## ANNEX I. ROBUSTNESS ANALYSIS

## Source details

**Canonical URL:** [ANNEX I. ROBUSTNESS ANALYSIS](https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024257-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2024/english/wpiea2024257-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2024/english/wpiea2024257-print-pdf.pdf.json)

---

### I. Introduction — context and motivation
- Financial inclusion defined as the level of access and use of financial services by households and firms; essential to fostering more sustainable and inclusive growth.
- Financial inclusion encourages shifts toward more efficient intertemporal allocation of resources: increased savings, reduced reliance on self-financing, enhanced incentives for productive investments, and deeper markets for goods and services.
- Closing financial inclusion gaps—especially financial deepening—remains a key challenge for countries in the Caucasus and Central Asia (CCA).
- The 2021 Findex Survey highlighted trust in banks as a major barrier to bank account penetration in the region.
- Multiple external shocks over the last decade have tested confidence in the CCA’s macroeconomic policy and institutional frameworks.

### II. Confidence as a driver of financial inclusion
- Public confidence in banking institutions and the broader economy drives both demand for and supply of financial services; banking is inherently based on trust amid asymmetric information.
- Higher confidence increases demand for formal bank savings.
- Confidence is fragile after macro-financial crises; loss of confidence can lead to contagious bank runs and durable retrenchment in financial intermediation.
- Conceptual channels emphasized:
  - Economic policy uncertainty weakens public confidence → reduces demand for formal bank savings.
  - Policy uncertainty incentivizes banks to reduce risk exposure → supply fewer financial services.
  - Reduced policy uncertainty fosters financial inclusion via improvements in banking stability, financial system efficiency, cross-border capital flows, and lower risk premia.
- Institutional and policy enablers highlighted:
  - Good governance mitigates economic policy uncertainty.
  - Inflation targeting provides a commitment device through transparency and accountability.
  - Fiscal rules improve sovereign risk perceptions and fiscal transparency.
  - Strong banking supervision, regulation, and depositor protection support confidence.
  - Deposit insurance effects are conditional: negative in lax regulatory environments, positive in sound regulatory environments.

### III. Literature gap and contribution
- The confidence–financial inclusion nexus remains relatively unexplored despite literature on policy drivers of financial inclusion.
- Related points cited:
  - Non-linear relationship between financial inclusion and financial stability (Sahay and others, 2015; Čihák and others, 2016).
  - Benefits of digital financial inclusion for growth (Khera and others, 2021; Aguilar and others, 2024).
  - SME financial inclusion boosts growth in CCA (Blancher and others, 2019).
  - Human development and rule of law essential for financial inclusion in the CCA (Poghosyan, 2022; 2023).
- This paper examines the nexus between public confidence and financial inclusion in the CCA region, considering confidence in banks and confidence in the broader economy.

### IV. Stylized facts (measurement and benchmarks)
- Financial inclusion measures:
  - Primary variable: composite financial institution development index (FI) (Sahey and others, 2015).
  - Subcomponents: (i) financial institution access (FIA), (ii) financial institution depth (FID), (iii) financial institution efficiency (FIE).
  - Robustness: financial development index (FD) including market component (Svirydzenka, 2016).
- Governance measurement: World Bank Worldwide Governance Indicators (VAE, POS, GEE, RQE, RLE, CCE).
- Benchmarking: gaps computed as deviations from global averages and standardized by annual global sample standard deviation → z-scores.
- Stylized findings:
  - CCA access surpassed global average by nearly ½ standard deviations; efficiency gap narrowed substantially; depth lags by close to 1 standard deviation.
  - Governance deficits: CCA lags global peers on voice and accountability, rule of law and control of corruption by between 0.6 and 1 standard deviations on average.
  - Country clusters by inclusion and governance gaps:
    - Smaller gaps: Armenia, Georgia, Kazakhstan — all have inflation targeting regimes and fiscal rules.
    - Larger gaps: Azerbaijan, the Kyrgyz Republic, Tajikistan, Turkmenistan — only two (Azerbaijan and Turkmenistan) have fiscal rules.
  - Composite confidence index (principal component of governance indicators + binary policy variables + fiscal rule index) shows decreasing but persistent confidence gaps; strong positive correlation between confidence and financial inclusion.
  - Confidence index normalized to vary between 0 and 1 by subtracting the minimum and dividing by new maximum.

