EUR.
- QPM-Gap development: over 3 years.
- Model documentation: working paper on QPM-g3 (Andrle and others, 2009); core model working paper age > 5 years as of 2014.
- CNB total score in 2017: 11.75 out of 20.

*Source: wp18210 - IMF working paper content provided.*

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

### References .............................................................................................................................. 68

### Figures
- 1. Monetary Policy Model: IFT Feedback Response and Transmission ...................................6
- 2. CBT-IT Index versus Dincer-Eichengreen Central Bank Transparency Index for the      
    Czech National Bank ...........................................................................................................16
- 3. Evolution of the CBT-IT Index for the Czech National Bank, 1998-2017 .........................17
- 4. Evolution of the Three Components of the CBT-IT Index for the Czech National  
    Bank, 1998-2017 ..................................................................................................................19

### Tables
- (no table entries listed in the provided content)

*Source: wp18210 - References, https://www.imf.org/-/media/files/publications/wp/2018/wp18210.pdf*

### 1. Stages of CNB’s Inflation Targeting .................................................................................

### 1. Stages of CNB’s Inflation Targeting

### I. Introduction — transparency and the need for a finer index
- Transparency improves the effectiveness of monetary policy and provides the basis for accountability.
- Dincer and Eichengreen (DE, 2014) measure central bank transparency across 5 broad criteria: political, economic, procedural, policy, and operational.
- Inflation-forecast-targeting (IFT) central banks score near the maximum of 15 on the DE index; the Czech National Bank (CNB) scores 14.5 out of 15 on the DE index.
- Limitations of the DE index motivating a new index:
  - Difficult to compare transparency across different monetary regimes.
  - DE index insufficiently emphasizes advanced forms of communications; most IFT countries cluster near the maximum score.
  - Post-global financial crisis emphasis on financial stability and macroprudential policies requires transparency about policy interactions.
- Purpose of the CBT-IT index:
  - Focuses only on inflation-targeting central banks and inspects practices in granular detail.
  - Establishes transparency benchmarks reflecting best practices at the most advanced IFT countries.
  - Guides improvements on policy objectives, the FPAS, and the policy process.
- Summary finding for the CNB:
  - Application of the index shows clear improvement in transparency over time.
  - Despite high scores, the CBT-IT index identifies several specific areas for further improvement:
    - (1) providing more information about how the CNB is managing the short-run output-inflation tradeoff;
    - (2) explaining better the role of the bands;
    - (3) providing more timely documentation of the core projection model;
    - (4) publishing a complete set of core macroeconomic variables that are included in their baseline forecasts with confidence bands, risk assessments, and decomposition of forecast revisions;
    - (5) explaining the methodology for constructing the forecast confidence bands;
    - (6) publishing more detailed contributions by monetary policy committee (MPC) members in the minutes;
    - (7) publishing external evaluations of their IFT framework and FPAS at least every 5 years.
  - Several of these initiatives are under consideration by the Board and management of the CNB.

*Italic line: Source: wp18210 - 1. Stages of CNB’s Inflation Targeting (IMF working paper content provided).*  

### II. Principles of Inflation-Forecast Targeting (IFT)
- Core principle:
  - Given a long-run objective for the rate of inflation, the central bank’s own forecast of inflation is an ideal intermediate target (Svensson, 1997).
  - The forecast embodies relevant information including policymakers’ preferences on the inflation-output tradeoff and the bank’s view of the transmission mechanism summarized in its core forecasting model.
- Defining features of IFT (for the purposes of this paper):
  - monetary policy is based on a long-run low inflation target, and a medium-term forecast path to this target;
  - the forecast path for the short-term interest rate—the policy instrument—is endogenous within the core projection model, with the rate varying to achieve the long-run inflation target and to eliminate any output gap;
  - the forecast is a key input into the MPC’s decision, but only an input—MPC members need not agree with the forecast but use it to explain their own views;
  - soon after the policy decision, the associated forecast path for key macroeconomic variables is disclosed, highlighting the path for the inflation rate;
  - the rationale for policy actions is explained in greater depth at regular intervals (usually quarterly) in a Monetary Policy Report (MPR);
  - the MPR outlines the endogenous forecast of the short-term interest rate, either qualitatively or with a numerical forecast path;
  - the central bank emphasizes risks verbally and with confidence bands for key published variables.
- Transition milestones illustrated by CNB experience:
  - IT-Lite: announcement of targets with a multi-year horizon; precision on the policy interest rate setting; transparent communications on policy implementation.
  - Inflation-Forecast Targeting (IFT): publication of the inflation forecast constructed under the assumption of an endogenous interest rate—clarity of the intermediate target.
  - Full-Fledged IFT: publication of a conditional numerical forecast path of the endogenous interest rate—clarity of how the central bank is managing the short-run output-inflation trade-off.
- Transmission mechanism emphasis:
  - Two main channels: domestic (short-term market interest rate) and external (exchange rate).
  - Importance of expectations: the whole expected Policy Rate Path (yield curve) matters; policy affects rates businesses and households face through expected future policy rates.
  - When constrained by the effective lower bound (ELB), credibility and management of expectations are central to effective anti-deflation policy.

