## Dirty Money: Does the Risk of Infectious Disease Lower Demand for Cash?

## Source details

**Canonical URL:** [Dirty Money: Does the Risk of Infectious Disease Lower Demand for Cash?](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020255-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020255-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020255-print-pdf.pdf.json)

---

### Key findings and contributions
- First empirical investigation of whether the risk of infectious diseases affects demand for physical cash across a large panel of countries.
- Main empirical result: the spread of infectious diseases with human-to-human transmission lowers demand for physical cash after controlling for macroeconomic, financial, and technological factors.
- Magnitude example: "a one percentage-point increase in the number of confirmed infectious-disease cases scaled by population is associated with a decline of 1.68 percent in currency-in-circulation as a share of GDP."
- Result is economically and statistically significant and robust to several checks.
- Trade-offs highlighted:
  - Electronic payments may substitute for cash during infectious-disease outbreaks, but are not universally available because of financial and technological bottlenecks.
  - Cash also functions as a store of value during economic uncertainty.

### Data and stylized facts
- Sample:
  - Unbalanced panel of annual observations for 133 countries during the period 1995–2017.
  - Excluded: Euro-area countries, Switzerland, the United Kingdom, and the United States.
- Main variables:
  - Dependent variable: currency-in-circulation as a share of GDP (from IMF’s International Financial Statistics).
  - Main explanatory variable: number of confirmed cases of infectious diseases with human-to-human transmission (including Ebola and SARS), obtained from the WHO database and scaled by population.
  - Controls: real GDP per capita (local currency), informality (Medina and Schneider, 2019), average interest rate on bank deposits, domestic credit to the private sector as a share of GDP, mobile phone penetration (subscribers per 1,000) from World Bank WDI.
- Sample features and stationarity:
  - The sample used in this paper has 2,806 observations for infectious diseases, 1,505 out of which are non-zero.
  - Currency-in-circulation increased from a global average of 6.2 percent in 1995 to 8.8 percent in 2017.
  - Panel unit root tests (Im-Pesaran-Shin, 2003) indicate variables are stationary after logarithmic transformation.
- Contextual COVID-19 metrics (as reported in the paper):
  - As of November 19, 2020, over 56.4 million confirmed COVID-19 cases in 190 countries, with more than 1.3 million deaths.

### Empirical methodology
- Baseline specification: currency-in-circulation/GDP regressed on infectious-disease incidence scaled by population and control variables, with country fixed effects and time fixed effects.
- Estimation strategy: two-stage least squares (2SLS) with instrumental variables (IV) to address omitted variable bias and potential endogeneity.
- Instrumental variable (robustness): health spending as a share of GDP; infectious-disease series also instrumented with their own lags.
- Standard errors: robust and clustered at the country level.

### Baseline 2SLS-IV estimates (summary)
- Dependent variable: Currency-in-circulation/GDP.
- Key coefficient estimates (selected, robust standard errors in brackets):
  - Real GDP: -5.943*** [3.540]; -4.399*** [3.656]; -6.915*** [3.744]
  - Informality: 0.547** [0.243]; 0.473** [0.338]; 0.414** [0.258]
  - Deposit interest rate: -0.028 [0.083]; -0.012 [0.089]; -0.013 [0.088]
  - Financial development: -0.012** [0.053]; -0.009** [0.056]; -0.010** [0.059]
  - Mobile phone penetration: -0.080** [0.031]; -0.056** [0.036]; -0.048** [0.034]
  - Disease coefficients:
    - Ebola: -1.630*** [1.380]
    - SARS: -1.439** [1.365]
    - All infectious diseases: -1.682*** [1.861]
- Sample and fit:
  - Number of countries: 133
  - Number of observations: 1,880
  - Country FE: Yes; Time FE: Yes
  - F-stat: 52.4; 54.0; 52.5
  - [p-value]: 0.00; 0.00; 0.00
  - Adjusted R2: 0.77; 0.78; 0.80
- Interpretation: Coefficients on infectious-disease incidence are negative and statistically significant across specifications; example interpretation given for Ebola: "a one percentage point increase in the number of confirmed cases [of Ebola] is associated with a decline of 1.63 percent in currency-in-circulation as a share of GDP."