### V. Empirical analysis — baseline model and main estimates
- Baseline panel model:
  - Annual data, 1996–2021.
  - Estimator: Hausman-Taylor instrumental variable estimator.
  - Dependent variable: FI_gap_{i,t} = gap between FI index and global average.
  - Key regressor: Confid_gap_{i,t} = gap between confidence index and global average.
  - Controls: detrended real GDP growth, inflation, banking crisis dummy, trade openness index (exports+imports/GDP), Chinn-Ito capital account liberalization index (country average), UN human development index, bank concentration ratio, overhead cost to total assets.
  - Endogeneity: financial inclusion and confidence gaps treated as endogenous; identification details addressed in Annex I.
- Composite confidence index results (full controls included):
  - CCA sample (Observations and Countries reported per outcome):
    - Confidence Gap → Financial Institution Gap: 0.739*** (Observations: 140; Countries: 8)
    - Confidence Gap → Financial Efficiency Gap: 1.384*** (Observations: 168; Countries: 8)
    - Confidence Gap → Financial Access Gap: 0.114** (Observations: 168; Countries: 8)
    - Confidence Gap → Financial Depth Gap: 0.329* (Observations: 168; Countries: 8)
  - Global sample:
    - Confidence Gap → Financial Institution Gap: 0.289*** (Observations: 2,605; Countries: 150)
    - Confidence Gap → Financial Efficiency Gap: 0.401*** (Observations: 2,605; Countries: 150)
    - Confidence Gap → Financial Access Gap: 0.139*** (Observations: 2,605; Countries: 150)
    - Confidence Gap → Financial Depth Gap: 0.193*** (Observations: 2,605; Countries: 150)
  - Significance notation: * p<0.05, ** p<0.01, *** p<0.001.
- Magnitude interpretation:
  - A 1-point decline in the confidence gap reduces the financial inclusion gap by 0.739 points in the CCA region and by 0.289 points in the global sample.

### VI. Institutional and policy drivers — interaction estimates and thresholds
- Selected coefficient estimates (Table 2; full controls included):
  - Governance gap (GG): CCA 0.023***; Global 0.021*** (Financial Institution Gap)
  - Deposit Insurance (DI): CCA 0.057***; Global 0.012*** (Financial Institution Gap)
  - Interaction GG × DI: CCA 0.021***; Global 0.008*** (Financial Institution Gap)
  - Inflation Targeting (IT): CCA 0.080***; Global 0.041*** (Financial Institution Gap)
  - Interaction GG × IT: CCA 0.049***; Global 0.023 (Financial Institution Gap)
  - Fiscal Rule (FR): CCA 0.036**; Global 0.017*** (Financial Institution Gap)
  - Interaction GG × FR: CCA 0.017**; Global 0.004*** (Financial Institution Gap)
- Table dimensions:
  - CCA Sample: Observations 176; Countries 8.
  - Global Sample: Observations 3,372; Countries 159.
  - Significance notation in Table 2: *** p<0.01, ** p<0.05, * p<0.1.
- Interpretation and counterfactuals:
  - An improvement in the governance gap of 1 standard deviation would reduce the average CCA financial institution gap by about 22 percent.
  - A commensurate and simultaneous improvement in CCA reliance on inflation targeting, fiscal rules and deposit insurance schemes would narrow the CCA financial institution gap by about 50 percent.
  - Threshold (governance gap) values above which IT, FR and DI effects on financial inclusion are positive:
    - Inflation targeting: -1.6
    - Fiscal rules: -2.1
    - Deposit insurance: -2.7
  - Effects would be negative if governance gap is below these thresholds.
  - CCA countries are more sensitive to confidence effects than global peers; policy rules and safety nets have larger impacts on financial institution development than on financial market development.

### VII. Governance subdimensions, heterogeneous effects, and extensions
- Governance subdimensions most influential: rule of law, regulatory quality, and control of corruption.
- Noted synergies:
  - Rule of law × Inflation Targeting.
  - Government effectiveness × Fiscal Rules.
  - Control of corruption × Deposit Insurance.
- Robustness and extensions:
  - Results robust to alternative specifications; extensive robustness analysis and synthetic-control case studies presented in Annexes I–III.

