## Introduction

## Source details

**Canonical URL:** [Introduction](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022060-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022060-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022060-print-pdf.pdf.json)

---

### Key premises and motivations
- Crypto-assets are private digital assets that depend primarily on cryptography and distributed ledger technology for record keeping.
- Crypto-assets present opportunities (efficiency, inclusiveness) and risks (consumer and investor protection, market integrity, tax evasion, money laundering, terrorist financing).
- Pseudonymity of crypto-assets makes them a potential vehicle for illicit flows, including proceeds from corruption; crypto-assets can move large amounts speedily and across borders.
- Potential policy responses include regulation such as know-your-customer (KYC) approaches; there are also potential public-sector uses of distributed ledger technologies (e.g., procurement transparency, property registries) and centrally issued digital currencies.

### Data Description and Methodology
- Baseline crypto-usage data: Statista Global Consumer Survey (2020), 2,000–12,000 respondents per country, 55 countries covered; variable = share of respondents who indicated they either owned or used cryptocurrencies in 2020.
- Transformation used in analysis: logarithm of one plus the share of users in total population (log(1+crypto adoption rate)).
- Alternative datasets considered (not used as baseline): Chainalysis Global Crypto Adoption Index (July 2020–June 2021), Finder, Triple A, Coin Dance; Appendix I reports results using Chainalysis and notes weak or no correlation across datasets.
- Explanatory variables (all for 2020 unless stated): Control of corruption index (Worldwide Governance Indicators), Average consumer price inflation rate for 2011-20 (log(1+inflation rate)), Capital openness: Chinn-Ito Index (latest value 2019), Real GDP per capita (log, PPP 2017, international dollars), Average remittances to GDP (2017-2019), Commercial bank branches per 100,000 adults (log), Secure internet servers per 1 million people (log).
- Regression sample: 53 countries.
- Logarithmic transformations applied to skewed variables (crypto adoption, inflation, commercial bank branches, secure internet servers).
- Multicollinearity noted among explanatory variables (average VIF = 3.74; GDP and control of corruption VIFs above 4).

### Results

- Descriptive statistics (Table 1, 53 countries):
  - Crypto adoption (log of (1+crypto adoption rate)): Mean 0.093, Median 0.084, Std. dev. 0.045, Min 0.036, Max 0.277.
  - Control of corruption (index): Mean 0.579, Median 0.566, Std. dev. 1.031, Min -1.097, Max 2.270.
  - Inflation (log of 1+inflation rate (2011-2020)): Mean 1.217, Median 1.050, Std. dev. 0.650, Min -0.124, Max 3.317.
  - Real GDP per capita (log, PPP 2017, international dollars): Mean 10.196, Median 10.386, Std. dev. 0.775, Min 8.475, Max 11.445.
  - Capital openness (index): Mean 1.239, Median 2.322, Std. dev. 1.386, Min -1.226, Max 2.322.
  - Commercial bank branches (log, per 100,000 adults): Mean 2.462, Median 2.603, Std. dev. 0.906, Min -0.570, Max 3.818.
  - Average remittances to GDP (2017-2019): Mean 0.017, Median 0.004, Std. dev. 0.025, Min 0.000, Max 0.098.
  - Secure internet servers (log, per 1 million people): Mean 8.870, Median 9.576, Std. dev. 2.348, Min 3.784, Max 12.532.

- Pairwise correlations (Table 2, significance):
  - Crypto adoption vs control of corruption: -0.53*** (p-value 0.00).
  - Crypto adoption vs inflation: 0.47*** (p-value 0.00).
  - Crypto adoption vs real GDP per capita: -0.54*** (p-value 0.00).
  - Crypto adoption vs capital openness: -0.54*** (p-value 0.00).
  - Crypto adoption vs commercial bank branches: -0.18 (p-value 0.19).
  - Crypto adoption vs average remittances to GDP: 0.29** (p-value 0.03).
  - Crypto adoption vs secure internet servers: -0.49*** (p-value 0.00).
  - Note: significance markers: *** p<0.01, ** p<0.05, * p<0.1.