### III. Central Bank Transparency Index for IFT (CBT-IT): questions and motivations
- Index coverage:
  - Transparency about Objectives (4 questions: A1-A4).
  - Transparency about the Forecasting and Policy Analysis System (FPAS) (9 questions: B1-B9).
  - Transparency about Policy Process (7 questions: C1-C7).
  - Appendix II contains the questions and scoring system (not reproduced here).
- III.1 Transparency about Objectives (A1-A4):
  - Examines: statement of objectives; clear communication of objectives; communication of performance managing the short-run output-inflation tradeoff; role of financial stability in monetary policy communication.
  - Highest score conditions:
    - explicit that inflation is the primary objective and the bank is managing the short-run output-inflation tradeoff;
    - well-defined point target preferred; alternative definitions include point target with band (e.g., 2 percent +/- 1 percentage point) or tolerance/control range (e.g., 1 percent to 3 percent).
  - Observations and evidence:
    - Point targets tend to better anchor long-term inflation expectations than range targets (Goretti and Laxton, 2005; Freedman and Laxton, 2009a,b,c).
    - Example of precise language: Riksbank variation band around 2 percent, described as a variation band between 1 and 3 per cent that does not affect the formulation of monetary policy.
  - Financial stability considerations:
    - Transparency about how macroprudential tools will be adjusted to achieve financial stability without jeopardizing monetary policy objectives is critical.
    - Both central banks with formal financial stability mandates and those without can score fully on the CBT-IT index provided they are transparent about their institutional design and frameworks.
  - Communication of achievement:
    - Central banks should provide clear communications about how well they have been managing the short-run output-inflation tradeoff—one approach is to calculate the historical value of a quadratic loss function penalizing inflation deviations from target, output deviations from potential, and changes in the policy interest rate.
- III.2 Transparency about the FPAS (B1-B9):
  - Core elements of FPAS:
    - a monetary policy model containing a policy reaction function or a loss function with an endogenous short-term interest rate response;
    - communication between policymakers and model builders;
    - communication between policymakers and forecasters to develop a consensus set of assumptions for baseline forecasts and alternatives.
  - Key FPAS outputs for decisions:
    - a baseline forecast—including an endogenous forecast path for the short-term interest rate;
    - risk assessments surrounding the baseline forecast;
    - alternative consistent scenarios each with an endogenous interest rate path.
  - Transparency expectations and scoring highlights:
    - Full score if data for time series used in forecasts are available (capacity utilization/output gap, inflation, inflation expectations, wages, unemployment, GDP); sharing both historical and key forecast databases is preferred.
    - Full score if the core model is transparently documented and replication code (including judgmental assumptions) is public; model documentation should be updated at least once every 5 years.
    - High score when the central bank publishes either the reaction function or loss function (with coefficients) and communicates judgment involved in the forecast.
    - Models should incorporate an endogenous interest rate; publishing the endogenous interest rate path and confidence bands is frontier transparency.
    - Full score for publishing fan charts for all macro variables (inflation, GDP growth, endogenous interest rate path, output gap, exchange rate) and the methodology for constructing fan charts.
    - Full score for clearly communicating full-fledged alternative scenarios (not mere sensitivity tests).
    - Full score when the bank publishes forecast revisions for the full set of historical and projected macro variables: inflation, GDP growth, endogenous interest rate path, output gap, exchange rate.
    - Inclusion of measures of financial conditions in MPRs is tracked; attention to financial conditions improves forecasting performance.
- III.3 Transparency about Policy Process (C1-C7):
  - Communication modalities expected to achieve full score:
    - Following every policy meeting, and at pre-announced dates and times, the central bank should hold a press conference; press conference and Q&A should be webcast and available on the website; presentations and Q&A should be in English.
    - Present regular forecast updates to journalists, analysts, and market participants with opportunities for Q&A; presentations should be in English.
    - Publish detailed minutes with voting outcomes on the main policy instrument; individual MPC member contributions should be clearly attributed.
    - Define clearly the roles of staff and policymakers in the decision-making process; staff forecast as a coherent point of reference is recommended when committee members may have divergent views (CNB Foreword citation).
    - Review and publish forecast performance at least once a year to get full score; allow an external and independent evaluation of the policy framework and FPAS every five years to secure another full score.
  - Institutional considerations:
    - When the MPC votes rather than a single Governor, publishing individual member forecasts could be inefficient and confusing; publishing a staff forecast as a reference point is a high-transparency practice.

