## 8. Sensitivity Analysis

## Source details

**Canonical URL:** [8. Sensitivity Analysis](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05175.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05175.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05175.pdf.json)

---

### Dataset and approach
- Annual panel data covering 1960–1999 for around 100 countries; seigniorage data available for 144 countries but missing explanatory variables reduce estimations to a maximum of 104 countries.
- Seigniorage measured two ways: change in reserve money (IFS line 14a) divided by nominal GDP (IFS line 99b) and change in reserve money divided by government revenues (IFS line 81).
- Estimation method: panel regressions with country fixed effects; political instability proxy (cabinet changes) is lagged one period (PI_{i,t-1}).

### Main empirical findings — determinants of seigniorage
- Political instability:
  - An additional cabinet change increases seigniorage by around 0.24 (t-statistics ~1.8–1.97), which is an increase of 13 percent relative to the sample mean of 1.87 (dependent variable ∆RM/GDP).
- Economic freedom:
  - A one-point increase in the Index of Economic Freedom reduces seigniorage by roughly 1.2 (a decrease of 64 percent relative to the sample mean).
  - Of the Index’s five areas, Area III (access to sound money) and Area IV (freedom to exchange with foreigners) are statistically significant with negative signs.
- Democracy (Polity Scale): not robustly significant for seigniorage levels in baseline specifications.
- Economic structure and performance:
  - Agriculture (percent of GDP): positive association with seigniorage in several specifications.
  - Growth of real GDP per capita: statistically significant negative effect on seigniorage.
  - Real GDP per capita: negative coefficient in most specifications (often highly significant).
- Time effects:
  - Decade dummies indicate higher seigniorage in the 1980s, followed by the 1990s and the 1970s; the 1960s had the lowest levels.

### Circumstances under which political instability raises seigniorage more (interaction results)
- Inflation regime:
  - Political instability (cabinet changes) raises seigniorage primarily when annual inflation ≥ 50 percent.
  - Interaction coefficients: [Cabinet changes * (inflation ≥ 50 percent)] ≈ 1.632 (t-statistic 2.49) in one specification; other instability proxies yield comparable sizable, significant interaction terms (e.g., Government crises * inflation ≥ 50 percent = 3.908 (4.15) and Index of political cohesion * inflation ≥ 50 percent = 1.763 (3.73)).
- Development status:
  - Effects are stronger in developing countries; interaction [Cabinet changes * (dev. countries)] ≈ .366 (2.71) while [Cabinet changes * (ind. countries)] is insignificant.
- Social polarization:
  - Political instability increases seigniorage only for countries with average Gini > 40; [Cabinet changes * (Gini > 40)] ≈ .531 (2.67).
- Central bank de facto independence:
  - Higher effects when turnover rate of central bank presidents is high (dummy high turnover defined as turnover > sample median 0.20): [Cabinet changes * (high turnover)] ≈ .475 (1.97).
- Economic freedom:
  - Effects concentrated in low economic freedom countries (dummy high economic freedom = Index > 5); [Cabinet changes * (low econ. freedom)] ≈ .774 (3.41).
- Democracy:
  - Political instability has lower effects when Polity scale > 0 (democracies); [Cabinet changes * (polity scale ≤ 0)] ≈ .364 (2.13) and [Cabinet changes * (polity scale > 0)] ≈ .222 (1.89).
- Financing constraints:
  - Domestic debt: higher effects when average domestic debt/GDP is above median (median = 13.28); [Cabinet changes * (high domestic debt)] ≈ .676 (2.20).
  - Creditworthiness: higher effects when creditworthiness is low (high creditworthiness dummy = average Euromoney > 60, so low creditworthiness drives the effect); [Cabinet changes * (low creditworthiness)] ≈ .375 (2.59).
- Regional and temporal concentration:
  - Effects felt mainly during the 1970s and 1980s.
  - Regionally, the positive effect of political instability on seigniorage (∆RM/GDP) is statistically significant for Western Hemisphere (Latin America); when seigniorage is defined as ∆RM/GR, significant effects also appear for Africa.

### Robustness and sensitivity checks
- Alternative instability proxies:
  - Government crises, executive changes, and the DPI index of political cohesion produce similar interaction patterns — significant effects only when inflation ≥ 50 percent.
- Alternative seigniorage definition:
  - Results hold when seigniorage is defined as ∆RM/Government Revenues (sensitivity tables show consistent coefficient signs and significance in comparable specifications).
- Sample restrictions:
  - Estimates for developing-country-only samples produce similar conclusions.
- Persistence of instability:
  - Three-year moving averages of cabinet changes yield results consistent with annual measures; interaction patterns remain significant in high-inflation and developing-country contexts.
- Other robustness findings:
  - Currency inside banks (more currency deposited) is associated with lower seigniorage; de facto fixed exchange rate regimes are associated with lower seigniorage.
  - Lower international creditworthiness is associated with higher seigniorage (Click, 1998 result confirmed).