- Multivariate regressions (Table 3, general-to-specific approach):
  - Estimation approach: OLS with robust standard errors; general-to-specific variable elimination to address multicollinearity and redundancy.
  - Final specification (regression 6 in Table 3) includes:
    - Control of corruption (index): coefficient -0.014* (t-statistics (-1.984)); text reports p-value 0.053.
    - Capital openness (index): coefficient -0.011* (t-statistics (-2.008)); text reports p-value 0.05.
    - Commercial bank branches, real GDP per capita, secure internet servers, remittances, and inflation were sequentially dropped or found redundant.
  - Regression diagnostics and fit:
    - Observations: 53.
    - R-squared values across specifications: 0.377, 0.377, 0.376, 0.376, 0.374, 0.340 (Table 3 columns (1)–(6)).

- Main empirical findings:
  - Crypto-asset usage is significantly and positively associated with corruption (weaker control of corruption) and with capital controls (lower capital openness).
  - In the last-stage regression, p-values for coefficients: control of corruption = 0.053; capital account openness = 0.05.
  - Interpretation: Countries with weaker control of corruption (more corruption) and lower degree of capital openness (more capital controls) tend to have a larger share of crypto adoption.
  - Magnitude example: A move from the 25th percentile to the 75th percentile in control of corruption and capital openness (other things being equal) is associated with a decline in crypto adoption by around 2 and 4 percentage points, respectively.
  - Commercial bank branches per capita has the expected (negative) sign but is not statistically significant.

- Robustness:
  - Findings robust to alternative measures of explanatory variables.
  - Results not driven by influential observations or outliers.
  - Results remain significant when using an alternative crypto adoption definition (Chainalysis), but the alternative measure has important deficiencies (e.g., geographic attribution issues).

### Conclusion and policy implication
- Main empirical findings:
  - Cross-country regression analysis finds that more crypto usage is empirically associated with higher perceived corruption and more intensive capital controls.
  - Using alternative variables (Corruption Perceptions Index from Transparency International and Capital Control Measures index from Fernandez et al.) has not changed the findings.
  - The evidence supports prudence given the rapid increase in macroeconomic relevance of crypto assets.
- Policy implications and recommendations:
  - Evidence adds to the case for regulating crypto usage—for example, by requiring intermediaries to implement know-your-customer procedures.
  - Work should continue in using the technologies underlying crypto assets to realize potential benefits to financial inclusion and the efficiency of governments.
  - The analysis shows the need for better data to understand the dynamics and the key driving factors behind crypto adoption.
- Robustness and alternative data (Chainalysis 2021 Global Crypto Adoption Index):
  - Chainalysis’ 2021 Global Crypto Adoption Index covers 154 countries but relies on web traffic data; usage of VPNs and other masking products can lead to errors in estimating crypto use and assigning it to specific countries.
  - Statista’s measure based on survey responses is skewed but has a more normally shaped distribution; the two datasets have a correlation of 0.43 and only four countries appear in the top ten of both lists.
  - Chainalysis sample increases regression sample size from 53 to 126 and yields similar coefficient signs and significance for control of corruption and capital controls; Chainalysis regressions indicate a stronger negative association between crypto usage and financial development (bank branches).
  - Descriptive statistics and regression results for the Chainalysis sample are reported for 126/127 countries with specific means, medians, standard deviations, coefficient estimates, t-statistics, observations, and R-squared values preserved in the source.

*Source: wpiea2022060-print-pdf - Introduction.*

### Introduction ...........................................................................................................

### Introduction

### Key premises and motivations
- Crypto-assets are private digital assets that depend primarily on cryptography and distributed ledger technology for record keeping.
- Crypto-assets present opportunities (efficiency, inclusiveness) and risks (consumer and investor protection, market integrity, tax evasion, money laundering, terrorist financing).
- Pseudonymity of crypto-assets makes them a potential vehicle for illicit flows, including proceeds from corruption; crypto-assets can move large amounts speedily and across borders.
- Potential policy responses include regulation such as know-your-customer (KYC) approaches; there are also potential public-sector uses of distributed ledger technologies (e.g., procurement transparency, property registries) and centrally issued digital currencies.