### IV. Application of the CBT-IT index to the Czech National Bank (evolution over time)
- Comparative scores:
  - DE index: CNB scored 14.5 out of 15 (Dincer and Eichengreen, 2014).
  - CBT-IT index: also scores the CNB at high levels but reveals areas for improvement (Figure 2 referenced).
- Historical context:
  - CNB adopted IT in 1998 after a currency crisis forced floating exchange rates.
  - Initial IT (1998) was an IT-lite strategy with low monetary policy transparency due to lack of FPAS.
- Index trajectory insights (figures and detailed historical CBT-IT scores referenced):
  - The aggregate quantitative magnitude of the CBT-IT index and component scores are reported (Figure 3 and Figure 4 referenced).
  - The index was equal to 1 out of 20 between 1998 and (text continues beyond provided excerpt).
- Appendix III (referenced) provides:
  - Explanation of the 2017 scores,
  - Recommendations for the CNB,
  - CNB’s comments for some questions.

*Italic line: Source: wp18210 - 1. Stages of CNB’s Inflation Targeting (IMF working paper content provided).*

### 2001.  This  low  score  reflects  the  fact  that  the  CNB  revealed  very  little  about  its  monetary

### wp18210 - 2001

### Overview and evolution of transparency
- The CNB’s overall CBT-IT index increased from 1 in 1998-2001 to 11.75 out of 20 in 2017.
- Initial low transparency (1998-2001) reflected very limited disclosure about monetary policy objectives and the absence of an FPAS.
- Key milestones:
  - 2002: Law amended; price stability becoming the primary objective; press conferences held after each policy decision.
  - Mid-2002: Implementation of a new FPAS (development over a 3-year time frame).
  - 2003: FPAS book published (Coats, Laxton and Rose, 2003).
  - 2006: Point inflation target at 3 percent announced in spring 2004 and became effective at the beginning of 2006.
  - March 8, 2007: CNB announced new inflation target of 2 percent, effective from January 2010, with a “tolerance band” of one percentage point on either side.
  - Mid-2008: QPM-Gap replaced by a DSGE model (QPM-g3).
  - 2008: Endogenous interest rate forecast published; historical and key forecast databases released.
  - 2009: Alternative scenarios and loss function values no longer published; exchange rate forecast published.
  - 2014: Exchange rate forecast no longer published; CNB became formally responsible for macroprudential policy by 2014.
  - February (year not specified in excerpt): CNB resumed publication of the exchange rate forecast of koruna to euro (expected to increase Category B score for 2018).

### Category A — Transparency about Objectives (Questions A1-A4)
- Score evolution:
  - Category A increased from 0.5 points in 1998 to 2.5 points by 2014 and remained at 2.5 through 2017.
- Facts on objectives and targeting:
  - Pre-1998 framework: pegged exchange rate + money targeting.
  - Post-May 1997: move to IT; IT-lite introduced in 1998 with objective defined as “monetary stability” (internal and external stability).
  - Statutory language (Central Bank Act No. 6/1993, amended in 2002) states: “The primary objective of the Czech National Bank shall be to maintain price stability. In addition, the Czech National Bank shall work to ensure financial stability and the safe and sound operation of the financial system in the Czech Republic. Without prejudice to its primary objective, the Czech National Bank shall support the general economic policies of the Government leading to sustainable economic growth and the general economic policies in the European Union with a view to contributing to the achievement of the objectives of the European Union.”
  - Inflation target evolution:
    - Started as a range in 1998.
    - Point target at 3 percent announced in spring 2004; effective beginning of 2006.
    - New target announced March 8, 2007: annual consumer price index growth of 2 percent, effective from January 2010; CNB will “strive to ensure that actual inflation does not differ from the target by more than one percentage point on either side.”
  - Financial stability mandate introduced de facto in 2004; macroprudential responsibilities and a list of macroprudential tools introduced and used actively by 2014.
- Recommendations to reach perfect Category A score (4 points):
  - Change language to clearly explain the role of the bands around the 2 percent target to avoid implying range-targeting.
  - Publish values of a loss function to show performance in managing the output-inflation tradeoff and progress in monetary policy efficiency.