### VIII. Annex I — robustness, endogeneity, and investor risk perceptions (summary)
- Identification approach: use structural confidence indicators (governance, policy transparency, depositor protection) to explain fluctuations in sovereign spreads — avoids using sovereign spreads as explanatory variables to reduce endogeneity.
- Key finding: structural confidence indicators explain more than 50 percent of the variability in CCA sovereign spreads, compared to 8 percent for the global sample.
- Results robust to controls including GDP growth, inflation, and lagged sovereign spreads.
- Data constraints: sovereign spread data at monthly frequency for six out of eight CCA countries starting from 2012; governance series available annual up to 2021 (limits observations).
- Granger-causality (Dumitrescu & Hurlin panel test) for CCA:
  - Evidence of unidirectional causality from governance to financial inclusion.
  - Financial inclusion does not Granger-cause governance.
  - Test statistics reported: Z-bar tilde = 4.5819 (p-value = 0.0000); Z-bar = 6.1403 (p-value = 0.0000); W-bar = 4.0702; Lag order: 1.
- Hausman-Taylor / Fixed Effects regressions reported in Annex Table 2 (coefficients and significance for governance gap, DI, IT, FR across CCA and Global samples).
- Additional checks: Logit models testing whether adoption decisions for IT, FR, and DI are affected by the financial inclusion gap are insignificant.

### IX. Annex II — deposit insurance quality index and nonlinear governance interactions
- Deposit insurance quality index:
  - Constructed from annual AIDI surveys covering 2011-2021.
  - Covers 22 sub-components with equal weights (type of system, legal structure, member banks/institutions, eligible deposit products, coverage).
- Estimation approach: panel smooth transition regression with logistic transition function F(GOV) to capture gradual transition from weak to strong governance regimes.
  - Model form: FI_i,t = α + δ DI_index_i,t + β DI_index_i,t * F(GOV_i,t) + γ GOV_i,t + Σ θ_k X_i,t,k + n_i + μ_t + ε_i,t.
  - Marginal impact: ∂FI_i,t / ∂DI_index_i,t = δ + β * F(GOV).
- Quantitative result:
  - Marginal impact of a 1-point improvement in the deposit insurance index on financial inclusion increases from 0.19 points to 0.39 points as governance improves.
  - Annex Table 3 estimates (standardized indexes vary between 0 and 1):
    - Deposit insurance index coefficient 0.1992* (t = 1.8578)
    - Deposit insurance index × F(Governance; γ,c) coefficient 0.2034** (t = 2.5064)
    - Governance coefficient 0.0453*** (t = 3.0828)
    - Location parameter (Threshold) = 1.5593
  - Note: governance and deposit insurance indexes standardized to vary between 0 and 1; estimates not comparable to non-standardized analyses.

### X. Annex III — synthetic control case studies (IT and FR adoption)
- Methodology:
  - Synthetic control method for causal treatment effects of adopting inflation targeting (IT) and fiscal rules (FR) on financial inclusion.
  - Predictors include lagged financial inclusion, macro variables, composite governance indicator, deposit insurance dummy, fiscal rules index, banking crisis dummy, UN human development index.
  - Dataset spans 1995-2021.
- Inflation targeting results:
  - Treated countries: Armenia (2006), Georgia (2009), Kazakhstan (2015).
  - Comparator pool listed in source.
  - Finding: adoption of inflation targeting had a positive effect on financial inclusion in the three adopters; effect manifests with a lag of four to six years and shows strong persistence.
  - Interpretation: positive effects linked to relatively small governance gaps in the adopters and time needed for regimes to gain credibility via governance reforms.
- Fiscal rules results:
  - Treated countries: Armenia (2002), Azerbaijan (2019), Georgia (2013), Kazakhstan (2013), Turkmenistan (2015).
  - Comparator pool listed in source.
  - Finding: positive and statistically significant treatment effects of fiscal rules on financial inclusion in Georgia, Armenia, and Kazakhstan; insignificant effects in Azerbaijan and Turkmenistan.
  - Observations:
    - Short transmission lag (1 year) in Georgia, attributed to positive governance gap during the period.
    - Armenia required about 10 years of governance reforms after introducing a fiscal rule in 2002 before benefits for financial inclusion materialized.
    - Azerbaijan’s 2019 fiscal rule has limited post-adoption observations; effect significance improved over the evaluation period.
  - Conclusion: positive causal impacts of IT and FR are contingent on relatively small governance gaps; otherwise effects are limited.

