## Trust What You Hear: Policy Communication, Expectations, and Fiscal Credibility — Working Paper No. WP/22/36

## Source details

**Canonical URL:** [Trust What You Hear: Policy Communication, Expectations, and Fiscal Credibility — Working Paper No. WP/22/36](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022036-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022036-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022036-print-pdf.pdf.json)

---

### Executive summary
- Paper focus: credibility and communication of fiscal policy; how policy communications on future fiscal targets affect market expectations and beliefs about future conduct of fiscal policy.
- Credibility definition: extent to which private agents trust that policymakers will deliver on their mandate or stated objectives; measured by how closely private expectations align with official forecasts.
- Indicators and methodology:
  - Data sources: professional forecasters’ projections on fiscal variables and official budget documents and multi-year fiscal plans.
  - Three fiscal credibility indicators: bias, skepticism, and unanchoring — capturing different aspects of the discrepancy between private and official expectations.
  - Approach: analyze changes in credibility around months when new fiscal targets are announced to capture anchoring effects and reduce confounding from differing update frequencies.
  - Fiscal news measure: difference between the new target and what private forecasts expected a month before announcement.
- Key empirical findings:
  - Official announcements of new fiscal targets improve fiscal credibility on average by 0.2-0.4 percent of GDP when new targets are announced.
  - Only 10 percent of the unexpected component of the new targets is immediately incorporated into private forecasts.
  - Less than half of the unexpected component is eventually incorporated into market beliefs.
  - Disagreements about the macroeconomic outlook (growth projections) are not as important for fiscal credibility as might be expected.
  - Fiscal rules and strong fiscal frameworks reduce the discrepancy between professional and official forecasts and reduce disagreement among private forecasts; budget deficit rules have a stronger effect.
  - Fiscal slippages and downward revisions to fiscal targets worsen credibility.
  - Countries with high fiscal credibility enjoy sovereign yields on average 70 basis points lower than less credible governments.
  - Anchoring effect of credible policy announcements is associated with a sizable decrease in financing costs for high-credibility countries; low-credibility countries show no material change in sovereign yields around announcements.
- Policy implications:
  - Governments should credibly communicate fiscal strategy and demonstrate administrative capacity and political will to anchor private expectations.
  - Strengthening fiscal frameworks and adopting effective fiscal rules—especially budget deficit rules—can improve fiscal credibility.
  - Avoiding fiscal slippages and downward revisions to targets is crucial because credibility is reputation-based and built over time.
  - Improving credibility can yield tangible market benefits, including lower sovereign borrowing costs.

*Source: Executive summary (wpiea2022036-print-pdf - Executive summary7).*

### 2.1 Definition — concept, measures, and treatment of horizons
- Fiscal credibility concept and scope:
  - Defined as the extent to which economic agents expect the government to try and fulfill its fiscal policy commitments, covering intention and ability to achieve targets.
  - Empirical focus on professional forecasters’ views on the fiscal balance.
- Reasons for divergence between private expectations and government announcements:
  - Multiple policy objectives, sustainability concerns, past policy slippages, reputational effects, asymmetric information, and differing macro parameter assessments.
- Anchoring and signaling:
  - Government anchors private beliefs when communications dominate other information and private forecasts put more weight on government information.
  - Focus on difference δ(h)_{i,t,f} ≡ E^f_t b^(h)_{i,t} − E^o_t b^(h)_{i,t}.
- Three credibility indicators:
  - Indicator 1 — Perceived bias:
    - Bias^(h)_{i,t} ≡ 〈 δ^(h)_{i,t,f} 〉_{f∈F} = E^p_t b^(h)_{i,t} − E^o_t b^(h)_{i,t}.
    - By convention, Bias is negative when private investors foresee a larger deficit than the government; this is the case in about 70 percent of the sample.
  - Indicator 2 — Average skepticism:
    - Skept^(h)_{i,t} ≡ || Bias^(h)_{i,t} || ≡ || E^p_t b^(h)_{i,t} − E^o_t b^(h)_{i,t} ||.
  - Indicator 3 — Unanchoring (dispersion):
    - Unanc^(h)_{i,t} ≡ √〈 [ δ^(h)_{i,t,f} ]^2 〉_{f∈F}.
    - Greater credibility associated with less dispersion among private forecasts.
- Cyclical adjustment and horizon treatment:
  - Cyclically-adjusted δ*^(h) = δ^(h) − η_i ∑_{`=0}^h (E^f_t γ^(`)_{i,t} − E^o_t γ^(`)_{i,t}); indicators Bias*, Skept*, Unanc* computed accordingly.
  - Focus on horizons h ∈ {0, 1}; fiscal year data linearly interpolated into calendar years when necessary.
  - Fixed-horizon proxy for one-year-ahead forecasts:
    - b^(⊥)_{i,t} ≡ (12 − m(t)) b^(0)_{i,t} + m(t) b^(1)_{i,t}  /  12, where m(t) ∈ {1, ···, 12}.