### Category B — Transparency about the FPAS (Questions B1-B9)
- Score evolution:
  - Category B rose from 0 points at IT inception to 1.6 points after mid-2002 implementation of FPAS.
  - Category B rose further to 4.8 points following developments up to 2008–2009.
  - In 2017 the CNB scored 3.5 points out of 9 for Category B.
- Key developments and features:
  - Initial FPAS absence forced reliance on data-driven near-term forecasting and expert judgment, producing very large forecast errors in early years.
  - Development and implementation of QPM-Gap (small-scale calibrated monetary policy model) over 3 years with IMF technical assistance; QPM-Gap had medium-term focus and an endogenous short-term interest rate equation.
  - CNB began publishing model-based endogenous forecasts for inflation and GDP growth; in 2007 all data used in the Inflation Report made publicly available.
  - Mid-2008: CNB replaced QPM-Gap with DSGE model (QPM-g3) and documented/published it.
  - Beginning of 2008: CNB started publishing fan charts for the policy rate path.
  - 2009: CNB started publishing koruna-euro exchange rate projection with fan charts; published alternative scenarios and sensitivity analysis in each Inflation Report.
  - 2014 retrenchment: stopped publishing the exchange rate forecast and alternative scenarios; communicated exchange rate commitment (27 CZK/EUR) and expected duration instead; core model working paper became older than 5 years.
- Recommendations to reach perfect Category B score (9 points):
  - Publish the core model with coefficients and code to allow users to replicate forecasts.
  - Publish baseline forecasts with fan charts for the output gap and alternative scenarios regularly in the Inflation Reports.
  - Continue and formalize publication of exchange rate forecasts (resumption noted in February increases 2018 score).

### Category C — Transparency about Policy Process (Questions C1-C7)
- Score evolution:
  - Category C increased from 0.5 points (1998-2001) to 5.75 points in 2017.
- Practices and disclosures:
  - Board issues a press release immediately after decisions; Governor gives a press conference the same afternoon.
  - Monetary policy decision context:
    - Explained in the context of a new macroeconomic forecast four times a year.
    - Inter-forecast policy meetings: risk assessment to the previous quarterly forecast (four times a year since 2008).
  - Presentations include votes cast by Board members on interest rate decisions.
  - Since 2014: Governor provides a written explanation of the decision followed by Q&A.
  - Eight days after the policy meeting: CNB publishes Minutes (with individual votes since 2008) and the Inflation Report containing forecast details.
  - Quarterly meetings with financial market analysts one day after the policy decision, with senior staff and one or two Board members; meetings web-casted live on CNB’s website; Executive Summary of the Inflation Report released with a detailed forecast table.
  - Ownership of the forecast and staff/policymaker roles in decision-making are clearly defined.
- Recommendations to reach perfect Category C score (7 points):
  - Clearly attribute contributions of individual MPC members.
  - Carry out an external evaluation of the policy framework and FPAS at least once every five years.

### Index components and comparative observations
- All three components (Objectives, FPAS, Policy Process) showed significant improvements between 1998 and 2017.
- Transparency about the policy process is particularly high for the CNB.
- Transparency about the FPAS is the component with the largest scope for further improvement.
- The development of the FPAS and modeling framework provided the foundation for IFT adoption in 2002 and subsequent transparency gains; publishing the policy-rate path in 2008 marked the beginning of full-fledged IFT in the Czech Republic.
- Clinton and others (2017) find published interest rate paths provided forward guidance and argue that publishing interest rate forecasts with confidence bands is more robust than other forward guidance forms used during the Global Financial Crisis.

### Key numeric values (preserved exactly as in source)
- CBT-IT index: increased from 1 (1998-2001) to 11.75 out of 20 in 2017.
- Category A: 0.5 points (1998) → 2.5 points (2014–2017); perfect score target = 4 points.
- Category B: 0 points (initial) → 1.6 points (post-2002) → 4.8 points (post-2008) → 3.5 points out of 9 in 2017; perfect score target = 9.
- Category C: 0.5 points (1998-2001) → 5.75 points in 2017; perfect score target = 7.
- Inflation targets and dates:
  - Point inflation target at 3 percent announced spring 2004; effective beginning of 2006.
  - New inflation target: “annual consumer price index growth of 2 percent, effective from January 2010.” CNB will “strive to ensure that actual inflation does not differ from the target by more than one percentage point on either side.”
- Exchange rate commitment communicated as 27 CZK/EUR.
- QPM-Gap development: over 3 years.
- Model documentation: working paper on QPM-g3 (Andrle and others, 2009); core model working paper age > 5 years as of 2014.

*Source: wp18210 - 2001 (https://www.imf.org/-/media/files/publications/wp/2018/wp18210.pdf).*

### 3. Procedural Transparency

### 3. Procedural Transparency (Content unit: wp18210 - 3. Procedural Transparency)

### Procedural Transparency — question items and scoring
- (a) Explicit policy rule or strategy describing monetary policy framework  
  - No = 0.  
  - Yes = 1.
- (b) Comprehensive account of policy deliberations (or explanations in case of single central banker) within a reasonable amount of time  
  - No or only after a substantial lag (more than eight weeks) = 0.  
  - Yes, comprehensive minutes (although not necessarily verbatim or attributed) or explanations (in case of a single central banker), including a discussion of backward- and forward-looking arguments = 1.
- (c) Disclosure of how each decision on the level of main operating instrument or target was reached  
  - No or only after a substantial lag (more than eight weeks) = 0.  
  - Yes, comprehensive minutes (although not necessarily verbatim or attributed) or explanations (in case of a single central banker), including a discussion of backward- and forward-looking arguments = 1.