### XI. Building supportive institutions — substantive policy implications (from Sections IV & Conclusion)
- Core institutional priorities to foster confidence and financial inclusion in CCA countries:
  - Strong and equal property rights
    - Reduce cost and number of procedures for obtaining legal titles; remove restrictions on property ownership including for women and low-income households.
    - Expand the range of acceptable collateral to broaden bank credit access.
  - Sound regulatory frameworks and robust financial safety nets
    - Reliable enforcement of contracts, strong banking regulations, effective risk-based supervision, and macroprudential oversight.
    - Well-designed bank resolution and depositor protection systems to protect creditor and depositor rights.
  - Adequate control of corruption
    - Strengthen corporate governance to reduce corruption that erodes property rights protection and public confidence.
  - Transparent policies and institutions
    - Central bank independence and well-designed inflation targeting regimes to ensure low inflation and adequate real returns.
    - Transparent, rules-based fiscal frameworks to limit deficit bias, fiscal dominance, and financial repression.
    - Strong accounting, disclosure practices, and effective credit bureaus to reduce informational asymmetries.
- Implementation support:
  - Cooperation with international financial institutions (IMF, World Bank, others) is essential to provide technical assistance and financial resources given capacity constraints in CCA countries.
- Summary conclusion:
  - Public confidence in macro-financial stability is crucial for financial inclusion in CCA countries.
  - The confidence channel is particularly sensitive to governance levels and reliance on transparent policy rules and robust financial safety nets.
  - Good governance enhances the credibility and effectiveness of inflation targeting, fiscal rules, and deposit insurance, thereby promoting financial inclusion.
  - Strengthening institutional frameworks—risk-based supervision, macroprudential oversight, regulatory and bank resolution frameworks, enforcement of creditor/depositor rights, control of corruption, central bank governance, and fiscal transparency—is paramount.

*Source: wpiea2024257-print-pdf (Annex I and Annexes II–III summaries as provided).*

### ANNEX I. ROBUSTNESS ANALYSIS ____________________________________________________________________ 13

### ANNEX I. ROBUSTNESS ANALYSIS

### I. Introduction — context and motivation
- Financial inclusion defined as the level of access and use of financial services by households and firms; essential to fostering more sustainable and inclusive growth.
- Financial inclusion incentivizes shifts away from short-term decision making toward more efficient intertemporal allocation of resources, promoting increased savings, reduced reliance on self-financing, enhanced incentives for productive investments, and deeper markets for goods and services (Rojas-Suarez and Amado, 2014).
- Despite progress over the last decade, closing financial inclusion gaps—especially financial deepening—remains a key challenge for countries in the Caucasus and Central Asia (CCA).
- The 2021 Findex Survey highlighted trust in banks as a major barrier to bank account penetration in the region.
- Multiple external shocks over the last decade have resulted in economic and financial distress, testing confidence in the CCA’s macroeconomic policy and institutional frameworks.

### Confidence as a driver of financial inclusion
- Public confidence in banking institutions and the broader economy is an important driver of both demand for and supply of financial services.
- Banking is inherently based on trust in the presence of asymmetric information.
- Higher levels of confidence in banks bolster demand for formal bank savings (Allen and others, 2016).
- Confidence is fragile after macro-financial crises; once lost, it is difficult to recover.
- Past confidence crises have been associated with contagious bank runs and durable retrenchment in financial intermediation.

### Literature gap and contribution
- A growing literature examines development policy drivers of financial inclusion, but the confidence–financial inclusion nexus remains relatively unexplored.
- Related findings cited:
  - Non-linear relationship between financial inclusion and financial stability at the global level (Sahay and others, 2015; Čihák and others, 2016).
  - Benefits of digital financial inclusion for economic growth (Khera and others, 2021; Aguilar and others, 2024).
  - Enhanced financial inclusion of small and medium-sized enterprises has a robust positive impact on growth in CCA countries (Blancher and others, 2019).
  - Human development and the rule of law are essential for financial inclusion in the CCA region (Poghosyan, 2022; 2023).
- This paper examines the nexus between public confidence and financial inclusion in the CCA region, considering both confidence in banks and confidence in the broader economy.