*Source: 2.1 Definition (wpiea2022036-print-pdf).*

### 2.3 Stylized characteristics of fiscal credibility — patterns, announcements, and institutional determinants
- Cross-country patterns:
  - Perceived bias often negative, ranging between -2.0 and +0.5 percent ofGDP.
  - Bias is within the [−12;+12] percent ofGDP interval for actual fiscal balance context.
  - Skepticism: cross-country average around 0.8 percent ofGDP, range 0.2 to 2.0 percent ofGDP.
  - Unanchoring: on average 1 percent ofGDP.
  - Pairwise correlations between Bias, Skept, and Unanc are relatively high at 0.8–0.9.
- Cyclical adjustment:
  - Cyclically-adjusted indicators filter out cyclical effects; cyclical effects explain a minor share of average Bias and Unanc.
  - Cyclically-adjusted Bias* has a higher (or less negative) value than non-adjusted Bias for most countries.
- Macroeconomic conditions and past performance:
  - Official targets tend to be more credible when real GDP growth has been high; correlation weakens with cyclically-adjusted bias.
  - Higher government debt-to-GDP associated with worse fiscal credibility.
  - Forecast performance correlations (regression findings, Table C.1):
    - Over-performance: a decrease in Err− by one percentage point ofGDP leads to deterioration of Bias by 0.26 percentage point ofGDP.
    - Fiscal slippage: an increase in Err+ by one percentage point ofGDP lowers Bias by 0.39 percentage point ofGDP.
  - Bin-scatter slope coefficients (controlling for controls; p-value < 1%):
    - Bias: -0.24
    - Skept: 0.20
    - Unanc: 0.26
- Announcements, surprises, and anchoring effects:
  - Private expectations converge toward official targets upon announcement (anchoring effect).
  - Average behavior around release:
    - Difference between private and official forecasts widens up to 0.7 percent ofGDP on the month preceding the announcement.
    - Difference shrinks to less than 0.4 percent ofGDP on impact at T.
  - Announcement associated with reductions in unanchoring and skepticism, but anchoring is partial and fades over subsequent months.
  - At fiscal consolidation announcements, private forecasters expect about 20 percent of announced adjustments to materialize.
- Fiscal news and pass-through:
  - News_{i,t} ≡ E^o_t b(⊥) − E^p_{t−1} b(⊥).
  - Update model: E^p_t b(⊥) − E^p_{t−1} b(⊥) = β News_t; β is the share of policy intentions that agents trust.
  - Estimated pass-through:
    - One month after announcement, only 10 percent of the fiscal surprise is reflected in private forecasts (after controls).
    - Private projections gradually lean toward the official target but even after several months reflect less than half of the fiscal surprise.
    - Heterogeneity by credibility group:
      - High-credibility group: ˆβ_h reaches 0.6.
      - Low-credibility group: pass-through about half of high-credibility group.
- Role of revisions and noise:
  - Controlling for Rev(⊥)_t ≡ E^o_t b(⊥) − E^o_{t−1} b(⊥):
    - β coefficients remain positive and even stronger when controlling for revision size.
    - γ coefficients (on Rev) are negative: large revisions can confuse private forecasters and harm credibility.
    - Implication: more frequent, smaller revisions might be preferable for building credibility.
- Institutional supports:
  - Regression approach: changes in credibility indicators at months of new fiscal targets regressed on macro controls and institutional variables using FGLS.
  - Fiscal rules and frameworks:
    - Debt rules decrease discrepancy between private and official targets by about 2 percentage point ofGDP based on Bias and Skept.
    - Budget balance rule has a similar impact on Skept.
    - Interaction: debt rules dampen negative impact of fiscal surprise on Bias by 0.4 percent ofGDP (≈ 54.6% ≈ 0.498/0.911 of the negative impact).
    - Debt rules reduce disagreement among private forecasters by about 0.8 percent ofGDP.
  - Fiscal agencies and legal frameworks:
    - Existence of fiscal agencies or independent watchdogs diminishes the negative impact of fiscal surprises on fiscal credibility by more than 1 percentage point ofGDP (except for Unanc).
    - Enshrining fiscal rules in law (versus political commitment alone) improves credibility and reduces skepticism.
  - Multiyear spending ceilings (top-down budgeting) dampen negative impact of fiscal surprises for all credibility indicators.
  - Effects hold after controlling for the political cycle.

*Source: 2.3 Stylized characteristics of fiscal credibility (wpiea2022036-print-pdf).*

### 4.2 Fiscal transparency and communication — channels, measures, and empirical links
- Role of fiscal transparency and communication:
  - Fiscal transparency is essential for fiscal credibility because credibility concerns convincing market participants of future fiscal policy.
  - Format of policy communications (posting in official bulletin vs. proactive speeches/blogs/media outreach) could matter for credibility.
- Data source and measurement:
  - Uses the Open Budget Survey (OBS) and Open Budget Index (OBI) on a 0–100 scale.
  - Focus on ten OBS questions related to publication of key documents, citizens budgets, independent control institutions, and content/accountability of fiscal projections.
  - Caution on measurement error for individual country scores.
- Empirical findings:
  - Effective public communication of budgets and fiscal objectives correlates with higher fiscal credibility across the three indicators.
  - Higher scores on publication of pre-budget statements (PBSs) associated with higher credibility.
  - Quality and accessibility of citizens budgets (OBI sub-index, Q65, Q67) associated with higher credibility.
  - No significant link found between credibility and other budget documents (year-end reports, budget proposals, audit reports) likely due to little variation across countries.
- Independent budgetary institutions (IBIs):
  - Countries where IBIs can conduct macro-fiscal projections (Q104), carry out objective budget cost analysis (Q105), and publicly present views at budget committees (Q106) exhibit higher credibility.
- Content of official communications:
  - Explicit recognition of expenditure arrears (Q41) and publishing debt projections (Q57) decrease Bias increases and Skept and Unanc indicators.
  - Early PBS articulation of revenue and spending intentions helps reduce perceived bias and disagreements.
- Summary implications:
  - High-quality PBSs and citizens budgets, empowered IBIs publishing forecasts/costings, and publishing concrete fiscal content reduce private-forecaster bias and disagreement.

