## _wp0842

## Source details

**Canonical URL:** [_wp0842](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp0842.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp0842.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp0842.pdf.json)

---

### I. INTRODUCTION — Scope and motivation
- Literature on whether budget deficits affect interest rates is heterogeneous despite about a hundred papers for advanced economies.
- For the U.S., Gale and Orszag (2003) count about 30 studies that find robustly positive effects of deficits on interest rates, and about 30 that do not.
- Bernheim (1989): “it is easy to cite a large number of studies that support any conceivable position.”
- Paper extends the literature by addressing angles that received little prior attention.

### Main contributions
- Uses a panel dataset of 60 advanced and emerging economies.
- Exploits cross-country variation to examine interactions between budget deficits and structural characteristics.
- Employs system GMM estimation as an alternative to VAR or conventional instrumental variable techniques.

### Key empirical findings (overview)
- Highly significant positive effect of budget deficits on interest rates:
  - Estimated magnitude about 26 basis points per 1 percent of GDP for the complete panel.
- Heterogeneity by country group and period:
  - Effects larger and more robust in emerging markets and in later periods than in advanced economies and earlier periods.
- Effect of deficits on interest rates significant only under certain conditions:
  - Deficits are high.
  - Deficits are mostly domestically financed.
  - Deficits interact with high domestic debt.
  - Financial openness is low.
  - Interest rates are liberalized.
  - Financial depth is low.

### Interpretation and policy implications (overview)
- Interaction effects may explain much of the heterogeneity in the empirical literature.
- Fiscal stabilization is more effective when the initial budget deficit and level of debt are lower, and when financial openness and financial depth are greater, because the effect of deficits on interest rates is smaller under these conditions (implying less crowding out).

### Source: _wp0842 - References

### Model and Methodology (Section II)
- Estimated equation (reduced form of nominal interest rate it):
  - log it = δ0 + δ1 i* t + δ2 e_it + δ3 π^e_it + δ4 m_it + δ5 y_it + δ6 d_it + δ7 r_it + δ8 s_it + δ9 lagged_it + ε_it
  - Lagged dependent variable accounts for delayed adjustment.
- Theoretical polar cases:
  - Perfect capital mobility: i determined by i*, expected depreciation (e), and country-specific risk spread (r).
  - Closed economy: i determined by expected inflation (π^e), real money supply (m), and real interest rate (via real GDP growth y, δ0, and budget deficit d).
- Null hypothesis: deficits have no effect, i.e., δ7 = 0.
- Determinants of deficit–interest rate effect discussed:
  - Degree of capital mobility.
  - Share of deficits financed domestically vs externally.
  - Degree of debt neutrality (Ricardian equivalence).
  - Extent to which interest rates are market-determined.
- Identification and estimation:
  - Multicollinearity among inflation, depreciation, and budget deficit addressed by marginal effects and staggered regressions.
  - Potential endogeneity addressed via two-step system GMM (Arellano and Bover (1995); Blundell and Bond (1998)), implemented with Roodman (2006) program and Windmeijer (2005) standard error correction.
  - Time dummies included; orthogonal deviations used.
  - Foreign interest rate and time dummies treated as exogenous; all other variables treated as endogenous.
  - Instruments: up to six lags, collapsed and limited to number of groups to address Hansen test weakness.
  - Pooled estimators preferred over heterogeneous panel estimators to exploit cross-country and cross-period heterogeneity via interaction terms.