### Key findings (as reported)
- Financial inclusion in CCA countries is particularly sensitive to confidence effects.
- Institutional drivers of confidence—such as governance and policy transparency—play a much larger role in the CCA relative to global peers.
- Confidence depends on broad-based improvements to institutional quality and greater reliance on transparent policy frameworks and robust financial safety nets.
- Effects of policy tools on financial inclusion depend non-linearly on governance:
  - Inflation targeting, fiscal rules, and deposit insurance have effects on financial inclusion that depend on the level of governance.
  - The positive effects of these tools on financial inclusion are reinforced by better governance and eroded by governance slippages.
- Consequently, advancements in governance are key to unlocking benefits of policy rules and financial safety nets for financial inclusion and should be prioritized in future reforms.

### Structure of the paper (as stated)
- The remainder of the paper is organized with Section II providing a brief overview of related literature.

*IMF Working Papers — A Confidence-Financial Inclusion Nexus in Caucasus and Central Asia?*

### Section III presents stylized facts. Section IV discusses the empirical analysis. Section V focuses on policy

### Section III–V: Stylized Facts, Empirical Analysis, and Policy Implications

### II. Confidence–Financial Inclusion Nexus
- Focus: confidence in macro-financial stability (macro-level) rather than trust in individual financial institutions.
- Conceptual model:
  - Confidence modeled as a function of sentiment and policy uncertainty.
  - Sentiment drivers: income level, GDP growth, inflation, banking crises, trade openness, capital account liberalization.
  - Policy-driven, forward-looking confidence affected by uncertainty over policy makers’ actions and effects on the economy.
- Channels:
  - Economic policy uncertainty weakens public confidence in banks → reduces demand for formal bank savings.
  - Policy uncertainty incentivizes banks to reduce risk exposure → supply fewer financial services.
  - Reduced policy uncertainty fosters financial inclusion via improvements in banking stability, financial system efficiency, cross-border capital flows, and lower risk premia.
- Institutional and policy enablers:
  - Good governance mitigates economic policy uncertainty.
  - Inflation targeting: commitment device through transparency and accountability; supports financial inclusion by promoting transparency and ensuring interest rates offer adequate real returns.
  - Fiscal rules: improve sovereign risk perceptions and fiscal transparency; reduce deficit bias, public debt burden, and crowding out of private credit.
  - Strong banking supervision, regulation, and depositor protection support confidence in banks.
  - Interaction effects: deposit insurance can have negative effects in lax regulatory environments but positive effects in sound regulatory environments.

### III. Stylized Facts
- Financial inclusion measurement:
  - Primary variable: composite financial institution development index (FI) (Sahey and others, 2015).
  - Subcomponents: (i) financial institution access (FIA), (ii) financial institution depth (FID), (iii) financial institution efficiency (FIE).
  - Robustness: financial development index (FD) including market component (Svirydzenka, 2016).
- Governance measurement:
  - World Bank Worldwide Governance Indicators (WGI): voice and accountability (VAE), political stability (POS), government effectiveness (GEE), regulatory quality (RQE), rule of law (RLE), control of corruption (CCE).
- Benchmarking approach:
  - Gaps computed as deviations from global averages and standardized by annual global sample standard deviation → z-scores.
- Key stylized findings:
  - CCA progress but persistent gaps: access surpassed global average by nearly ½ standard deviations; efficiency gap narrowed substantially; depth lags by close to 1 standard deviation.
  - Governance deficits: CCA lags global peers on voice and accountability, rule of law and control of corruption by between 0.6 and 1 standard deviations on average.
  - Country clusters by inclusion and governance gaps:
    - Smaller gaps: Armenia, Georgia, Kazakhstan — all have inflation targeting regimes and fiscal rules.
    - Larger gaps: Azerbaijan, the Kyrgyz Republic, Tajikistan, Turkmenistan — only two (Azerbaijan and Turkmenistan) have fiscal rules.
  - Composite confidence index (principal component of governance indicators + binary policy variables + fiscal rule index) shows decreasing but persistent confidence gaps relative to global peers and CEE; strong positive correlation between confidence and financial inclusion.
  - Confidence index normalized to vary between 0 and 1 by subtracting the minimum and dividing by new maximum.