### Policy Transparency — question items and scoring
- (a) Prompt announcement of adjustments to main operating instrument or target  
  - No or only after the day of implementation = 0.  
  - Yes, on the day of implementation = 1.
- (b) Provision of explanation when policy decisions are announced  
  - No = 0.  
  - Yes, when policy decisions change, or only superficially = 1/2.  
  - Yes, always and including forwarding-looking assessments = 1.
- (c) Disclosure of explicit policy inclination after every policy meeting or explicit indication of likely future policy actions (at least quarterly)  
  - No = 0.  
  - Yes = 1.

### Operational Transparency — question items and scoring
- (a) Regular evaluation of whether main policy operating targets have been achieved  
  - No or not very often (at less than annual frequency) = 0.  
  - Yes but without providing explanations for significant deviations = 1/2.  
  - Yes, accounting for significant deviations from target (if any); or, (nearly) perfect control over main operating instrument/target = 1.
- (b) Regular provision of information on (unanticipated) macroeconomic disturbances affecting policy transmission  
  - No or not very often = 0.  
  - Yes but only through short-term forecasts or analysis of current macroeconomic developments (at least quarterly) = 1/2.  
  - Yes, including a discussion of past forecast errors (at least annually) = 1.
- (c) Regular evaluation of policy outcome in light of macroeconomic objectives  
  - No or not very often (at less than annual frequency) = 0.  
  - Yes but superficially = 1/2.  
  - Yes, with an explicit account of the contribution of monetary policy in meeting the objectives = 1.

---

### Appendix II: CBT-IT INDEX QUESTIONS — structure and scoring rubrics

### Category A: Transparency about Objectives (A1–A4)
- A1. Formal statement of objectives of monetary policy emphasizing inflation as primary objective; accessibility on website  
  - Single inflation objective or multiple policy objectives without prioritization. = 0.0  
  - Inflation as the primary objective such that any other objective (output, etc.) cannot be inconsistent with the primary objective of anchoring inflation and inflation expectations. = 1.0  
  - Example quoted: CNB under Article 98 of the Constitution.
- A2. Clarity of inflation target definition  
  - No medium-term numerical target over a horizon of 2-3 years or more = 0.0  
  - Inflation target defined as a “tolerance” or “control range” target; or defined as a medium-term target but meaning of range/band not clear = 0.5  
  - Inflation target defined as a well-defined point target. If a band is used, it is clearly communicated = 1.0  
  - Example quoted: Riksbank 2 percent target with 1–3 percent variation band.
- A3. Interaction between financial stability and primacy of inflation objective  
  - (i) If another institution is responsible for financial stability:  
    - Central bank cares about financial stability to extent it affects stabilization objectives but unclear that inflation is primary = 0.0  
    - Central bank makes clear that inflation is the primary objective = 1.0  
  - (ii) If central bank is at least partly responsible for financial stability:  
    - Borderlines unclear, creating confusion about primary objective = 0.0  
    - Central bank has both tools and it is clear how it adjusts them to achieve monetary policy and financial stability objectives = 1.0
- A4. Use of a loss function evaluation to show management of short-run output-inflation tradeoff  
  - No = 0.0  
  - Yes = 1.0