*Source: 4.2 Fiscal transparency and communication (wpiea2022036-print-pdf).*

### Market rewards, data, and regression evidence — magnitudes and statistical findings
- Market rewards of fiscal credibility:
  - Lower credibility (higher absolute difference between private and official targets) associated with higher sovereign yields, controlling for macro factors.
  - Fiscal announcements on average trigger a 40 basis point drop in sovereign yields for high- and low-credibility countries alike.
  - Reduction in yields associated with consolidation announcements smaller for low-credibility countries relative to high-credibility countries: 20 vs. 40 bps.
  - Large consolidations lower sovereign yields only for more credible governments.
  - Fiscal expansion announcements reduce financing costs for credible governments but do not change financing costs for less credible governments.
- Empirical magnitudes and selected regression coefficients (Table C.8 and related):
  - CDS spreads are around 2-3 percentage points higher on average for a level of skepticism or a perceived bias of 0.5 percent of GDP.
  - Selected coefficients (exact values reported in Table C.8):
    - Skept(⊥) (lagged): 6.059*** (5-yr CDS), 5.356*** (10-yr CDS), 0.138*** (2-yr yield), 0.383*** (10-yr yield), -0.0313*** (Rating).
    - Bias(⊥) (lagged): -4.707*** (5-yr CDS), -4.028*** (10-yr CDS), -0.0248 (2-yr yield), -0.116*** (10-yr yield), 0.0216** (Rating).
    - Unanc(⊥) (lagged): 12.22*** (5-yr CDS), 9.636*** (10-yr CDS), 0.230*** (2-yr yield), 0.468*** (10-yr yield), -0.0224* (Rating).
- Key regression and robustness results (selected):
  - Non-linear relationship between forecast errors and fiscal credibility (Table C.1):
    - Err_{-} coefficient: -0.261***; Err_{+}: -0.390***. Observations: 4,728 in reported columns.
  - Fiscal rules, frameworks, and credibility (Table C.2):
    - News coefficient large and negative across specifications (e.g., News = -0.775***; News = -1.468***).
    - Interaction effects (Debt Rule × News, Budget Bal. Rule × News, Monitoring Institution × News, Legal basis × News, Multiyear expenditure ceiling × News) show significant moderation of News impacts.
  - Public engagement and publication (Table C.3):
    - OBI of Citizen budget × News: 0.00597*** (ΔBias).
    - Q65 × News: 0.00363*** (ΔBias).
    - Q67 × News: 0.00458** (ΔBias).
  - Independent fiscal institutions (Table C.4):
    - News strongly associated with reductions in Bias and Unanc (e.g., News -0.685*** in ΔBias).
    - Q104 × News: 0.007*** (ΔSkept); Q106 × News: 0.009*** (ΔBias).
  - Transparency and content (Table C.5):
    - News strong negative coefficients (e.g., News -0.950*** in ΔBias).
    - Q41 × News: 0.0108*** (ΔSkept) and -0.00625*** (ΔUnanc).
    - Q57 × News: 0.00822*** (ΔBias) and -0.00592*** (ΔUnanc).
  - Government track records (Table C.6):
    - News: -0.284*** (ΔBias).
    - Rev(⊥)_t: -0.862*** (ΔBias).
  - Composition of announced plans (Table C.7):
    - Consolid = 1: -0.486*** (ΔBias); (Consolid = 1) × News: 0.319*** (ΔBias).
    - LargeConsolid = 1: -0.616*** (ΔBias); (LargeConsolid = 1) × News: 0.484*** (ΔBias).
- Data sources and coverage:
  - Official budget documents: country-specific sources; examples include Argentina January 2002; Australia July 1994; United States October 1996 (see Table B.1).
  - Private forecasts: Monthly Consensus Economics publications, Bloomberg surveys.
  - Official forecasts: Budget documents, IMF program reviews; EU Stability and Convergence Programs, Draft budgetary plans.
  - Macro/fiscal variables: WEO or Eurostat.
  - Fiscal rule design: IMF fiscal rule database (Lledó et al., 2017).
  - Fiscal councils: IMF Fiscal council dataset (Debrun et al., 2013; Beetsma et al., 2019).
  - Sovereign CDS spreads and currency asset swap spreads: Bloomberg.
  - Sovereign credit ratings: DBRS Morningstar, Moody’s, Fitch, S&P.
  - Uncertainty index: Ahir, Bloom, and Furceri (2018).
- Open Budget Survey questionnaire items (excerpts relevant to findings):
  - Citizens budgets / communication: Q65, Q67.
  - IBIs: Q104, Q105, Q106.
  - Accountability / official projections: Q41, Q54, Q55, Q56, Q57.

*Source: wpiea2022036-print-pdf — References and appendices (A–D and Tables).*

### Policy implications (as presented)
- Strengthen fiscal communication and demonstrate capacity and political will to anchor private expectations.
- Adopt and enforce fiscal rules and supporting institutions—especially budget-deficit and debt rules, fiscal agencies, and legally enshrined rules—to reduce bias, skepticism, and unanchoring.
- Publish high-quality pre-budget statements (PBSs) and accessible citizens budgets; empower IBIs to publish forecasts and costings and participate publicly in budget scrutiny.
- Provide concrete fiscal content early (expenditure arrears recognition, debt projections, early revenue and spending intentions).
- Prefer predictability through stable, small revisions over large unexpected revisions to build credibility over time.
- Improving credibility can materially lower sovereign borrowing costs and expand fiscal space, enabling more effective fiscal policy implementation.