### Data Description and Methodology

### Data sources and primary dependent variable
- Baseline crypto-usage data: Statista Global Consumer Survey (2020), 2,000–12,000 respondents per country, 55 countries covered; variable = share of respondents who indicated they either owned or used cryptocurrencies in 2020.
- Transformation used in analysis: logarithm of one plus the share of users in total population (log(1+crypto adoption rate)).

### Alternative datasets considered (not used as baseline)
- Chainalysis Global Crypto Adoption Index (July 2020–June 2021): based on web traffic/on-chain metrics; covers 154 countries but likely distorted by VPNs and has many near-zero observations.
- Finder survey data: larger samples but covers only 27 countries.
- Triple A and Coin Dance: have serious shortfalls (assumptions based on Canada, bitcoin-only volume).
- Appendix I reports results using Chainalysis and notes weak or no correlation across datasets.

### Explanatory variables (all for 2020 unless stated)
- Control of corruption index (Worldwide Governance Indicators), range -2.5 to 2.5.
- Average consumer price inflation rate for 2011-20, transformed as log(1+inflation rate).
- Capital openness: Chinn-Ito Index (latest value 2019), range in sample -1.9 to 2.3 (higher = greater openness).
- Real GDP per capita (log, PPP 2017, international dollars).
- Average remittances to GDP (2017-2019), percent of GDP.
- Commercial bank branches per 100,000 adults (log) from IMF Financial Access Survey.
- Secure internet servers per 1 million people (log) from World Bank population estimates (control for ability to engage in crypto transactions).

### Sample and transformations
- Regression sample: 53 countries (list provided in source).
- Logarithmic transformations applied to skewed variables (crypto adoption, inflation, commercial bank branches, secure internet servers).
- Multicollinearity noted among explanatory variables (average VIF = 3.74; GDP and control of corruption VIFs above 4).

### Results

### Descriptive statistics (Table 1, 53 countries)
- Crypto adoption (log of (1+crypto adoption rate)): Mean 0.093, Median 0.084, Std. dev. 0.045, Min 0.036, Max 0.277.
- Control of corruption (index): Mean 0.579, Median 0.566, Std. dev. 1.031, Min -1.097, Max 2.270.
- Inflation (log of 1+inflation rate (2011-2020)): Mean 1.217, Median 1.050, Std. dev. 0.650, Min -0.124, Max 3.317.
- Real GDP per capita (log, PPP 2017, international dollars): Mean 10.196, Median 10.386, Std. dev. 0.775, Min 8.475, Max 11.445.
- Capital openness (index): Mean 1.239, Median 2.322, Std. dev. 1.386, Min -1.226, Max 2.322.
- Commercial bank branches (log, per 100,000 adults): Mean 2.462, Median 2.603, Std. dev. 0.906, Min -0.570, Max 3.818.
- Average remittances to GDP (2017-2019): Mean 0.017, Median 0.004, Std. dev. 0.025, Min 0.000, Max 0.098.
- Secure internet servers (log, per 1 million people): Mean 8.870, Median 9.576, Std. dev. 2.348, Min 3.784, Max 12.532.

### Pairwise correlations (Table 2, significance)
- Crypto adoption vs control of corruption: -0.53*** (p-value 0.00).
- Crypto adoption vs inflation: 0.47*** (p-value 0.00).
- Crypto adoption vs real GDP per capita: -0.54*** (p-value 0.00).
- Crypto adoption vs capital openness: -0.54*** (p-value 0.00).
- Crypto adoption vs commercial bank branches: -0.18 (p-value 0.19).
- Crypto adoption vs average remittances to GDP: 0.29** (p-value 0.03).
- Crypto adoption vs secure internet servers: -0.49*** (p-value 0.00).
- Note: p-values in Table 2 are reported in parentheses; significance markers: *** p<0.01, ** p<0.05, * p<0.1.