### Robustness checks (selected results)
- Approaches: truncated sample (5th and 95th percentiles removed); sub-sample (1995-2007); developing & low-income countries; alternative IV (health spending); excluding informality.
- Selected coefficient estimates (robust standard errors in brackets; dependent variable: Currency-in-circulation/GDP):
  - Real GDP:
    - Truncated sample: -1.478*** [0.262]
    - Sub-sample (1995-2007): -15.19*** [2.994]
    - Developing & low-income: -7.916*** [3.744]
    - Alternative IV (health spending): -6.092*** [3.165]
    - Excluding informality: -7.931*** [3.147]
  - Informality:
    - Truncated sample: 0.110*** [0.017]
    - Sub-sample (1995-2007): 0.145*** [0.135]
    - Developing & low-income: 0.415** [0.258]
    - Alternative IV: 0.509** [0.296]
  - Mobile phone penetration:
    - Truncated sample: -0.004** [0.002]
    - Sub-sample (1995-2007): -0.097** [0.035]
    - Developing & low-income: -0.019** [0.034]
    - Alternative IV: -0.046** [0.087]
    - Excluding informality: -0.013** [0.029]
  - All infectious diseases:
    - Truncated sample: -0.150*** [0.085]
    - Sub-sample (1995-2007): -0.242** [0.135]
    - Developing & low-income: -1.158*** [0.268]
    - Alternative IV (health spending): -1.852*** [1.110]
    - Excluding informality: -1.155*** [0.243]
- Robustness summary:
  - Negative and economically significant relationship between demand for cash and confirmed infectious-disease cases remains across checks.
  - Impact is larger in developing and low-income countries.
  - Alternative IV (health spending) confirms the negative impact.

### Interpretation and behavioral mechanisms
- Two potential behavioral responses to infectious-disease outbreaks:
  - Hoarding of cash for precautionary/store-of-value reasons.
  - Reduced cash use due to precautionary avoidance of potentially contaminated cash, boosting alternative payments.
- Empirical evidence supports the latter channel: infectious-disease spread reduces measured demand for cash.
- Transmission risk considerations:
  - Evidence is mixed on whether banknotes and coins pose greater health risk than other surfaces; perception of risk can drive behavior even absent definitive transmission evidence.
- Payment substitution constraints:
  - Electronic payment adoption may accelerate during pandemics, but access constraints (financial and technological) limit substitution in some countries.
  - Cash’s store-of-value role can become prominent during financial stress, potentially counteracting substitution effects.

### Policy implications and recommendations
- Support access to electronic payment methods while addressing financial and technological bottlenecks to avoid exclusion when cash use declines due to public-health concerns.
- Consider design and deployment of digital payment solutions (including central bank digital currency) that preserve financial inclusion and resilience.
- Address vulnerabilities of digital financial services to cyberattacks, digital fraud, and money-laundering when expanding digital payment infrastructure.
- Balance short-run contingency measures (e.g., sanitizing banknotes) and long-run payment system strategy, recognizing behavioral changes during outbreaks may not be permanent.
- Strengthen know-your-customer procedures and anti-money laundering compliance alongside digital transformation efforts.

### Conclusions
- This paper is the first empirical attempt to test whether risk of infectious diseases affects cash use intensity; using 2SLS-IV, results indicate increased confirmed cases of infectious disease with human-to-human transmission reduce demand for physical cash, controlling for other factors.
- Example summary reiterated: "a one percentage point increase in the number of confirmed infectious-disease cases is associated with a decline of 1.68 percent in currency-in-circulation as a share of GDP."
- The COVID-19 outbreak, spreading to 190 countries, can be expected to have a greater effect on currency-in-circulation over the long term.
- Findings do not amount to a recommendation to abolish cash; policy should manage digital transition benefits and risks, including inclusion, financial stability, cyber and AML vulnerabilities.