*Source: wpiea2022036-print-pdf — Working Paper No. WP/22/36.*

### Executive summary7

### Executive summary

### Overview
- Paper focus: credibility and communication of fiscal policy; how policy communications on future fiscal targets affect market expectations and beliefs about future conduct of fiscal policy.
- Credibility definition: extent to which private agents trust that policymakers will deliver on their mandate or stated objectives; measured by how closely private expectations align with official forecasts.

### Indicators and methodology
- Data sources: professional forecasters’ projections on fiscal variables and official budget documents and multi-year fiscal plans.
- Three fiscal credibility indicators introduced: bias, skepticism, and unanchoring — capturing different aspects of the discrepancy between private and official expectations.
- Approach: analyze changes in credibility around months when new fiscal targets are announced to capture anchoring effects and reduce confounding from differing update frequencies of private vs. official forecasts.
- Fiscal news measure: constructed as the difference between the new target and what private forecasts expected a month before it was announced.

### Key empirical findings
- Anchoring effect of announcements:
  - Official announcements of new fiscal targets improve fiscal credibility on average by 0.2-0.4 percent of GDP when new targets are announced.
  - Only 10 percent of the unexpected component of the new targets is immediately incorporated into private forecasts.
  - Less than half of the unexpected component is eventually incorporated into market beliefs.
- Role of macroeconomic disagreements:
  - Disagreements about the macroeconomic outlook (growth projections) are not as important for fiscal credibility as might be expected.
- Institutional and historical determinants:
  - Fiscal rules and strong fiscal frameworks reduce the discrepancy between professional and official forecasts and reduce disagreement among private forecasts.
  - Budget deficit rules have a stronger effect on credibility because they directly target the fiscal balance linked to the credibility indicators.
  - Fiscal slippages and downward revisions to fiscal targets worsen credibility, consistent with reputation effects built over time.
- Market rewards:
  - Countries with high fiscal credibility enjoy sovereign yields on average 70 basis points lower than less credible governments.
  - The anchoring effect of credible policy announcements is associated with a sizable decrease in financing costs for high-credibility countries, whereas low-credibility countries show no material change in sovereign yields around announcements.

### Interpretation and links to literature
- Credibility operates through an expectation channel analogous to monetary policy forward guidance; anchoring fiscal expectations can reduce risk premia and facilitate fiscal solvency.
- The findings align with literature showing that fiscal rules and frameworks provide discipline and predictability, though causal inference is challenging due to endogeneity.
- The paper extends prior work by expanding country coverage beyond Europe, combining disagreement and skepticism into a summary credibility measure, and deepening analysis of policy announcements’ effects on professional forecasts.

### Policy implications
- Governments should aim to credibly communicate fiscal strategy and demonstrate administrative capacity and political will to anchor private expectations.
- Strengthening fiscal frameworks and adopting effective fiscal rules—especially budget deficit rules—can improve fiscal credibility.
- Avoiding fiscal slippages and downward revisions to targets is crucial because credibility is reputation-based and built over time.
- Improving credibility can yield tangible market benefits, including lower sovereign borrowing costs.

*Source: Executive summary (wpiea2022036-print-pdf - Executive summary7).*

### 2.1    Definition

### 2.1    Definition

### Fiscal credibility: concept and scope
- Fiscal credibility is defined as the extent to which economic agents expect the government to try and fulfill its fiscal policy commitments, at least within a reasonable margin of error and with a certain likelihood.
- The concept covers two aspects: the intention and ability to achieve targets.
- This paper focuses on the credibility of budgets and the anchoring of expectations around government fiscal targets.
- The empirical focus is on educated agents—professional forecasters—and their views on the fiscal balance, the most visible headline indicator of fiscal policy.

### Reasons for divergence between private expectations and government announcements
- Fiscal policy has many objectives that are less clearly spelled out and prioritized than monetary policy, leading to uncertainty about government intentions.
- Agents may expect governments not to keep promises when stated policy objectives appear unsustainable or unrealistic; they use past policy slippages and government reputation (e.g., preference for austerity or profligacy) to assess likelihood of meeting targets.
- Agents may believe they have better information or forecasting tools than governments (e.g., proprietary market data); governments are often expected to have a longer time horizon and tend to be more optimistic.
- Disagreements about parameters underpinning the budget (macroeconomic environment, commodity prices) are common and can stem from diverging assessments of the current situation and the distribution of risks.
- Informational imperfections and asymmetries in the expectation formation process produce disagreement among private forecasters; disagreement can be modeled as Bayesian learning amid noisy information.

### Anchoring and the government’s signaling role
- A credible government anchors private beliefs around official targets when its communications dominate other sources of information and private forecasts put more weight on government information than on other signals.
- The methodology expands on End (2020) and focuses on the difference between private forecasters’ expectations and official policy targets in terms of the fiscal balance expected at time t for country i for forecast horizon h:
  - δ(h)_{i,t,f} ≡ E^f_t b^(h)_{i,t} − E^o_t b^(h)_{i,t}
- Three complementary indicators of credibility are used, all measuring the anchoring of private expectations.