### Multivariate regressions (Table 3, general-to-specific approach)
- Estimation approach: OLS with robust standard errors; general-to-specific variable elimination to address multicollinearity and redundancy.
- Final specification (regression 6 in Table 3) includes:
  - Control of corruption (index): coefficient -0.014* (t-statistics reported as (-1.984)); significance * (p<0.1 in table; text reports p-value 0.053).
  - Capital openness (index): coefficient -0.011* (t-statistics (-2.008)); significance * (p<0.1 in table; text reports p-value 0.05).
  - Commercial bank branches, real GDP per capita, secure internet servers, remittances, and inflation were sequentially dropped or found redundant.
- Regression diagnostics and fit:
  - Observations: 53.
  - R-squared values across specifications: 0.377, 0.377, 0.376, 0.376, 0.374, 0.340 (Table 3 columns (1)–(6)).
  - Robust t-statistics presented in Table 3; significance markers: *** p<0.01, ** p<0.05, * p<0.1.

### Main empirical findings
- Crypto-asset usage is significantly and positively associated with corruption (weaker control of corruption) and with capital controls (lower capital openness).
- In the last-stage regression, p-values for coefficients: control of corruption = 0.053; capital account openness = 0.05.
- Interpretation: Countries with weaker control of corruption (more corruption) and lower degree of capital openness (more capital controls) tend to have a larger share of crypto adoption.
- Magnitude example: A move from the 25th percentile to the 75th percentile in control of corruption and capital openness (other things being equal) is associated with a decline in crypto adoption by around 2 and 4 percentage points, respectively.
- Commercial bank branches per capita has the expected (negative) sign but is not statistically significant.

### Robustness
- Findings robust to alternative measures of explanatory variables.
- Results not driven by influential observations or outliers (e.g., high crypto usage outliers, average inflation outliers).
- Results remain significant when using an alternative crypto adoption definition (Chainalysis), but the alternative measure has important deficiencies (e.g., geographic attribution issues) making inference less reliable.

### Conclusion and policy implication
- Given the associations found, the analysis adds to the case for regulating crypto-assets (including KYC approaches) rather than adopting a laissez-faire stance, especially in light of risks related to corruption proceeds and circumvention of capital controls.
- Results are presented with caveats: small sample size (53 countries), data-quality concerns, and multicollinearity challenges. Measurement error likely biases against finding significance; therefore significant associations merit attention despite data limitations.

*Source: wpiea2022060-print-pdf - Introduction.*

### Conclusion

### Conclusion

### Main empirical findings
- Cross-country regression analysis using a general-to-specific approach finds that more crypto usage is empirically associated with higher perceived corruption and more intensive capital controls.
- Using alternative variables, the Corruption Perceptions Index from Transparency International and Capital Control Measures index from Fernandez et al., has not changed our findings.
- The evidence supports prudence given the rapid increase in macroeconomic relevance of crypto assets.

### Policy implications and recommendations
- This evidence adds to the case for regulating crypto usage—for example, by requiring intermediaries to implement know-your-customer procedures.
- Work should continue in using the technologies underlying crypto assets to realize potential benefits to financial inclusion and the efficiency of governments.
- The analysis also shows the need for better data to understand the dynamics and the key driving factors behind crypto adoption.

### Robustness and alternative data (Chainalysis 2021 Global Crypto Adoption Index)
- Chainalysis’ 2021 Global Crypto Adoption Index covers 154 countries but relies on web traffic data; usage of VPNs and other masking products can lead to errors in estimating crypto use and assigning it to specific countries.
- Values of the Chainalysis index have a highly skewed distribution with a long right tail; the largest outlier is excluded in regressions using the index.
- Statista’s measure based on survey responses is also skewed but has a more normally shaped distribution; the two datasets have a correlation of 0.43 and only four countries appear in the top ten of both lists.
- The authors consider the Statista measure to have a more plausible distribution and to avoid likely mis-allocation due to VPN use.