### IV. Empirical Analysis
- Baseline panel model (annual data, 1996–2021) estimated with Hausman-Taylor instrumental variable estimator:
  - Dependent variable: FI_gap_{i,t} = gap between FI index and global average.
  - Key regressor: Confid_gap_{i,t} = gap between confidence index and global average.
  - Controls (X_{i,t,k}): detrended real GDP growth, inflation, banking crisis dummy, trade openness index (exports+imports/GDP), Chinn-Ito capital account liberalization index (country average), UN human development index, bank concentration ratio, overhead cost to total assets.
  - Endogeneity addressed by treating financial inclusion and confidence gaps as endogenous; additional detail in Annex I.
- Composite confidence index results (Table 1, full controls included):
  - CCA sample:
    - Confidence Gap coefficient on Financial Inclusion Gap: 0.739*** (Observations: 140; Countries: 8)
    - Financial Efficiency Gap: 1.384*** (Observations: 168; Countries: 8)
    - Financial Access Gap: 0.114** (Observations: 168; Countries: 8)
    - Financial Depth Gap: 0.329* (Observations: 168; Countries: 8)
  - Global sample:
    - Confidence Gap coefficient on Financial Inclusion Gap: 0.289*** (Observations: 2,605; Countries: 150)
    - Financial Efficiency Gap: 0.401*** (Observations: 2,605; Countries: 150)
    - Financial Access Gap: 0.139*** (Observations: 2,605; Countries: 150)
    - Financial Depth Gap: 0.193*** (Observations: 2,605; Countries: 150)
  - Significance notation: * p<0.05, ** p<0.01, *** p<0.001.
- Magnitude interpretation:
  - A 1-point decline in the confidence gap reduces the financial inclusion gap by 0.739 points in the CCA region and by 0.289 points in the global sample.
- Institutional and policy drivers (Table 2: Governance gap and interactions with policy dummies)
  - Selected coefficient estimates (full controls included):
    - Governance gap (GG): CCA 0.023***; Global 0.021*** (Financial Institution Gap)
    - Deposit Insurance (DI): CCA 0.057***; Global 0.012*** (Financial Institution Gap)
    - Interaction GG x DI: CCA 0.021***; Global 0.008*** (Financial Institution Gap)
    - Inflation Targeting (IT): CCA 0.080***; Global 0.041*** (Financial Institution Gap)
    - Interaction GG x IT: CCA 0.049***; Global 0.023 (Financial Institution Gap)
    - Fiscal Rule (FR): CCA 0.036**; Global 0.017*** (Financial Institution Gap)
    - Interaction GG x FR: CCA 0.017**; Global 0.004*** (Financial Institution Gap)
  - Table dimensions:
    - CCA Sample: Observations 176; Countries 8.
    - Global Sample: Observations 3,372; Countries 159.
  - Significance notation in Table 2: *** p<0.01, ** p<0.05, * p<0.1.
- Interpretation and counterfactuals:
  - An improvement in the governance gap of 1 standard deviation would reduce the average CCA financial institution gap by about 22 percent.
  - A commensurate and simultaneous improvement in CCA reliance on inflation targeting, fiscal rules and deposit insurance schemes would narrow the CCA financial institution gap by about 50 percent.
  - Threshold effects: the overall effects of IT, FR and DI on financial inclusion are positive when the governance gap exceeds:
    - Inflation targeting: -1.6
    - Fiscal rules: -2.1
    - Deposit insurance: -2.7
    - Otherwise the effects would be negative.
  - CCA countries are more sensitive to confidence effects than global peers; policy rules and safety nets have larger impacts on financial institution development than on financial market development.
- Governance subdimensions:
  - Financial inclusion particularly sensitive to improvements in rule of law, regulatory quality, and control of corruption.
  - Strong synergies found for:
    - Rule of law × Inflation Targeting.
    - Government effectiveness × Fiscal Rules.
    - Control of corruption × Deposit Insurance.
- Robustness and extensions:
  - Results robust to alternative specifications; extensive robustness analysis and synthetic-control case studies presented in Annexes I–III.