### Key numeric descriptors (selected)
- Seigniorage (∆ Reserve money as percent of GDP): Obs 4,761; Mean 1.87; Std. Dev. 3.62; Min -29.40; Max 65.53 (IMF-IFS).
- Seigniorage (∆ Reserve money as percent of government revenues): Obs 3,889; Mean 122.22; Std. Dev. 3,355.86; Min -380.78; Max 151,882.8 (IMF-IFS).
- Cabinet changes (CNTS): Obs 5,667; Mean 0.44; Std. Dev. 0.60; Min 0; Max 5.
- Index of Economic Freedom: Obs 2,958; Mean 5.70; Std. Dev. 1.19; Min 2.30; Max 9.05.
- Gini coefficient: Obs 693; Mean 37.49; Std. Dev. 10.64; Min 16.63; Max 74.33.
- Turnover rate governors (Cukierman & Webb): Obs 1,990; Mean 0.24; Std. Dev. 0.20; Min 0; Max 1.08.
- Median domestic debt/GDP used to define high domestic debt = 13.28.

### Policy implications (from empirical results)
- Institutional and policy reforms that can limit seigniorage reliance when political instability is present:
  - Increase economic freedom (higher Index of Economic Freedom), notably access to sound money and freedom to exchange with foreigners.
  - Strengthen de facto central bank independence (reduce turnover of central bank presidents).
  - Improve international creditworthiness (reduce borrowing constraints).
  - Reduce income inequality (lower Gini coefficients) to limit social polarization.
  - Strengthen democratic institutions, budgeting procedures, and transparency to reduce the transmission from political instability to seigniorage.
- Expected outcome: reduced dependence on seigniorage should help lower inflation over time and support higher growth and economic prosperity.

*Source: _wp05175 - 8. Sensitivity Analysis (IMF working paper content provided).*

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

### References................................................................................................................................24

### Tables
- 1.         Seigniorage         Across         Countries......................................................................................14
- 2.         Descriptive         Statistics....................................................................................................16
- 3.         Results         for         Seigniorage ................................................................................................18
- 4.         Robustness         Tests ..........................................................................................................19
- 5. Results for Interactions of Cabinet Changes................................................................20
- 6. More Results for Interactions of Cabinet Changes ......................................................21
- 7. Results for Interactions of Other Proxies of Political Instability .................................22

* _wp05175 - References..............................................................................................................*

### 8.         Sensitivity         Analysis ................................................................................

### 8. Sensitivity Analysis

### Dataset and approach
- Annual panel data covering 1960–1999 for around 100 countries; seigniorage data available for 144 countries but missing explanatory variables reduce estimations to a maximum of 104 countries.
- Seigniorage measured two ways: change in reserve money (IFS line 14a) divided by nominal GDP (IFS line 99b) and change in reserve money divided by government revenues (IFS line 81).
- Estimation method: panel regressions with country fixed effects; political instability proxy (cabinet changes) is lagged one period (PI_{i,t-1}).

### Main empirical findings — determinants of seigniorage
- Political instability:
  - An additional cabinet change increases seigniorage by around 0.24 (t-statistics ~1.8–1.97), which is an increase of 13 percent relative to the sample mean of 1.87 (dependent variable ∆RM/GDP).
- Economic freedom:
  - A one-point increase in the Index of Economic Freedom reduces seigniorage by roughly 1.2 (a decrease of 64 percent relative to the sample mean).
  - Of the Index’s five areas, Area III (access to sound money) and Area IV (freedom to exchange with foreigners) are statistically significant with negative signs.
- Democracy (Polity Scale): not robustly significant for seigniorage levels in baseline specifications.
- Economic structure and performance:
  - Agriculture (percent of GDP): positive association with seigniorage in several specifications.
  - Growth of real GDP per capita: statistically significant negative effect on seigniorage.
  - Real GDP per capita: negative coefficient in most specifications (often highly significant).
- Time effects:
  - Decade dummies indicate higher seigniorage in the 1980s, followed by the 1990s and the 1970s; the 1960s had the lowest levels.