*Prepared by Serhan Cevik; IMF Working Paper WP/20/255, November 2020.*

### Section 1

### Dirty Money: Does the Risk of Infectious Disease Lower Demand for Cash?

### Key findings and contributions
- First empirical investigation of whether the risk of infectious diseases affects demand for physical cash across a large panel of countries.
- Main empirical result: the spread of infectious diseases with human-to-human transmission lowers demand for physical cash after controlling for macroeconomic, financial, and technological factors.
- Magnitude example: "a one percentage-point increase in the number of confirmed infectious-disease cases scaled by population is associated with a decline of 1.68 percent in currency-in-circulation as a share of GDP."
- Result is economically and statistically significant and robust to several checks.
- Highlights trade-offs: while electronic payments may substitute for cash during infectious-disease outbreaks, electronic payment methods are not universally available because of financial and technological bottlenecks; cash also functions as a store of value during economic uncertainty.

### Data and stylized facts
- Sample: unbalanced panel of annual observations for 133 countries during the period 1995–2017.
- Excluded from the sample: Euro-area countries, Switzerland, the United Kingdom, and the United States (to limit complications from foreign demand for reserve currencies).
- Main dependent variable: currency-in-circulation as a share of GDP (from IMF’s International Financial Statistics).
- Main explanatory variable: number of confirmed cases of infectious diseases with human-to-human transmission (including Ebola and SARS), obtained from the WHO database and scaled by population.
- Control variables: real GDP per capita (local currency), a novel measure of informality in economic activity (Medina and Schneider, 2019), average interest rate on bank deposits, domestic credit to the private sector as a share of GDP, and mobile phone penetration (subscribers per 1,000) from World Bank WDI.
- Sample features and stationarity:
  - The sample used in this paper has 2,806 observations for infectious diseases, 1,505 out of which are non-zero.
  - Currency-in-circulation shows significant dispersion across countries and over time.
  - Currency-in-circulation increased from a global average of 6.2 percent in 1995 to 8.8 percent in 2017.
  - Panel unit root tests (Im-Pesaran-Shin, 2003) indicate variables are stationary after logarithmic transformation (results available upon request).

- Contextual COVID-19 metrics (as reported in the paper):
  - As of November 19, 2020, over 56.4 million confirmed COVID-19 cases in 190 countries, with more than 1.3 million deaths.

### Empirical methodology
- Baseline specification: currency-in-circulation as a share of GDP (dependent variable) regressed on infectious-disease incidence scaled by population and vector of control variables, with country fixed effects and time fixed effects.
- Estimation strategy: two-stage least squares (2SLS) with instrumental variables (IV) to address omitted variable bias and potential endogeneity (reverse causality between cash use and disease spread).
- Instrumental variable used (robustness check): health spending as a share of GDP.
- Standard errors: robust and clustered at the country level to account for heteroskedasticity.

### Interpretation and mechanisms
- Two potential behavioral responses to infectious-disease outbreaks:
  - Hoarding of cash by consumers and businesses for precautionary reasons (store-of-value channel).
  - Reduced cash use due to precautionary avoidance of potentially contaminated cash, boosting alternative payment methods (cards, mobile payments, central bank digital currency).
- Empirical evidence in the paper supports the latter channel: infectious-disease spread reduces measured demand for cash.
- Considerations on transmission risk: evidence is mixed on whether banknotes and coins pose greater health risk than other surfaces (credit cards, mobile phones); perception of risk can drive behavior even absent definitive transmission evidence.
- Policy-relevant nuances:
  - Electronic payment adoption may accelerate during pandemics, but access constraints in some countries can limit substitution from cash to electronic payments.
  - Cash’s role as a store of value can gain prominence in episodes of financial stress (e.g., global financial crisis), potentially counteracting substitution effects in some settings.
  - Emerging instruments (e.g., central bank digital currency) can improve payment efficiency and distribution of transfers, but raise inclusion and cybersecurity concerns.