*Italic: Source: wpiea2024257-print-pdf (Sections II–IV as provided).*

### Annex I addresses endogeneity and shows that the structural confidence indicators discussed above can

### wpiea2024257-print-pdf - Annex I addresses endogeneity and shows that the structural confidence indicators discussed above can

### Robustness, endogeneity, and investor risk perceptions
- Identification approach: use structural confidence indicators (governance, policy transparency, depositor protection) to explain fluctuations in sovereign spreads rather than using sovereign spreads as explanatory variables to avoid endogeneity.
- Key findings:
  - Structural confidence indicators explain more than 50 percent of the variability in CCA sovereign spreads, compared to 8 percent for the global sample.
  - Results are statistically significant and robust to controls including GDP growth, inflation, and lagged sovereign spreads.
  - Sovereign spread data: monthly frequency for six out of eight CCA countries starting from 2012; governance series available annual up to 2021 (limits observations).
- Granger-causality evidence:
  - Dumitrescu & Hurlin panel test results indicate unidirectional causality from governance to financial inclusion in the CCA sample.
  - Financial inclusion does not Granger-cause governance in the CCA sample.
  - Dumitrescu & Hurlin test statistics reported: Z-bar tilde = 4.5819 (p-value = 0.0000); Z-bar = 6.1403 (p-value = 0.0000); W-bar = 4.0702; Lag order: 1.
  - Hausman-Taylor / Fixed Effects regressions reported (Annex Table 2) with coefficients and significance levels for governance gap, Deposit Insurance, Inflation Targeting, Fiscal Rule across CCA and Global samples (see source for full table).
- Additional robustness checks:
  - Logit models used to test whether the decision to introduce inflation targeting, fiscal rules, and deposit insurance is affected by the financial inclusion gap; results are insignificant.

### Governance, depositor protection, and heterogeneous effects (Annex II)
- Methodology:
  - Constructed a novel deposit insurance quality index based on annual AIDI surveys covering 2011-2021.
  - Deposit insurance index covers 22 sub-components (type of system, legal structure, member banks/institutions, eligible deposit products, coverage) with equal weights.
  - Employed a panel smooth transition regression with a logistic transition function F(GOV) to capture gradual transition from weak to strong governance regimes:
    - Model: FI_i,t = α + δ DI_index_i,t + β DI_index_i,t * F(GOV_i,t) + γ GOV_i,t + Σ θ_k X_i,t,k + n_i + μ_t + ε_i,t.
    - Marginal impact: ∂FI_i,t / ∂DI_index_i,t = δ + β * F(GOV).
- Quantitative results:
  - Marginal impact of a 1-point improvement in the deposit insurance index on financial inclusion increases from 0.19 points to 0.39 points as governance improves.
  - Annex Table 3 reporting estimates: Deposit insurance index coefficient 0.1992* (t = 1.8578); Deposit insurance index × F(Governance; γ,c) coefficient 0.2034** (t = 2.5064); Governance coefficient 0.0453*** (t = 3.0828); Location parameter (Threshold) = 1.5593. Controls included. Note: governance and deposit insurance indexes standardized to vary between 0 and 1; estimates not comparable to non-standardized analyses.

### Case studies: synthetic control evaluation of inflation targeting and fiscal rules (Annex III)
- Methodology:
  - Synthetic control method to estimate causal treatment effects of adopting inflation targeting (IT) and fiscal rules (FR) on financial inclusion.
  - Predictors include lagged financial inclusion, macro variables (GDP growth/volatility, CPI inflation, GDP per capita, trade openness, Chinn-Ito index, broad money-to-GDP), composite governance indicator, deposit insurance dummy, fiscal rules index, banking crisis dummy, UN human development index.
  - Dataset spans 1995-2021.
- Inflation targeting results:
  - Treated countries: Armenia (2006), Georgia (2009), Kazakhstan (2015).
  - Comparator pool: Azerbaijan, Kyrgyzstan, Tajikistan, Turkmenistan, Uzbekistan, Albania, Bulgaria, Belarus, Croatia, Northern Macedonia, Ukraine.
  - Finding: adoption of inflation targeting had a positive effect on financial inclusion in the three adopters; effect manifests with a lag of four to six years and shows strong persistence.
  - Interpretation: positive effects likely due to relatively small governance gaps in Armenia, Georgia, Kazakhstan and the time needed for regimes to gain credibility via governance reforms.
- Fiscal rules results:
  - Treated countries: Armenia (2002), Azerbaijan (2019), Georgia (2013), Kazakhstan (2013), Turkmenistan (2015).
  - Comparator pool: Kyrgyzstan, Tajikistan, Uzbekistan, Albania, Belarus, Moldova, Northern Macedonia, Turkey, Ukraine.
  - Finding: positive and statistically significant treatment effects of fiscal rules on financial inclusion in Georgia, Armenia, and Kazakhstan; insignificant effects in Azerbaijan and Turkmenistan.
  - Observations:
    - Short transmission lag (1 year) in Georgia, attributed to positive governance gap during the period.
    - Armenia required about 10 years of governance reforms after introducing a fiscal rule in 2002 before benefits for financial inclusion materialized.
    - Azerbaijan’s 2019 fiscal rule has limited post-adoption observations; effect significance improved over the evaluation period.
  - Conclusion: positive causal impacts of IT and FR are contingent on relatively small governance gaps; otherwise effects limited.