### Indicator 1 — Perceived bias
- The perceived bias of country i at time t is the average, across the set of forecasters F, of the relative difference between private forecasts and official targets:
  - Bias^(h)_{i,t} ≡ 〈 δ^(h)_{i,t,f} 〉_{f∈F} = E^p_t b^(h)_{i,t} − E^o_t b^(h)_{i,t}
- By convention, Bias is negative when on average private investors foresee a larger deficit than the government; this is the case in about 70 percent of the sample.
- The bias measure treats government (perceived) optimism and pessimism asymmetrically; whether markets should expect more or less deficit than governments depends on policy context (e.g., fiscal consolidation vs fiscal stimulus).

### Indicator 2 — Average skepticism
- Average skepticism is the absolute discrepancy between government and market forecasts:
  - Skept^(h)_{i,t} ≡ || Bias^(h)_{i,t} || ≡ || E^p_t b^(h)_{i,t} − E^o_t b^(h)_{i,t} ||
- This indicator captures the magnitude of disagreement irrespective of sign.

### Indicator 3 — Unanchoring (dispersion)
- Unanch^(h)_{i,t} captures the dispersion of private forecasts around official government targets:
  - Unanc^(h)_{i,t} ≡ √〈 [ δ^(h)_{i,t,f} ]^2 〉_{f∈F}
- Greater credibility should be associated with less dispersion among private forecasts; the more credible the government, the more expectations should converge.
- This measure links disagreements to the credibility of policy announcements rather than viewing dispersion only as intrinsic forecast uncertainty.

### Cyclical adjustment and horizon treatment
- Measures implicitly include disagreements about the macroeconomic environment and about budget aggregates given macro assumptions.
- Cyclically-adjusted indicators are computed as a robustness check by replacing δ^(h) with δ*^(h) = δ^(h) − η_i ∑_{`=0}^h (E^f_t γ^(`)_{i,t} − E^o_t γ^(`)_{i,t}), where η_i is the country-specific elasticity of the fiscal balance to the output gap.
- Cyclically-adjusted versions of Bias, Skept, and Unanc are denoted Bias*, Skept*, and Unanc*.
- Focus is on horizons h ∈ {0, 1} (current and next years, calendar years); when necessary, fiscal year data are linearly interpolated into calendar years.
- To minimize time-series discontinuities, a fixed-horizon proxy for one-year-ahead forecasts (denoted h = ⟂) is defined as a weighted average of current and next year forecasts:
  - b^(⊥)_{i,t} ≡ (12 − m(t)) b^(0)_{i,t} + m(t) b^(1)_{i,t}  /  12
  - m(t) ∈ {1, ···, 12} stands for the calendar month of date t.

*Source: wpiea2022036-print-pdf — 2.1 Definition*

### 2.3    Stylized characteristics of fiscal credibility

### 2.3    Stylized characteristics of fiscal credibility

### Cross-country patterns of credibility indicators
- Perceived bias:
  - Often negative, ranging between -2.0 and +0.5 percent ofGDP.
  - Bias is within the [−12;+12] percent ofGDP interval for actual fiscal balance context.
- Skepticism (not cyclically adjusted):
  - Cross-country average hovers around an average of 0.8 percent ofGDP and ranges from 0.2 to 2.0 percent ofGDP.
- Unanchoring (disagreement among private forecasters):
  - Unancis on average 1 percent ofGDP.
- Correlations and heterogeneity:
  - Pairwise correlations between the three credibility indicators are relatively high at 0.8–0.9.
  - Time-series variations are non-negligible.
- Cyclical adjustment:
  - Cyclically-adjusted indicators (red bars in Figure 4) filter out cyclical effects to proxy credibility for given macroeconomic assumptions.
  - Cyclical effects explain a minor share of the average Bias and Unanc indicators, implying expectation gaps are mostly driven by divergent views on fiscal projections rather than growth.
  - Cyclically-adjusted Bias* has a higher (or less negative) value than non-adjusted Bias for most countries.
  - Skept is higher than Skept* for some countries, but lower for others.

### Macroeconomic conditions, past performance, and credibility
- Macro correlations:
  - Official targets tend to be more credible when real GDP growth has been high; correlation weakens when cyclically-adjusted bias is used.
  - Higher government debt-to-GDP level is associated with worse fiscal credibility (negative correlation).
  - Professional forecasters penalize governments that indulge in large fiscal deficits.
- Forecast performance:
  - Credibility measures strongly correlate with past forecast performance.
  - Regression findings (Table C.1):
    - Making forecast errors, in either direction, worsens fiscal credibility in a non-linear manner.
    - An over-performance of fiscal outcome (a decrease in Err− by one percentage point ofGDP) leads to a deterioration of Bias by 0.26 percentage point ofGDP.
    - A fiscal slippage (an increase in Err+ by one percentage point ofGDP) lowers Bias by 0.39 percentage point ofGDP.
  - Bin-scatter slope coefficients (Figure 5), controlling for lagged real GDP growth, public debt-to-GDP ratio, primary balance, and fixed effects:
    - Bias: -0.24
    - Skept: 0.20
    - Unanc: 0.26
    - p-value lower than 1%.