### Data (Section III)
- Sample: 1970–2006 panel of 60 advanced and emerging economies; U.S. excluded.
- Data sources and variable construction:
  - Most data from IMF International Financial Statistics (IFS); budget deficit and GDP from IMF World Economic Outlook Database (WEO).
  - Foreign rate: yield of one-year U.S. treasury bills (IFS line 61).
  - Depreciation: percent increase in the U.S. dollar exchange rate (IFS line rf).
  - Risk spread proxied by International Country Risk Guide (ICRG) rating (available from 1984).
  - Inflation: average annual change in consumer price index (IFS line 64).
  - Real money supply: growth rate of ratio of broad money (IFS line 35l) to GDP (nominal GDP used as deflator).
  - Real GDP growth: WEO line NGDP_R.
- Interest rates:
  - Preferred long rates limited; use short rates with maturity of three to twelve months.
  - Pooling order: T-bill rates (IFS line 60c), money market rates (IFS line 60b), savings rates (IFS line 60l), lending rates (IFS line 60p); ratio splicing where needed.
- Temporal aggregation:
  - Data averaged over five-year non-overlapping periods (including 2005–06 in last window).
- Interaction and institutional measures:
  - Domestic financing (IFS line 84a), domestic debt (IFS line 88a), financial depth measured by liquid liabilities of banking system (IFS line 55l or 34+35 if unavailable), all in percent of GDP.
  - Indices: capital account liberalization (Chinn and Ito (2006)) and interest rate liberalization (Abiad et al. (2007)); higher values imply greater liberalization.
- Expectations:
  - Adaptive expectations equal to current conditions; five-year averaging and limited survey data motivate this choice.
- Data cleaning:
  - Hyperinflations: 24 observations removed where five-year average nominal interest rate exceeded 100 percent.
- Descriptive statistics (Table 1, preserved exactly):
  - Interest rate Mean 10.44 Std. Dev. 8.64 Min. 0.11 Max. 49.99
  - Foreign rate Mean 7.71 Std. Dev. 2.29 Min. 4.73 Max. 12.38
  - Depreciation Mean 9.22 Std. Dev. 28.51 Min. -50.39 Max. 262.56
  - Risk spread Mean -71.23 Std. Dev. 12.16 Min. -93.58 Max. -28.27
  - Inflation Mean 15.41 Std. Dev. 41.01 Min. -0.38 Max. 626.07
  - Money supply Mean 1.68 Std. Dev. 5.12 Min. -14.75 Max. 24.82
  - GDP growth Mean 3.65 Std. Dev. 2.82 Min. -11.24 Max. 12.68
  - Budget deficit Mean 3.11 Std. Dev. 5.24 Min. -16.07 Max. 36.27
  - Domestic financing Mean 2.23 Std. Dev. 2.93 Min. -8.86 Max. 15.05
  - Domestic debt Mean 27.93 Std. Dev. 20.57 Min. 0.06 Max. 109.32
  - Financial depth Mean 53.78 Std. Dev. 55.53 Min. 8.84 Max. 416.82
  - Capital account liberalization Mean 0.36 Std. Dev. 1.55 Min. -1.75 Max. 2.62
  - Interest rate liberalization Mean 1.78 Std. Dev. 1.30 Min. 0.00 Max. 3.00

### Results (Section IV)

A. Baseline
- Marginal effects assessed to address multicollinearity among deficits, inflation, and depreciation.
- Baseline numeric findings:
  - Initial regression: increase in the deficit by 1 percent of GDP increases the interest rate by 44 basis points (coefficient 0.438).
  - Adding other variables progressively reduces deficit coefficient to 26 basis points (final baseline coefficient 0.259).
  - U.S. literature comparison: range between 20-60 basis points per 1 percent of GDP (Gale and Orszag (2003)).
  - Lagged interest rate never significant (five-year averaging removes dynamic effects).
  - Selected coefficients in some specifications:
    - Inflation coefficient about 0.41 in some specifications.
    - Foreign rate coefficients across specifications: 3.395, 1.498, 2.005, 1.747, 1.333, 2.374 (with respective standard errors).
  - Marginal effects when adding controls:
    - Adding international rate: deficit coefficient unchanged.
    - Adding depreciation: reduces deficit coefficient substantially → some deficit effect operates indirectly via depreciation.
    - Adding risk premium reduces observations; adding inflation reduces deficit coefficient (consistent with deficits being inflationary).
    - Adding money supply increases effect of deficits → suggests accommodative monetary policy dampens deficit effect when not controlled for.
    - GDP growth reduces effect of deficits (likely reflects positive short-run correlation between deficits and GDP growth).
- Table 2 baseline budget deficit coefficients across specifications: 0.438, 0.438, 0.312, 0.559, 0.484, 0.586, 0.259 (standard errors in parentheses).

B. Periods and Regions
- Time variation:
  - Time-dummy evidence: deficits had a negative effect on interest rates during 1985–1994 and became significantly positive from 1995 onwards.
  - Specific period effects (interpretation in text uses differences):
    - 2000–06 effect of 38 basis points.
    - 1995–99 effect of 170 basis points (large and possibly influenced by crises; low significance).
  - Table 3 reported combined baseline coefficient 3.284 and period interaction coefficients:
    - Budget deficit*1985-89 -3.395
    - Budget deficit*1990-94 -3.400
    - Budget deficit*1995-99 -1.578
    - Budget deficit*2000-06 -2.905
- Regional/country-group findings:
  - Emerging markets: increase in deficit ratio raises interest rates by 24 basis points.
  - Advanced economies: no significant effect when run separately.
  - Latin America: effect largest (magnitude described as implausibly large, possibly due to crisis episodes).
  - Emerging Asia and Emerging Europe: effect about 60 basis points each.
  - Middle East and Africa: effect highly significantly negative (possibly reflecting financial repression).
- Conclusion: Table 2 results for entire sample driven by emerging economies.