### Policy implications and recommendations
- Support access to electronic payment methods while addressing financial and technological bottlenecks to avoid exclusion when cash use declines due to public-health concerns.
- Consider design and deployment of digital payment solutions (including central bank digital currency) that preserve financial inclusion and resilience.
- Address vulnerabilities of digital financial services to cyberattacks, digital fraud, and money-laundering attempts when expanding digital payment infrastructure.
- Recognize that behavioral changes during outbreaks may not be permanent; policy should balance short-run contingency measures (e.g., sanitizing banknotes) and long-run payment system strategy.

*Prepared by Serhan Cevik; IMF Working Paper WP/20/255, November 2020.*

### Section 2

### Infectious Diseases and Demand for Cash—Baseline Estimations (Section 2)

### Baseline 2SLS-IV results (Table 2)
- Dependent variable: Currency-in-circulation/GDP.
- Key coefficient estimates (robust standard errors in brackets):
  - Real GDP: -5.943*** [3.540]; -4.399*** [3.656]; -6.915*** [3.744]
  - Informality: 0.547** [0.243]; 0.473** [0.338]; 0.414** [0.258]
  - Deposit interest rate: -0.028 [0.083]; -0.012 [0.089]; -0.013 [0.088]
  - Financial development: -0.012** [0.053]; -0.009** [0.056]; -0.010** [0.059]
  - Mobile phone penetration: -0.080** [0.031]; -0.056** [0.036]; -0.048** [0.034]
  - Ebola: -1.630*** [1.380] (introduced in column (1))
  - SARS: -1.439** [1.365] (introduced in column (2))
  - All infectious diseases: -1.682*** [1.861] (introduced in column (3))
- Sample and fit:
  - Number of countries: 133 (all columns)
  - Number of observations: 1,880 (all columns)
  - Country FE: Yes; Time FE: Yes
  - F-stat: 52.4; 54.0; 52.5
  - [p-value]: 0.00; 0.00; 0.00
  - Adjusted R2: 0.77; 0.78; 0.80
- Note: 2SLS-IV approach used; infectious disease series instrumented with their own lags and alternatively with health spending as a share of GDP. Robust standard errors reported. *, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively.

### Interpretation of baseline estimates
- Conventional determinants:
  - Currency-in-circulation as a share of GDP is inversely related to the level of income (Real GDP negative and significant).
  - Informality is positively associated with currency-in-circulation.
  - Deposit interest rate has a negative coefficient but is not statistically significant.
  - Financial development is negatively associated with demand for cash.
  - Mobile phone penetration lowers currency-in-circulation, consistent with enabling alternative payments.
- Main variable (infectious diseases):
  - Coefficients are negative and statistically significant across specifications.
  - Magnitudes reported:
    - SARS: -1.439
    - Ebola: -1.630
    - All infectious diseases (Ebola, malaria, SARS, yellow fever): -1.682
  - Example interpretation: "a one percentage point increase in the number of confirmed cases [of Ebola] is associated with a decline of 1.63 percent in currency-in-circulation as a share of GDP."
  - Conclusion: Higher risk of infectious disease with human-to-human transmission is associated with lower intensity of cash use, controlling for macroeconomic, financial, and technological factors.