### Announcements, surprises, and anchoring effects
- Credibility boost around announcements:
  - Private expectations converge toward official targets upon announcement (anchoring effect).
  - Average behavior around release (Figure 6a):
    - Difference between private and official forecasts widens up to 0.7 percent ofGDP on the month preceding the announcement.
    - Difference shrinks to less than 0.4 percent ofGDP on impact at T.
  - Announcement is associated with reductions in unanchoring and skepticism, but anchoring is partial and fades over subsequent months.
  - At fiscal consolidation announcements, private forecasters expect about 20 percent of announced adjustments to materialize (Figure 7).
- Fiscal news (surprise) construct:
  - News_{i,t} ≡ E^o_t b(⊥) − E^p_{t−1} b(⊥) (equation (5)).
  - News is treated as an exogenous shock reflecting the component of new official targets that private forecasters did not expect.
- Imperfect pass-through of fiscal news to private forecasts:
  - Model of update: E^p_t b(⊥) − E^p_{t−1} b(⊥) = β News_t (equation (6)); β is the share of policy intentions that agents trust.
  - Local projection regression (equation (7)) is used to estimate β_h for horizons h ∈ {0, ..., 5}.
  - Estimated pass-through (Figure 9):
    - One month after announcement, only 10 percent of the fiscal surprise is reflected in private forecasts (after controls).
    - Private projections gradually lean toward the official target but even after several months reflect less than half of the fiscal surprise.
    - Heterogeneity by credibility group (Appendix Figure D.2):
      - High-credibility group: ˆβ_h reaches 0.6.
      - Low-credibility group: pass-through is about half of the high-credibility group.
- Role of revisions and noise:
  - Controlling for size of official target revisions Rev(⊥)_t ≡ E^o_t b(⊥) − E^o_{t−1} b(⊥).
  - Regression (equation (8)) includes News and Rev(⊥).
  - Findings (Figure 10):
    - β coefficients remain positive and are even stronger when controlling for revision size.
    - γ coefficients (on Rev) are negative: large revisions can confuse private forecasters and bear negatively on credibility.
    - Implication: more frequent, smaller revisions might be preferable to building credibility over time.

### Institutional supports for fiscal credibility
- Empirical approach:
  - Change in credibility indicators at months of new fiscal target releases is regressed on macro controls and institutional variables using FGLS (equation (9)).
  - Controls include country, year, month fixed effects, lagged real GDP growth, private expectations for next year real GDP growth, lagged primary balance, public debt ratio, country-specific uncertainty index, policy rate, inflation, election dummy, IMF program dummy.
  - Z_{i,t} includes institutional factors: fiscal rules, frameworks, fiscal transparency, content of announced plans, past policy record, announced future plans.
- Fiscal rules and institutions (summary of regressions, Table C.2 and discussion):
  - Debt rules and budget-balance rules:
    - Found to exert direct effects improving credibility when new budgets are announced.
    - Debt rules decrease the discrepancy between private and official targets by about 2 percentage point ofGDP based on Bias and Skept.
    - Budget balance rule has a similar impact on Skept.
    - Interaction terms: debt rules dampen the negative impact of fiscal surprise on credibility (Bias) by 0.4 percent ofGDP, equivalent to about half (54.6% ≈ 0.498/0.911) of the negative impact from fiscal surprise.
    - Debt rules reduce disagreement among private forecasters by about 0.8 percent ofGDP.
  - Fiscal agencies and legal frameworks:
    - Existence of fiscal agencies or independent watchdogs that monitor budgets and fiscal rules diminishes the negative impact of fiscal surprises on fiscal credibility by more than 1 percentage point ofGDP (except for Unanc).
    - Enshrining fiscal rules in law (versus political commitment alone) improves credibility and reduces skepticism.
  - Multiyear spending ceilings (top-down budgeting):
    - Having multiyear spending limits contributes to fiscal credibility by dampening the negative impact of fiscal surprises for all indicators of fiscal credibility.
- Robustness to political cycle:
  - Effects of fiscal rules and institutions on credibility hold even after controlling for the political cycle.
- Interpretation:
  - Fiscal rules and strong fiscal institutions tend to improve the anchoring of private expectations on official targets, though different types of rules/frameworks affect different credibility aspects variably.
  - Large unexpected revisions can undermine credibility; institutional mechanisms that promote predictability and constrain discretionary fiscal deviations support credibility.

*Source: 2.3 Stylized characteristics of fiscal credibility, wpiea2022036-print-pdf*

### 4.2    Fiscal transparency and communication

### 4.2    Fiscal transparency and communication

### Role of fiscal transparency and communication
- Fiscal transparency is presented as an essential ingredient for fiscal credibility because credibility concerns the government’s ability to convince market participants of future fiscal policy.
- The format of policy communications (e.g., posting a document in the official bulletin vs. proactive speeches, blogs, and media outreach) could matter for credibility.
- Measuring fiscal transparency and communication is challenging; existing quantitative indicators examine features of budget documents and budget processes.

### Data source and measurement
- The analysis draws on the Open Budget Survey (OBS) published by the International Budget Partnership, a biennial survey covering availability and publication of fiscal information, information quality, and presence of independent experts on budget and tax policies.
- The OBS uses 100+ categorical questions that are spliced into the Open Budget Index (OBI) ranked on a 0–100 scale.
- The authors focus on a subset of ten OBS questions related to:
  - publication of key documents and the quality and importance of citizens budgets in the budget process;
  - existence, quality, and role of independent control institutions;
  - content and accountability of fiscal projections.
- Caution: “Caution is needed in interpreting scores for any individual country given measurement error because the quality of underlying data can vary across countries and data sources.”

### Empirical findings on communication and credibility
- Governments that communicate budgets and fiscal objectives effectively with the public have higher fiscal credibility.
- Higher scores on the publication of pre-budget statements (PBSs) are associated with higher credibility according to the three credibility indicators used in the study.
- The quality and accessibility of citizens budgets (OBI sub-index for Citizens Budgets and questions Q65 and Q67) are associated with higher credibility.
- No significant link was found between credibility and variables related to other budget documents (year-end reports, budget proposals, audit reports), likely because most countries release these documents, yielding little variation.
- The OBI sub-index reflecting the quality, comprehensiveness, and transparency of PBSs improves fiscal credibility.