C. Interactions
- Baseline with interaction terms using “high”/“low” dummies (above/below median) — selected key numeric results (Table 4, preserved exactly where presented):
  - Deficits matter more when:
    - High deficits: effect 52 basis points (Budget deficit*High budget deficit coefficient 0.516 (0.081)***).
    - High domestic financing: effect 92 basis points (Budget deficit*High domestic financing coefficient 0.923 (0.186)***).
    - High domestic debt: effect 65 basis points when controlling for financial depth (Budget deficit*High domestic debt coefficient 0.652).
  - Domestic vs external financing:
    - Effect much larger and significant only when deficits mostly financed domestically (coefficient 0.923).
  - Financial openness:
    - Low financial openness: deficit effect 67 basis points (Budget deficit*Low capital account openness coefficient 0.673 (0.069)***).
    - High financial openness: effect not significant (Budget deficit*High capital account openness coefficient 0.022).
  - Interest rate liberalization:
    - High liberalization: effect 67 basis points (Budget deficit*High interest rate liberalization coefficient 0.674 (0.061)***).
    - Low liberalization: effect 45 basis points (Budget deficit*Low interest rate liberalization coefficient 0.449). Difference significant at 13 percent level.
  - Financial depth:
    - Low financial depth: effect 200 basis points (Budget deficit*Low financial depth coefficient 2.108 (0.454)***).
    - High financial depth: effect not significant (Budget deficit*High financial depth coefficient -0.111).
    - Financial depth as continuous variable: coefficient -0.017 (significant at 10 percent), indicating deeper financial systems reduce deficit effect.
- Additional notes:
  - Nonlinearities and interaction with debt consistent with literature.
  - Several other interactions tested (exchange rate regime, political instability) yielded no significant effects.
  - Collinearity prevents including all interactions jointly; sample size limits region- or period-specific interaction analysis.
- Selected interaction coefficients (Table 4 reproduced values):
  - Budget deficit*High budget deficit 0.516 (0.081)***
  - Budget deficit*High domestic financing 0.923 (0.186)***
  - Budget deficit*High domestic debt 0.876 (0.135)***
  - Budget deficit*Low financial depth 2.108 (0.454)***
  - Budget deficit*Low capital account openness 0.673 (0.069)***
  - Budget deficit*High interest rate liberalization 0.674 (0.061)***

### Conclusions (Section V) — Findings and policy implications
- Three main conclusions:
  - For the complete panel, highly significant positive effect of budget deficits on interest rates in the order of about 26 basis points per 1 percent of GDP.
  - Effect varies by country group and time period: larger and more robust in emerging markets and in later periods than in advanced economies and earlier periods.
  - Effect depends on interaction terms and is significant only under certain conditions:
    - When deficits are high.
    - When deficits are mostly domestically financed.
    - When deficits interact with high domestic debt.
    - When financial openness is low.
    - Effect is larger when interest rates are more liberalized and when the domestic financial sector is less developed.
- Policy implications:
  - Fiscal stabilization more effective when initial budget deficit and level of debt are lower, and when financial openness and financial depth are greater, because the effect of deficits on interest rates is smaller under these conditions (implying less crowding out and a greater multiplier).

*Source — _wp0842 (sections II–V) as provided.*

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

### References

### I. INTRODUCTION — Scope and motivation
- The literature on whether budget deficits affect interest rates remains heterogeneous despite about a hundred papers examining the relationship for the advanced economies.
- For the U.S., Gale and Orszag (2003) count about 30 studies that find robustly positive effects of deficits on interest rates, and about 30 that do not.
- Bernheim (1989) is cited: “it is easy to cite a large number of studies that support any conceivable position.”
- Several important angles have received little or no attention prior to this work.

### Main contributions of the paper
- Extends the literature in three new directions:
  - Uses a panel dataset of 60 advanced and emerging economies, providing evidence on emerging markets and enabling comparisons between advanced and emerging economies.
  - Exploits cross-country variation to examine interactions between budget deficits and structural characteristics of the economy to help explain heterogeneity in prior findings.
  - Employs system GMM estimation, as an alternative to VAR or conventional instrumental variable techniques, to potentially achieve greater efficiency.