### Category B: Transparency about the FPAS (Forecasting, Policy Analysis, and System) (B1–B9)
- B1. Availability of basic economic data in downloadable format (minimum series list)  
  - No database publicly available = 0.0  
  - Minimal set of series publicly available (output gap/capacity utilization, inflation, inflation expectations, wages, unemployment, GDP) = 0.5  
  - All series used in producing the MPR published in downloadable format (includes at least the seven series above) = 1.0  
  - Example: CNB publishes Inflation Report data but history limited to report span.
- B2. Public availability of core quarterly projection model and documentation updated within last 5 years  
  - No = 0.00  
  - Yes, in a “working paper” format only, i.e., irreproducible = 0.25  
  - Yes, in a working paper and with code = 0.50  
  - Yes, working paper, with code, and web-based front-end to modify forecast assumptions = 1.00
- B3. Transparency of reaction functions or loss functions used to compute interest rate paths in regular projection exercises  
  - Does not publish either reaction function or loss function = 0.0  
  - Publishes reaction function and/or loss function (with coefficients) in an easily accessible place = 1.0
- B4. Variables for which consistent endogenous-instrument quarterly projections are published over horizon of at least two years  
  - None = 0.0  
  - Inflation = 0.2  
  - Inflation and GDP growth = 0.4  
  - Inflation, GDP growth, and the endogenous interest rate path = 0.6  
  - Inflation, GDP growth, the endogenous interest rate path, and the output gap = 0.8  
  - Inflation, GDP growth, the endogenous interest rate path, the output gap, and the exchange rate = 1.0
- B5. Regular publication of forecast densities (fan charts) to communicate forecast uncertainty  
  - No fan chart = 0.0  
  - Fan chart for inflation = 0.2  
  - Fan charts for inflation and GDP growth = 0.4  
  - Fan charts for inflation, GDP growth, and the endogenous interest rate path = 0.6  
  - Fan charts for inflation, GDP growth, the endogenous interest rate path, and the output gap = 0.8  
  - Fan charts for inflation, GDP growth, the endogenous interest rate path, the output gap, and the exchange rate = 1.0
- B6. Clarity and accessibility of underlying methodology constructing forecast densities (fan charts)  
  - No fan chart, or methodology not explained = 0.0  
  - Fan charts published in all monetary policy reports and methodology clearly explained and/or link to technical paper provided = 1.0
- B7. Regular publication of assessment of forecast revisions (decomposition of forecast changes vs. previous forecast)  
  - No = 0.0  
  - For inflation only with discussion of underlying causes = 0.2  
  - For inflation and GDP growth with discussion of underlying causes = 0.4  
  - For inflation, GDP growth, and endogenous interest rate path with discussion = 0.6  
  - For inflation, GDP growth, endogenous interest rate path, and output gap with discussion = 0.8  
  - For inflation, GDP growth, endogenous interest rate path, output gap, and exchange rate with discussion = 1.0
- B8. Publication of alternative scenarios in monetary policy reports to illustrate key risks in baseline forecast  
  - No alternative scenario = 0.0  
  - Major risk(s) communicated in an alternative scenario(s) = 1.0
- B9. Inclusion of historical data and forecasts for financial variables in monetary policy reports (downloadable)  
  - No data or forecast = 0.0  
  - Historical data on less than 5 of listed variables and forecasts for less than 5 = 0.1–0.9*  
  - Historical data on 5 or more variables and forecasts for 5 or more variables = 1.0  
  - * Scoring: 0.1 for each type of financial variable up to maximum 0.5 for historical series; 0.1 for each type up to maximum 0.5 for forecast series.

### Category C: Transparency about Policy Process (C1–C7)
- C1. Publication of a press statement immediately following policy decisions  
  - Does not publish immediately after decisions = 0.0  
  - Publishes press statements in native language only = 0.5  
  - Publishes press statements in English = 1.0
- C2. Policy decision explained at a press conference immediately after announcement; presentations available in English  
  - No = 0.0  
  - Yes, after all policy meetings at pre-announced dates/times; press conference with Q&A webcast and recording available; presentations downloadable only in native language = 0.5  
  - Yes, as above, presentations downloadable in English = 1.0  
  - Example: Norges Bank press conference practice referenced.
- C3. Regular presentation of forecast updates with Q&A to journalists, analysts, market participants; presentations available in English  
  - No = 0.0  
  - Yes, presentation and Q&A available only in native language = 0.5  
  - Yes, presentation and Q&A available in English = 1.0  
  - Example: Czech National Bank regular meetings with analysts referenced.
- C4. Public account of policy deliberations (“minutes”) published in less than one month after the meeting  
  - (i) When policy decisions made by a monetary policy committee:  
    - No = 0.00  
    - Yes, but condensed, non-attributed, and without voting results = 0.50  
    - Yes, detailed and with voting results on main policy instrument; contributions not attributed = 0.75  
    - Yes, detailed and with voting results; contributions attributed = 1.00  
  - (ii) When policy decisions made by a single policymaker:  
    - No = 0.00  
    - Yes, with arguments/explanations = 1.00
- C5. Clarity about role of staff and policymakers in the baseline forecast process  
  - No — not clear how forecast is constructed and used = 0.0  
  - Yes — ownership of forecast and its role in decision-making defined clearly = 1.0  
  - Example: CNB quote describing forecast role and Bank Board meeting schedule.
- C6. Forecasting performance reviewed at least once a year in monetary policy reports or separate document  
  - No = 0.0  
  - Yes = 1.0
- C7. External evaluation of policy framework and FPAS publicly available in last 5 years  
  - No evaluation in last 5 years = 0.0  
  - Either policy framework or FPAS evaluation in last 5 years = 0.5  
  - Both policy framework and FPAS evaluation in last 5 years = 1.0  
  - Example: None.