### Independent budgetary institutions (IBIs)
- IBIs theoretically act as counter-powers to government discretion and should improve predictability of fiscal policy.
- OBS questions related to IBIs show that countries allowing their IBIs to:
  - conduct macro-fiscal projections (Q104),
  - carry out an objective budget cost analysis (Q105),
  - present publicly their views at budget committees (Q106),
  also exhibit higher credibility.

### Content of official communications
- Specific content elements of official communications matter for fiscal credibility:
  - Governments that explicitly recognize expenditure arrears (Q41) and publish debt projections (Q57) see decreases in Biasincreases and Ske ptandUnancindicators.
  - Describing early, in PBSs, intentions in terms of revenue and spending helps reduce perceived bias and disagreements among private forecasters.

### Summary of implications from section
- Effective public communication of budget intentions and forecasts—through high-quality PBSs and citizens budgets—correlates with higher fiscal credibility.
- Empowered and transparent IBIs that perform projections, cost analyses, and publicly present views are associated with improved credibility.
- Publishing concrete fiscal content (expenditure arrears recognition, debt projections, early articulation of revenue and spending plans) reduces measures of private-forecaster bias and disagreement.

*Source: wpiea2022036-print-pdf - 4.2    Fiscal transparency and communication*

### References

### wpiea2022036-print-pdf - References (Appendices A–D and Tables)

### Market rewards of fiscal credibility
- Higher values of the absolute difference between private and official targets (lower credibility) are associated with higher sovereign yields, even when controlling for macroeconomic factors.
- Expected sovereign yields almost perfectly track actual market yields (Figure A.1b).
- Fiscal announcements on average trigger a 40 basis point drop in sovereign yields for high- and low-credibility countries alike (Figure A.2a).
- The reduction in yields associated with consolidation announcements is smaller for low-credibility countries relative to high-credibility countries: 20 vs. 40 bps (Figure A.2b).
- Large consolidations lower sovereign yields only for more credible governments (Figure A.2c).
- Fiscal expansion announcements reduce financing costs for credible governments but do not change financing costs for less credible governments (Figure A.2d).
- Panel regression specification used to relate market perceptions of sovereign risk to credibility:
  - y_{i,t} = φ_{y(t)} + φ_{i} + β Cred_{(⊥) i,t} + Θ X_{i,t} + ε_{i,t} (Equation A.1).
- Empirical magnitudes:
  - CDS spreads are around 2-3 percentage points higher on average for a level of skepticism or a perceived bias of 0.5 percent of GDP (Table C.8 discussion).
- Table C.8 (selected reported coefficients, lagged predictors):
  - Skept(⊥) (lagged): 6.059*** (5-yr CDS), 5.356*** (10-yr CDS), 0.138*** (2-yr yield), 0.383*** (10-yr yield), -0.0313*** (Rating) — exact values reported in Table C.8.
  - Bias(⊥) (lagged): -4.707*** (5-yr CDS), -4.028*** (10-yr CDS), -0.0248 (2-yr yield), -0.116*** (10-yr yield), 0.0216** (Rating) — exact values reported in Table C.8.
  - Unanc(⊥) (lagged): 12.22*** (5-yr CDS), 9.636*** (10-yr CDS), 0.230*** (2-yr yield), 0.468*** (10-yr yield), -0.0224* (Rating) — exact values reported in Table C.8.

### Data sources and coverage
- Official budget documents: country-specific sources and start coverage listed in Table B.1 (e.g., Argentina January 2002; Australia July 1994; United States October 1996).
- Data coverage (Table B.2) provides first available forecast dates, # of observations, and group classification for many countries (examples preserved in table).
- Macro, fiscal, and institutional variables (Table B.3) and their sources include:
  - Private forecast of general government’s overall balance: percent of GDP, Monthly Consensus Economics publications, Bloomberg surveys.
  - Official forecast of general government’s overall balance: percent of GDP, Budget documents, IMF program reviews; for EU countries, Stability and Convergence Programs, Draft budgetary plans.
  - Output gap, Real GDP growth, Public debt, Overall balance, Non-interest balance, Long-term yield (10-year T-bond), GDP per capita: WEO or Eurostat as indicated.
  - Fiscal rule design: IMF fiscal rule database (Lledó et al., 2017).
  - Fiscal councils: IMF Fiscal council dataset (Debrun et al., 2013; Beetsma et al., 2019).
  - Sovereign CDS spreads and currency asset swap spreads: last price, Bloomberg.
  - Sovereign credit ratings: DBRS Morningstar, Moody’s, Fitch, S&P.
  - Uncertainty index: Ahir, Bloom, and Furceri (2018).

### Open Budget Survey questionnaire items (excerpt)
- Citizens budget / Communication:
  - Q65: How is the Citizens budget disseminated to the public?
  - Q67: Are “citizens” versions of budget documents published throughout the budget process?
- Independent budget institution (IBI) as counter-power:
  - Q104: Does the IBI publish macroeconomic and/or fiscal forecasts?
  - Q105: Does the IBI publish its own costings of new policy proposals, to assess their impact on the budget?
  - Q106: In the past 12 months, how frequently did the head or a senior staff member of the IBI take part and testify in hearings of a committee of the legislature?
- Accountability: transparency and quality of official projections:
  - Q41: Do draft budget or budget doc present estimates of expenditure arrears for at least the budget year?
  - Q54: Do PBSs present information on the macroeconomic forecast upon which the budget projections are based?
  - Q55: Do PBSs present information on the government’s expenditure policies and priorities that will guide the development of detailed estimates for the upcoming budget?
  - Q56: Do PBSs present information on the government’s revenue policies and priorities that will guide the development of detailed estimates for the upcoming budget?
  - Q57: Do PBSs present three estimates related to government borrowing and debt: net new borrowing required during the budget year; total debt outstanding at the end of the budget year; and interest payments on the debt for the budget year?
- Note: www.internationalbudget.org/open-budget-survey provides further information (reference preserved from table note).