### Key empirical findings
- There is a highly significant positive effect of budget deficits on interest rates:
  - The estimated magnitude is about 26 basis points per 1 percent of GDP for the complete panel.
- Heterogeneity by country group and period:
  - Effects are larger and more robust in emerging markets and in later periods than in advanced economies and in earlier periods.
- Interaction-dependent significance:
  - The effect of budget deficits on interest rates is only significant under certain conditions, including:
    - Deficits are high.
    - Deficits are mostly domestically financed.
    - Deficits interact with high domestic debt.
    - Financial openness is low.
    - Interest rates are liberalized.
    - Financial depth is low.

### Interpretation and policy implications
- Interaction effects may explain much of the heterogeneity observed in the empirical literature.
- Fiscal policy effectiveness:
  - Fiscal stabilization is more effective when the initial budget deficit and level of debt are lower, and when financial openness and financial depth are greater.
  - Rationale: the effect of deficits on interest rates is smaller under these conditions, implying less crowding out.

*Source: _wp0842 - References*

### section II discusses the model and methodology; section III reviews the data; section IV

### _wp0842 - section II discusses the model and methodology; section III reviews the data; section IV

### Model and Methodology (Section II)
- Equation estimated: reduced form equation of the nominal interest rate it in small open economies (derived from Edwards and Khan (1985)):
  - log it = δ0 + δ1 i* t + δ2 e_it + δ3 π^e_it + δ4 m_it + δ5 y_it + δ6 d_it + δ7 r_it + δ8 s_it + δ9 lagged_it + ε_it (as presented in equation (1) with the lagged dependent variable accounting for delayed adjustment).
- Theoretical polar cases:
  - Perfect capital mobility: nominal interest rate determined by foreign nominal interest rate (i*), expected depreciation (e), and country-specific risk spread (r).
  - Closed economy: nominal interest rate determined by expected inflation (π^e), real money supply (m, liquidity), and the real interest rate (captured by real GDP growth y and the constant δ0 and the budget deficit d).
- Hypothesis of interest: test null that deficits have no effect, i.e., δ7 = 0.
- Determinants of the deficit-interest rate effect:
  - Degree of capital mobility (more mobility → adjustment via exchange rate rather than domestic real rates).
  - Share of deficits financed domestically vs externally (only domestic financing should affect domestic real rate).
  - Degree of debt neutrality (Ricardian equivalence) — increasing with financial development and planning horizon.
  - Extent to which interest rates are market-determined.
- Identification challenges:
  - Multicollinearity among inflation, depreciation, and budget deficit. Authors avoid forcing coefficients equal to one (would examine nominal rates rather than spreads) and use staggered regressions to assess bias.
  - Potential endogeneity of regressors. Literature approaches: VARs or instrumental variables; both have shortcomings (sensitivity to ordering; weak instruments).
- Estimation approach:
  - Two-step system GMM estimator (Arellano and Bover (1995); Blundell and Bond (1998)) implemented with program by Roodman (2006).
  - Windmeijer (2005) correction applied to reported standard errors.
  - Time dummies included; orthogonal deviations used to maximize sample with gaps.
  - Foreign interest rate and time dummies treated as exogenous; all other variables treated as endogenous.
  - Instrument selection: maximum possible number of up to six lags in almost all cases; instruments collapsed and limited to the number of groups to address Hansen test weakness.
- Modeling choices:
  - Do not use heterogeneous panel estimators because the paper focuses on exploiting cross-country and cross-period heterogeneity via interaction terms; pooled estimators preferred for between-group effects (citing Baltagi et al., 2000).