*Italic: Source — wp18210 - 3. Procedural Transparency (PDF chapter/section).*

### APPENDIX III: CBT-IT INDEX SCORE FOR THE CZECH NATIONAL BANK IN 2017

### APPENDIX III: CBT-IT INDEX SCORE FOR THE CZECH NATIONAL BANK IN 2017

### Overview
- CNB total score: 11.75 out of 20.
- Staff at the CNB were given a chance to comment and presented envisaged future plans to strengthen transparency modalities consistent with the frontiers of IT.

### A. Objectives and mandate (Questions A1–A4)
- A1. Formal statement of objectives (Score: 1/1)
  - Findings:
    - Central Bank Act No. 6/1993 quoted: “The primary objective of the Czech National Bank shall be to maintain price stability. In addition, the Czech National Bank shall work to ensure financial stability and the safe and sound operation of the financial system in the Czech Republic. Without prejudice to its primary objective, the Czech National Bank shall support the general economic policies of the Government leading to sustainable economic growth and the general economic policies in the European Union with a view to contributing to the achievement of the objectives of the European Union.”
    - Statement repeated on the first page of each Inflation Report.
  - Recommendation:
    - Consider the United States Federal Reserve Bank’s one-pager as an illustrative example for explicitly publishing and explaining the dual mandate (reference provided).
    - Consider renaming the Inflation Report to the Monetary Policy Report to reflect broader role of monetary policy.
  - CNB comment:
    - Board approved project to switch from an Inflation Report to a Monetary Policy Report in 2019-20, focused on explaining price stability, role of monetary policy, uncertainty, and tradeoffs.

- A2. Clarity of inflation target (Score: 0.5/1)
  - Findings:
    - CNB announced: “the new inflation target as annual consumer price index growth of 2 percent, effective from January 2010. As before, the CNB will strive to ensure that actual inflation does not differ from the target by more than one percentage point on either side. The CNB continues to view its inflation target as a medium-term one which actual inflation may deviate from temporarily as a result of exogenous shocks. Changes to indirect taxes are one such shock; monetary policy does not react to the first-round effects of such changes and concentrates only on their second-round effects.”
  - Recommendation:
    - Change language to explicitly announce a well-defined point target and clearly explain the role of bands; Riksbank wording offered as an example.
  - CNB comment:
    - Staff judge the bank has systematically behaved as a point targeter; band of +/- one percentage point communicates uncertainty, not inaction; may reconsider terminology in future.

- A3. Interaction of financial stability and inflation objectives (Score: 1/1)
  - Findings:
    - CNB states financial stability is also an objective and is responsible for setting macroprudential policy.
    - Published list of macroprudential tools: countercyclical capital buffers, systemic risk buffer, capital conservation buffer, recommendations on loan-to-value ratios.
  - CNB comment:
    - Macroprudential toolkit deemed sufficient so far; CNB aims to strengthen tools by adding binding regulations on LTV, DTI and DSTI ratios. Amendment to law not passed by previous Parliament; plans to resubmit after general elections in October 2017.

- A4. Use of loss function evaluation (Score: 0/1)
  - Findings:
    - A 2008 research paper evaluated performance using a loss function (Evaluation of the Fulfillment of the CNB’s Inflation Targets 1998-2007, Chapter 2, page 25), but it was not a policy document.
  - Recommendation:
    - Include regular loss function evaluations in future monetary policy reports.

### B. Data, models, and forecast transparency (Questions B1–B9)
- B1. Availability of basic economic data (Score: 1/1)
  - Findings:
    - All series used in producing the Inflation Report were published in downloadable Excel format, including capacity utilization, inflation, inflation expectations, wages, unemployment, and GDP.

- B2. Publication and documentation of core quarterly projection model (Score: 0/1)
  - Findings:
    - Published paper from 2009: “Implementing the New Structural Model of the Czech National Bank” exists but is not referenced in the Inflation Report.
  - Recommendation:
    - Publish model code and web front-end; update documentation within last 5 years; Riksbank’s Ramses II provided as example.
  - CNB comment:
    - Provided core model equations and coefficients upon request; plan to introduce extended forecasting model, publish its documentation, and keep it up to date on CNB webpage in 2018, with material on usage in FPAS.

- B3. Transparency of reaction/loss functions in projections (Score: 0/1)
  - Findings:
    - Core model working paper is old and not easily accessible; Inflation Report does not reference documentation that contains loss or reaction functions.
  - CNB comment:
    - Reaction function published in general form but without coefficient values; coefficients to be provided in extended model documentation; CNB added no judgments to interest rate path unless ELB binding.

- B4. Publication of consistent endogenous-instrument quarterly projections (Score: 0.8/1)
  - Findings:
    - CNB published inflation, GDP growth, endogenous interest rate path, and output gap projections; did not publish exchange rate series in 2017 (hence less than full score).
  - CNB comment:
    - Exchange rate projection (CZK/EUR) published in 2013, interrupted during exchange rate commitment from November 2013, and resumed publication of exchange rate path in February 2018.