### Building supportive institutions — substantive policy implications (Section IV & Conclusion)
- Core institutional priorities to foster confidence and financial inclusion in CCA countries:
  - Strong and equal property rights
    - Reduce cost and number of procedures for obtaining legal titles; remove restrictions on property ownership including for women and low-income households.
    - Expand the range of acceptable collateral to broaden bank credit access.
  - Sound regulatory frameworks and robust financial safety nets
    - Reliable enforcement of contracts, strong banking regulations, effective risk-based supervision, and macroprudential oversight.
    - Well-designed bank resolution and depositor protection systems to protect creditor and depositor rights.
  - Adequate control of corruption
    - Strengthen corporate governance to reduce corruption that erodes property rights protection and public confidence.
  - Transparent policies and institutions
    - Central bank independence and well-designed inflation targeting regimes to ensure low inflation and adequate real returns.
    - Transparent, rules-based fiscal frameworks to limit deficit bias, fiscal dominance, and financial repression.
    - Strong accounting, disclosure practices, and effective credit bureaus to reduce informational asymmetries.
- Implementation support:
  - Cooperation with international financial institutions (IMF, World Bank, others) is essential to provide technical assistance and financial resources given capacity constraints in CCA countries.
- Summary conclusion:
  - Public confidence in macro-financial stability is crucial for financial inclusion in CCA countries.
  - The confidence channel is particularly sensitive to governance levels and reliance on transparent policy rules and robust financial safety nets.
  - Good governance enhances the credibility and effectiveness of inflation targeting, fiscal rules, and deposit insurance, thereby promoting financial inclusion.
  - Strengthening institutional frameworks—risk-based supervision, macroprudential oversight, regulatory and bank resolution frameworks, enforcement of creditor/depositor rights, control of corruption, central bank governance, and fiscal transparency—is paramount.

*Source: IMF Working Paper (Annexes I–III content as provided in the supplied PDF content).*

### Annex Figure 1. Treatment Effects of Inflation Targeting on Financial Inclusion

### Annex Figure 1. Treatment Effects of Inflation Targeting on Financial Inclusion

### Figure description
- Title shown: "Annex Figure 1. Treatment Effects of Inflation Targeting on Financial Inclusion"
- Caption/source line shown: "Source: Authors’ estimates."
- Appears within the IMF Working Paper "A Confidence-Financial Inclusion Nexus in Caucasus and Central Asia?"
- Located in the sequence of annex figures in the PDF (followed by "Annex Figure 2. Treatment Effects of Fiscal Rules on Financial Inclusion").

### Context within the working paper
- Document series: IMF WORKING PAPERS
- Working Paper title: "A Confidence-Financial Inclusion Nexus in Caucasus and Central Asia?"
- Working Paper No.: WP/2024/257
- Pages around the figure in the provided excerpt: 19–21 (annex figures and surrounding captions/references appear across these pages).

### Key labels and provenance
- Figure label: "Annex Figure 1"
- The figure attribution is explicitly "Authors’ estimates."

*Source: wpiea2024257-print-pdf (Annex Figure 1) — IMF Working Paper No. WP/2024/257*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024257-print-pdf.pdf_