### Circumstances under which political instability raises seigniorage more (interaction results)
- Inflation regime:
  - Political instability (cabinet changes) raises seigniorage primarily when annual inflation ≥ 50 percent.
  - Interaction coefficients: [Cabinet changes * (inflation ≥ 50 percent)] ≈ 1.632 (t-statistic 2.49) in one specification; other instability proxies yield comparable sizable, significant interaction terms (e.g., Government crises * inflation ≥ 50 percent = 3.908 (4.15) and Index of political cohesion * inflation ≥ 50 percent = 1.763 (3.73)).
- Development status:
  - Effects are stronger in developing countries; interaction [Cabinet changes * (dev. countries)] ≈ .366 (2.71) while [Cabinet changes * (ind. countries)] is insignificant.
- Social polarization:
  - Political instability increases seigniorage only for countries with average Gini > 40; [Cabinet changes * (Gini > 40)] ≈ .531 (2.67).
- Central bank de facto independence:
  - Higher effects when turnover rate of central bank presidents is high (dummy high turnover defined as turnover > sample median 0.20): [Cabinet changes * (high turnover)] ≈ .475 (1.97).
- Economic freedom:
  - Effects concentrated in low economic freedom countries (dummy high economic freedom = Index > 5); [Cabinet changes * (low econ. freedom)] ≈ .774 (3.41).
- Democracy:
  - Political instability has lower effects when Polity scale > 0 (democracies); [Cabinet changes * (polity scale ≤ 0)] ≈ .364 (2.13) and [Cabinet changes * (polity scale > 0)] ≈ .222 (1.89).
- Financing constraints:
  - Domestic debt: higher effects when average domestic debt/GDP is above median (median = 13.28); [Cabinet changes * (high domestic debt)] ≈ .676 (2.20).
  - Creditworthiness: higher effects when creditworthiness is low (high creditworthiness dummy = average Euromoney > 60, so low creditworthiness drives the effect); [Cabinet changes * (low creditworthiness)] ≈ .375 (2.59).
- Regional and temporal concentration:
  - Effects felt mainly during the 1970s and 1980s.
  - Regionally, the positive effect of political instability on seigniorage (∆RM/GDP) is statistically significant for Western Hemisphere (Latin America); when seigniorage is defined as ∆RM/GR, significant effects also appear for Africa.

### Robustness and sensitivity checks
- Alternative instability proxies:
  - Government crises, executive changes, and the DPI index of political cohesion produce similar interaction patterns — significant effects only when inflation ≥ 50 percent.
- Alternative seigniorage definition:
  - Results hold when seigniorage is defined as ∆RM/Government Revenues (sensitivity tables show consistent coefficient signs and significance in comparable specifications).
- Sample restrictions:
  - Estimates for developing-country-only samples produce similar conclusions.
- Persistence of instability:
  - Three-year moving averages of cabinet changes yield results consistent with annual measures; interaction patterns remain significant in high-inflation and developing-country contexts.
- Other robustness findings:
  - Currency inside banks (more currency deposited) is associated with lower seigniorage; de facto fixed exchange rate regimes are associated with lower seigniorage.
  - Lower international creditworthiness is associated with higher seigniorage (Click, 1998 result confirmed).

### Key numeric descriptors (selected)
- Seigniorage (∆ Reserve money as percent of GDP): Obs 4,761; Mean 1.87; Std. Dev. 3.62; Min -29.40; Max 65.53 (IMF-IFS).
- Seigniorage (∆ Reserve money as percent of government revenues): Obs 3,889; Mean 122.22; Std. Dev. 3,355.86; Min -380.78; Max 151,882.8 (IMF-IFS).
- Cabinet changes (CNTS): Obs 5,667; Mean 0.44; Std. Dev. 0.60; Min 0; Max 5.
- Index of Economic Freedom: Obs 2,958; Mean 5.70; Std. Dev. 1.19; Min 2.30; Max 9.05.
- Gini coefficient: Obs 693; Mean 37.49; Std. Dev. 10.64; Min 16.63; Max 74.33.
- Turnover rate governors (Cukierman & Webb): Obs 1,990; Mean 0.24; Std. Dev. 0.20; Min 0; Max 1.08.
- Median domestic debt/GDP used to define high domestic debt = 13.28.

### Policy implications (from empirical results)
- Institutional and policy reforms that can limit seigniorage reliance when political instability is present:
  - Increase economic freedom (higher Index of Economic Freedom), notably access to sound money and freedom to exchange with foreigners.
  - Strengthen de facto central bank independence (reduce turnover of central bank presidents).
  - Improve international creditworthiness (reduce borrowing constraints).
  - Reduce income inequality (lower Gini coefficients) to limit social polarization.
  - Strengthen democratic institutions, budgeting procedures, and transparency to reduce the transmission from political instability to seigniorage.
- Expected outcome: reduced dependence on seigniorage should help lower inflation over time and support higher growth and economic prosperity.

*Source: _wp05175 - 8. Sensitivity Analysis (IMF working paper content provided).*

### References

### References

### Reference list omitted

*Source: _wp05175 - References (https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05175.pdf)*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05175.pdf_