### Key results using Chainalysis index (expanded sample)
- Using the Chainalysis index increases the regression sample size from 53 to 126.
- The coefficients of explanatory variables largely maintain their signs compared with previous results.
- For control of corruption and capital controls the coefficients are very similar in magnitude to those reported elsewhere and remain significant and robust to redundancy tests.
- The Chainalysis regressions indicate a stronger and more significant negative association between crypto usage and the level of financial development (as proxied by the number of bank branches).

### Descriptive statistics (Chainalysis sample of 127 countries)
- Crypto adoption (log of (1+crypto adoption rate)): Mean 0.062; Median 0.039; Std. dev. 0.083; Min 0.000; Max 0.693
- Control of corruption (index): Mean 0.126; Median -0.066; Std. dev. 0.985; Min -1.572; Max 2.270
- Inflation (log of 1+inflation rate (2011-2020)): Mean 1.400; Median 1.292; Std. dev. 0.708; Min -0.124; Max 4.429
- Real GDP per capita (log, PPP 2017, international dollars): Mean 9.623; Median 9.694; Std. dev. 1.031; Min 6.987; Max 11.445
- Capital openness (index): Mean 0.729; Median 1.049; Std. dev. 1.543; Min -1.924; Max 2.322
- Commercial bank branches (log, per 100,000 adults): Mean 2.367; Median 2.581; Std. dev. 1.102; Min -3.413; Max 4.231
- Average of remittances to GDP (2017-2019): Mean 0.043; Median 0.019; Std. dev. 0.060; Min 0.000; Max 0.306
- Secure internet servers (log, per 1 million people): Mean 7.326; Median 7.140; Std. dev. 2.826; Min 1.349; Max 12.532
- Note: Descriptive statistics are for the 127 countries in the regression sample.

### Pairwise correlations (selected)
- Crypto adoption with Control of corruption: -0.17** (p-value 0.04)
- Crypto adoption with Inflation (log of 1+inflation rate (2011-2020)): 0.27*** (p-value 0.00)
- Crypto adoption with Real GDP per capita (log, PPP 2017): -0.17** (p-value 0.04)
- Crypto adoption with Capital openness (index): -0.22** (p-value 0.01)
- Crypto adoption with Commercial bank branches (log, per 100,000 adults): -0.18** (p-value 0.04)
- Note: p-values in parentheses. *** p<0.01, ** p<0.05, * p<0.1

### Empirical regression results (Chainalysis crypto adoption as dependent variable)
- Control of corruption (index) coefficients:
  - (1) -0.019** (-2.118)
  - (2) -0.023** (-2.577)
  - (3) -0.026*** (-2.775)
  - (4) -0.026*** (-2.659)
- Capital openness (index) coefficients:
  - (1) -0.010** (-2.274)
  - (2) -0.011** (-2.442)
  - (3) -0.013*** (-2.807)
  - (4) -0.012*** (-2.719)
- Commercial bank branches (log, per 100,000 adults):
  - (1) -0.011 (-1.366)
  - (2) -0.013 (-1.604)
  - (3) -0.014* (-1.778)
  - (4) -0.015** (-2.070)
- Secure internet servers (log, per 1 million people):
  - (1) 0.014** (2.339)
  - (2) 0.014** (2.371)
  - (3) 0.011** (2.350)
  - (4) 0.011** (2.596)
- Average of remittances to GDP (2017-2019): coefficients not significant in reported specifications (e.g., -0.140, -0.135, -0.092 with t-statistics shown).
- Constant terms and other included variables reported in table; robust t-statistics in parentheses. *** p<0.01, ** p<0.05, * p<0.1
- Observations: 126, 126, 126, 127 (for columns (1)–(4))
- R-squared: 0.159, 0.153, 0.143, 0.140

*IMF Working Paper: Crypto, Corruption, and Capital Controls: Cross-Country Correlations (Conclusion and Appendix I).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022060-print-pdf.pdf_