- B5. Publication of forecast densities (fan charts) (Score: 0.6/1)
  - Findings:
    - Fan charts published for inflation, GDP growth, and endogenous interest rate.
    - Fan charts not published for exchange rate and output gap in 2017.
  - CNB comment:
    - No plan to publish forecast densities for the output gap; exchange rate fan chart publication resumed in February 2018.

- B6. Methodology for constructing forecast densities (Score: 0/1)
  - Findings:
    - Methodology not explained in each Inflation Report; confined to brief notes with each chart.
  - CNB comment:
    - Methodology explained in a box when fan charts were introduced; no plan to include full methodology in every Inflation Report; brief notes deemed sufficient.

- B7. Assessment of forecast revisions (Score: 0.6/1)
  - Findings:
    - Each 2017 Inflation Report included a “Comparison with the previous forecast” section discussing causes of forecast revisions. No revisions published for exchange rate or output gap.
  - CNB comment:
    - Assessment of exchange rate forecast changes reintroduced with resumed publication in February 2018; reintroduced chart breaking down changes in interest rate path by contributions of fundamental factors; no plan to introduce assessment of output gap forecast changes (output gap plays no direct role in core forecasting model).

- B8. Publication of alternative scenarios (Score: 0/1)
  - Findings:
    - Alternative scenarios not regularly published in first 3 Inflation Reports in 2017; last Inflation Report of 2017 included a sensitivity scenario on higher domestic inflation pressures.
  - CNB comment:
    - Occasional publication of alternative scenarios or sensitivity analyses if key risk identified or Board requests it; some scenarios produced for internal use only, particularly during exchange rate commitment period and immediately after exit.

- B9. Publication of financial variables and forecasts (Score: 0.5/1)
  - Findings:
    - Historical data published for: government bond yield curve, consumer lending rates, mortgage rates, property prices, and credit standards (loan officer surveys).
    - Forecast coverage for 5 or more financial variables not fully documented in 2017.

### C. Communication practices and decision-accountability (Questions C1–C6)
- C1. Press statement immediately after policy decisions (Score: 1/1)
  - Findings:
    - CNB published statements after each policy decision in both Czech and English.

- C2. Post-decision press conference and availability in English (Score: 1/1)
  - Findings:
    - Following every policy meeting, CNB held press conferences at pre-announced dates/times; press conferences and Q&A were webcast and made available.
    - Presentations available in Czech and English.
  - CNB comment:
    - Producing an English transcript of the Governor’s remarks may be considered in the future.

- C3. Presentation of regular forecast updates with Q&A; English availability (Score: 1/1)
  - Findings:
    - CNB presented regular forecast updates with Q&A to journalists, analysts, and market participants; presentations published in English; videos available on website.

- C4. Publication of minutes with voting attribution (Score: 0.75/1)
  - Findings:
    - CNB provided detailed minutes with voting results on the main policy instrument attributed to MPC members; contributions by individual MPC members were not recorded in the minutes (voting results also presented separately in an Excel file).
  - Recommendation:
    - Consider fully-attributed minutes as in Riksbank’s “Monetary policy minutes.”
  - CNB comment:
    - Plans to discuss switching to fully-attributed minutes during first half of 2018.

- C5. Clarity of roles of staff and policymakers in the baseline forecast (Score: 1/1)
  - Findings:
    - Ownership of the forecast and roles of staff versus policymakers are clearly defined.
    - Foreword statement in Inflation Report clarifies forecasting schedule and that the forecast is the key, but not the only, input to monetary policy decision-making; Bank Board meets eight times a year with four meetings discussing a new forecast and four meetings discussing risks/uncertainties.

- C6. Annual review of forecasting performance (Score: 1/1)
  - Findings:
    - CNB reviewed forecasting performance in every Inflation Report and in the Annual Report.

*Source: APPENDIX III: CBT-IT INDEX SCORE FOR THE CZECH NATIONAL BANK IN 2017 (wp18210 - APPENDIX III).*

### 7. C7. When was the last time the central bank or the government held or invited an

### 7. C7. When was the last time the central bank or the government held or invited an external evaluation of the policy framework and the FPAS, and made the results publicly available?

### Perfect score requirement
- There is an external and independent evaluation of the policy framework and the FPAS within the last five years. The results are publicly available.

### CNB assessment
- CNB Score: 0/1

### Findings
- There has been no external or independent evaluation of the policy framework or FPAS held at the CNB.

### CNB comments and potential next steps
- Staff from the CNB mentioned that the bank may consider asking the IMF to do an external evaluation in the future, after switching to its planned extended core model.

*Source: wp18210 - 7. C7. When was the last time the central bank or the government held or invited an external evaluation of the policy framework and the FPAS, and made the results publicly available?*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2018/wp18210.pdf_