### Key regression and robustness results (selected)
- Non-linear relationship between forecast errors and fiscal credibility (Table C.1):
  - Err_{-} coefficient: -0.261*** (column (1)); Err_{+}: -0.390*** (column (2)). Observations reported: 4,728 (columns 1,2) and other counts as in Table C.1.
- Fiscal rules, frameworks, and credibility (Table C.2):
  - News coefficient is large and negative across many specifications (e.g., News = -0.775*** in column (1); News = -1.468*** in column (3); exact coefficients preserved in Table C.2).
  - Interactions (Debt Rule × News, Budget Bal. Rule × News, Monitoring Institution × News, Legal basis × News, Multiyear expenditure ceiling × News) show statistically significant moderation effects in several specifications (exact coefficients in Table C.2).
- Public engagement and publication of budget documents (Table C.3):
  - News coefficients remain significant and negative in many specifications (e.g., News = 0.0565 in column (1) but significant negatives in other ΔSkept/ΔUnanc columns; exact values in Table C.3).
  - OBI of Citizen budget × News: 0.00597*** (column (8), ΔBias) and other interaction coefficients reported in Table C.3.
  - Q65 × News: 0.00363*** (column (6), ΔBias) with related coefficients for ΔSkept and ΔUnanc in Table C.3.
  - Q67 × News: 0.00458** (column (10), ΔBias) and negative significant interactions for ΔSkept and ΔUnanc (exact values in Table C.3).
- Independent fiscal institutions (Table C.4):
  - News is strongly associated with reductions in Bias and Unanc in several specifications (e.g., News -0.685*** in column (1) ΔBias).
  - Q104 × News: 0.007*** (column (2) ΔSkept), Q105 × News: not significant for ΔBias but negative for ΔSkept in column (5), Q106 × News: 0.009*** (column (6) ΔBias) — see Table C.4 for full set.
- Transparency and quality of budget forecasts (Table C.5):
  - News coefficients are strongly negative in many specifications (e.g., News -0.950*** in column (1) ΔBias).
  - Q41 × News: 0.0108*** (column (2) ΔSkept) and -0.00625*** (column (3) ΔUnanc).
  - Q55 × News: 0.00685** (column (8) ΔBias) and negative interactions for ΔSkept and ΔUnanc in Table C.5.
  - Q56 × News: 0.00902*** (column (11) ΔBias) and negative interactions for ΔSkept and ΔUnanc.
  - Q57 × News: 0.00822*** (column (13) ΔBias) and negative interaction -0.00592*** (column (15) ΔUnanc).
- Government track records and credibility (Table C.6):
  - News: -0.284*** (column (1) ΔBias).
  - Rev(⊥)_t: -0.862*** (column (1) ΔBias), Rev(⊥)_t × News: -0.005 (column (1) ΔBias) and positive significant interactions for ΔSkept and ΔUnanc (exact values in Table C.6).
- Composition of announced fiscal plans and credibility (Table C.7):
  - Consolid = 1: -0.486*** (ΔBias, column (1)); (Consolid = 1) × News: 0.319*** (column (2) ΔBias).
  - LargeConsolid = 1: -0.616*** (ΔBias, column (3)); (LargeConsolid = 1) × News: 0.484*** (column (4) ΔBias).
  - LargeMTConsolid = 1: -0.283 (ΔBias, column (13)); (LargeMTConsolid = 1) × News: 0.290*** (column (14) ΔBias).
  - Observations for these regressions: 424 (columns covering announcements); full counts in Table C.7.
- Sovereign borrowing costs and fiscal credibility (Table C.8 summary):
  - High explanatory power in models (reported R^2 values up to 0.994 in some specifications).
  - Uncertainty, growth (lagged), and inflation show large, statistically significant coefficients across sovereign spreads and yields (exact coefficient values preserved in Table C.8).

### Policy implications and interpretation (as presented)
- Fiscal credibility reduces sovereign borrowing costs and is associated with lower CDS spreads and sovereign yields.
- Credible governments enjoy more fiscal space and flexibility, including the ability to implement fiscal expansions without immediate market penalties.
- Credibility anchors expectations, which improves the effectiveness of fiscal policy (e.g., public spending shocks have more effective impact on growth when expectations are more strongly anchored, citing Ricco, Callegari, and Cimadomo, 2016).
- Loss of credibility can lead to delayed consumption and investment, higher precautionary savings, informality, erosion of tax bases, and worse fiscal outcomes.
- Institutional features (fiscal rules, monitoring institutions, publication of citizens budgets, IBIs publishing forecasts/costings, transparency of PBSs) interact with fiscal news to influence measures of credibility (Bias, Skept, Unanc) as shown in the interaction terms across multiple regression tables.

*Trust What You Hear: Policy Communication, Expectations, and Fiscal Credibility — Working Paper No. WP/22/36*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022036-print-pdf.pdf_