### Data (Section III)
- Sample: 1970–2006 panel of 60 advanced and emerging economies; U.S. excluded due to no plausibly exogenous international rate.
- Data sources:
  - Most data from IMF International Financial Statistics (IFS).
  - Budget deficit and GDP from IMF World Economic Outlook Database (WEO).
  - Foreign rate: yield of one-year U.S. treasury bills (IFS line 61).
  - Depreciation: percent increase in the U.S. dollar exchange rate (IFS line rf).
  - Risk spread proxied by International Country Risk Guide (ICRG) rating (available from 1984 onward).
  - Inflation: average annual change in consumer price index (IFS line 64).
  - Real money supply: growth rate of ratio of broad money (IFS line 35l) to GDP (nominal GDP used as deflator).
  - Real GDP growth: WEO line NGDP_R.
- Interest rates:
  - Preferred: long rates, but limited availability; thus use short rates with maturity of three to twelve months.
  - Pooling order (to maximize sample): T-bill rates (IFS line 60c), money market rates (IFS line 60b), savings rates (IFS line 60l), lending rates (IFS line 60p); ratio splicing where needed.
- Temporal aggregation:
  - Data averaged over five-year non-overlapping periods (including 2005–06 in last window) to reduce short-run effects and focus on cross-sectional and long-run effects.
- Interaction variables and institutional measures:
  - Domestic financing (IFS line 84a), domestic debt (IFS line 88a), financial depth measured by liquid liabilities of banking system (IFS line 55l or 34+35 if unavailable), all in percent of GDP.
  - Indices of capital account liberalization (Chinn and Ito (2006)) and interest rate liberalization (Abiad et al. (2007)); higher values imply greater liberalization.
- Expectations:
  - Use adaptive expectations equal to current conditions (expectations = current variable) due to limited survey data on expected deficits; rationale includes high correlation between current and expected inflation over one-year horizon and five-year averaging.
  - Hyperinflations: 24 observations removed where five-year average nominal interest rate exceeded 100 percent.
- Table 1 — Descriptive statistics (as reported):
  - Interest rate Mean 10.44 Std. Dev. 8.64 Min. 0.11 Max. 49.99
  - Foreign rate Mean 7.71 Std. Dev. 2.29 Min. 4.73 Max. 12.38
  - Depreciation Mean 9.22 Std. Dev. 28.51 Min. -50.39 Max. 262.56
  - Risk spread Mean -71.23 Std. Dev. 12.16 Min. -93.58 Max. -28.27
  - Inflation Mean 15.41 Std. Dev. 41.01 Min. -0.38 Max. 626.07
  - Money supply Mean 1.68 Std. Dev. 5.12 Min. -14.75 Max. 24.82
  - GDP growth Mean 3.65 Std. Dev. 2.82 Min. -11.24 Max. 12.68
  - Budget deficit Mean 3.11 Std. Dev. 5.24 Min. -16.07 Max. 36.27
  - Domestic financing Mean 2.23 Std. Dev. 2.93 Min. -8.86 Max. 15.05
  - Domestic debt Mean 27.93 Std. Dev. 20.57 Min. 0.06 Max. 109.32
  - Financial depth Mean 53.78 Std. Dev. 55.53 Min. 8.84 Max. 416.82
  - Capital account liberalization Mean 0.36 Std. Dev. 1.55 Min. -1.75 Max. 2.62
  - Interest rate liberalization Mean 1.78 Std. Dev. 1.30 Min. 0.00 Max. 3.00

### Results (Section IV)
A. Baseline
- Marginal effects assessment due to potential multicollinearity (deficits, inflation, depreciation).
- Key baseline findings (Table 2 and text):
  - Initial regression: increase in the deficit by 1 percent of GDP increases the interest rate by 44 basis points (coefficient 0.438).
  - Adding other variables progressively reduces deficit coefficient to 26 basis points (final baseline coefficient 0.259).
  - Comparison: U.S. literature range between 20-60 basis points per 1 percent of GDP (citing Gale and Orszag (2003)).
  - Final baseline specification passes Arellano-Bond and Hansen tests; instruments jointly highly relevant and far smaller in number than groups.
  - Lagged interest rate never significant (five-year averaging removes dynamic effects).
  - Coefficients:
    - Inflation coefficient about 0.41 in some specifications, consistent with gradual incorporation of inflation into yields.
    - Foreign rate coefficient noted as somewhat high.
    - Depreciation significant with expected sign only when inflation not included.
  - Marginal effects on deficit when adding variables:
    - Adding international rate: deficit coefficient unchanged.
    - Adding depreciation: reduces deficit coefficient substantially → some deficit effect operates indirectly via depreciation.
    - Adding risk premium reduces observations; adding inflation reduces deficit coefficient (consistent with deficits being inflationary).
    - Adding money supply increases effect of deficits → suggests accommodative monetary policy dampens deficit effect when not controlled for.
    - GDP growth reduces effect of deficits (likely reflects positive short-run correlation between deficits and GDP growth).