### Robustness checks (Table 3)
- Approaches applied:
  - Truncated sample (5th and 95th percentiles removed)
  - Sub-sample (1995-2007)
  - Developing & low-income countries
  - Alternative IV (health spending)
  - Excluding informality
- Selected coefficient estimates (robust standard errors in brackets; dependent variable: Currency-in-circulation/GDP):
  - Real GDP:
    - Truncated sample: -1.478*** [0.262]
    - Sub-sample (1995-2007): -15.19*** [2.994]
    - Developing & low-income: -7.916*** [3.744]
    - Alternative IV (health spending): -6.092*** [3.165]
    - Excluding informality: -7.931*** [3.147]
  - Informality:
    - Truncated sample: 0.110*** [0.017]
    - Sub-sample (1995-2007): 0.145*** [0.135]
    - Developing & low-income: 0.415** [0.258]
    - Alternative IV: 0.509** [0.296]
  - Deposit interest rate:
    - Truncated sample: -0.012 [0.007]
    - Sub-sample (1995-2007): -0.044 [0.051]
    - Developing & low-income: -0.013 [0.009]
    - Alternative IV: -0.029 [0.085]
    - Excluding informality: -0.007 [0.082]
  - Financial development:
    - Truncated sample: -0.011** [0.004]
    - Sub-sample (1995-2007): -0.006** [0.015]
    - Developing & low-income: -0.049** [0.059]
    - Alternative IV: -0.028** [0.056]
    - Excluding informality: -0.040** [0.052]
  - Mobile phone penetration:
    - Truncated sample: -0.004** [0.002]
    - Sub-sample (1995-2007): -0.097** [0.035]
    - Developing & low-income: -0.019** [0.034]
    - Alternative IV: -0.046** [0.087]
    - Excluding informality: -0.013** [0.029]
  - All infectious diseases:
    - Truncated sample: -0.150*** [0.085]
    - Sub-sample (1995-2007): -0.242** [0.135]
    - Developing & low-income: -1.158*** [0.268]
    - Alternative IV (health spending): -1.852*** [1.110]
    - Excluding informality: -1.155*** [0.243]
- Sample and fit for robustness checks:
  - Number of countries: 126; 128; 105; 130; 133 (columns respectively)
  - Number of observations: 1,661; 919; 1,840; 1,521; 2,069
  - Country FE: Yes (all)
  - Time FE: Yes (all)
  - F-stat: 78.9; 31.3; 46.9; 41.7; 48.0
  - [p-value]: 0.000; 0.000; 0.000; 0.000; 0.000
  - Adjusted R2: 0.86; 0.79; 0.77; 0.77; 0.77
- Robustness summary:
  - Negative and economically significant relationship between demand for cash and number of confirmed infectious-disease cases remains across checks.
  - Impact is larger in developing and low-income countries.
  - Using health spending as an alternative instrument confirms the negative impact.

### Conclusions and policy considerations (Section IV)
- Empirical findings:
  - This paper is the first empirical attempt to test whether risk of infectious diseases affects cash use intensity.
  - Using 2SLS-IV to address endogeneity, results indicate increased confirmed cases of infectious disease with human-to-human transmission (e.g., Ebola, SARS) reduce demand for physical cash, controlling for other factors.
  - Example summary: "a one percentage point increase in the number of confirmed infectious-disease cases is associated with a decline of 1.68 percent in currency-in-circulation as a share of GDP."
  - The COVID-19 outbreak, spreading to 190 countries, can be expected to have a greater effect on currency-in-circulation over the long term.
- Policy-relevant observations:
  - Findings are not a recommendation to abolish cash.
  - Electronic payment methods may not be universally available because of financial infrastructure and technological bottlenecks.
  - Cash remains a store of value, particularly important during economic uncertainty or when banks are vulnerable.
  - COVID-19 has catalyzed a shift toward digital money (including central bank digital currency), driven by:
    - Significant increase in online shopping where cash is not an option.
    - Businesses and consumers preferring digital payments to minimize cash use.
    - Preliminary data showing a decline in ATM withdrawals and an increase in mobile applications for transactions.
- Risks and trade-offs of digital transformation:
  - Digital transformation should not be treated as binary—constraints in many countries limit access to electronic payments.
  - Potential benefits: faster processing, greater variety, efficiency, convenience, improved disbursement of cash transfers during disasters and crises.
  - Potential costs and vulnerabilities:
    - Greater cross-border spillovers.
    - Risks to financial stability.
    - Complications for macroeconomic policy.
    - Increased vulnerability to cyberattacks and fraud, especially in developing countries with lagging digital infrastructure.
    - Need to strengthen know-your-customer procedures and anti-money laundering compliance.

*Source: IMF; World Bank; WHO; author's calculations.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020255-print-pdf.pdf_