B. Periods and Regions
- Table 3 and text findings:
  - Time dummies for periods show deficits had a negative effect on interest rates during 1985–1994 (combined coefficients e.g., 1985–89 equal to 3.284 – 3.395 = –0.111), and the combined coefficient is highly significantly different from zero by Wald tests.
  - Effect becomes significantly positive from 1995 onwards:
    - 2000–06 effect of 38 basis points.
    - 1995–99 effect of 170 basis points (large and possibly influenced by many crises; low significance).
  - Overall: effect turned from negative during 1985–94 to positive during 1995–2006.
  - Regional/country-group findings:
    - Emerging markets: increase in deficit ratio raises interest rates by 24 basis points.
    - Advanced economies: no significant effect when run separately.
    - Latin America: effect of deficits on interest rates largest (magnitude described as implausibly large, possibly due to crisis episodes).
    - Emerging Asia and Emerging Europe: effect about 60 basis points each.
    - Middle East and Africa: effect highly significantly negative (possibly reflecting financial repression).
  - Conclusion: results in Table 2 for entire sample driven by emerging economies.

C. Interactions
- Baseline with interaction terms coded as “high” and “low” dummies (above/below median) — Table 4 key results:
  - Deficits matter more when:
    - High deficits: effect 52 basis points (Budget deficit*High budget deficit coefficient 0.516).
    - High domestic financing: effect 92 basis points (Budget deficit*High domestic financing coefficient 0.923).
    - High domestic debt: effect 65 basis points when controlling for financial depth (Budget deficit*High domestic debt coefficient 0.652).
  - Domestic vs external financing:
    - Effect much larger and significant only when deficits mostly financed domestically (coefficient 0.923).
  - Financial openness:
    - Low financial openness: deficit effect 67 basis points (Budget deficit*Low capital account openness coefficient 0.673).
    - High financial openness: effect not significant (Budget deficit*High capital account openness coefficient 0.022).
  - Interest rate liberalization:
    - High liberalization: effect 67 basis points (Budget deficit*High interest rate liberalization coefficient 0.674).
    - Low liberalization: effect 45 basis points (Budget deficit*Low interest rate liberalization coefficient 0.449). Difference significant at 13 percent level.
  - Financial depth:
    - Low financial depth: implausibly large effect of 200 basis points (Budget deficit*Low financial depth coefficient 2.108).
    - High financial depth: effect not significant (Budget deficit*High financial depth coefficient -0.111).
    - Financial depth enters also as continuous variable with coefficient -0.017 (significant at 10 percent, showing deeper financial systems reduce deficit effect).
  - Additional notes:
    - Nonlinearities and interaction with debt consistent with literature.
    - Several other interactions tested (exchange rate regime, political instability) yielded no significant effects.
    - Collinearity prevents including all interactions jointly; sample size limits region- or period-specific interaction analysis.

- Selected numeric results from Tables (preserved exactly as reported):
  - Table 2 baseline budget deficit coefficients across specifications: 0.438, 0.438, 0.312, 0.559, 0.484, 0.586, 0.259 (standard errors in parentheses).
  - Table 2 foreign rate coefficients across specifications: 3.395, 1.498, 2.005, 1.747, 1.333, 2.374 (with respective standard errors).
  - Table 3 reported combined baseline coefficient 3.284 and period interaction coefficients: Budget deficit*1985-89 -3.395; *1990-94 -3.400; *1995-99 -1.578; *2000-06 -2.905. (Interpretation in text uses differences yielding negative or positive marginal effects across periods.)
  - Table 4 interaction coefficients (selected): Budget deficit*High budget deficit 0.516 (0.081)***; *High domestic financing 0.923 (0.186)***; *High domestic debt 0.876 (0.135)***; *Low financial depth 2.108 (0.454)***; *Low capital account openness 0.673 (0.069)***; *High interest rate liberalization 0.674 (0.061)***.

### Conclusions (Section V)
- Three main conclusions:
  - For the complete panel, highly significant positive effect of budget deficits on interest rates in the order of about 26 basis points per 1 percent of GDP.
  - Effect varies by country group and time period: larger and more robust in emerging markets and in later periods than in advanced economies and earlier periods.
  - Effect depends on interaction terms and is significant only under certain conditions:
    - When deficits are high.
    - When deficits are mostly domestically financed.
    - When deficits interact with high domestic debt.
    - When financial openness is low.
    - Effect is larger when interest rates are more liberalized and when the domestic financial sector is less developed.
- Policy implications:
  - Fiscal stabilization more effective when initial budget deficit and level of debt are lower, and when financial openness and financial depth are greater, because the effect of deficits on interest rates is smaller under these conditions (implying less crowding out and a greater multiplier).

*Italic: Source — _wp0842 (sections II–V) as provided.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp0842.pdf_