## _wp08137

## Source details

**Canonical URL:** [_wp08137](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08137.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08137.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08137.pdf.json)

---

### Introduction and context
- Latin America’s average fiscal deficit hovered around 5 percent in the last decade, declined beginning in 2003, and reached a surplus in 2006.
- Fiscal developments (unweighted averages for listed countries, in percent of GDP):
  - Public sector revenue: 24.6 (1995-01), 25.0 (2002), 25.5 (2003), 26.0 (2004), 27.1 (2005), 28.5 (2006), 28.9 (2007)
  - Public sector expenditures: 27.6, 29.6, 28.7, 27.9, 27.9, 28.2, 28.7
  - Noninterest: 24.2, 25.2, 25.0, 24.6, 24.9, 25.4, 26.2
  - Interest: 3.4, 4.3, 3.7, 3.3, 3.0, 2.8, 2.5
  - Public sector overall balance: -2.6, -4.6, -3.2, -1.8, -0.7, 0.3, 0.2
  - Public sector primary balance: 0.8, -0.3, 0.5, 1.5, 2.2, 3.1, 2.7
- Central policy question: how much of recent revenue growth is “permanent” versus cyclical or commodity-driven?

### Methodology (overview)
- Main innovations relative to standard structural balance methodology:
  - Distinguish noncommodity and commodity revenues; estimate structural level separately.
  - Adjust noncommodity revenue series for identifiable tax regime changes before cyclical regressions; residuals treated under two alternative views (structural vs. temporary).
- Noncommodity structural revenue approaches:
  - Standard formulation: Rnc_t = A_t Y_t^{ε}. Structural estimate via cyclical correction using potential output and estimated ε (equation (2)).
  - Alternative: adjust revenue series for tax policy changes, test for cointegration between adjusted revenue and GDP; structural revenue = fitted value from regression evaluated at cyclically neutral output (equation (3)).
- Commodity structural revenue:
  - Assume commodity revenues proportional to commodity prices (γ = 1).
  - Compute structural commodity revenue by valuing actual commodity revenues at medium-term expected prices (five-year ahead expectations) using IMF and World Bank forecasts (equation (4)).
- Structural balances:
  - Structural balance = structural revenues − structural expenditures; expenditures assumed structural (no automatic cyclical link).
  - Structural public sector primary balance computed using either conventional (equation (6)) or alternative (equation (7)) treatment of noncommodity residuals, combined with IMF or World Bank commodity price projections.

### Key parameter estimates and statistical tests
- Sample with detailed tax-adjusted analysis limited to eight countries: Argentina, Brazil, Colombia, Costa Rica, Chile, El Salvador, Panama, and Peru.
- Dynamic OLS estimated long-run income elasticities of central government tax revenues (selected estimates; standard errors in parentheses):
  - Argentina: 1.00 (0.01)
  - Brazil: 1.03 (0.01)
  - Chile: 0.96 (0.03)
  - Colombia: 1.10 (0.04)
  - Costa Rica: 1.11 (0.02)
  - El Salvador: 1.36 (0.17)
  - Panama: 0.80 (0.16)
  - Peru: 1.11 (0.07)
- Statistical test outcomes (summary):
  - For most countries there is at least weak evidence of cointegration between adjusted revenues and GDP (ADF and KPSS test columns summarized in Table 2).
  - Exceptions and mixed results: Peru (ADF cannot reject nonstationary residuals; KPSS rejects stationarity), Argentina and Costa Rica (contradictory test indications).

### Results — Structural noncommodity revenues
- Empirical findings:
  - In 5 out of 8 countries, both actual and structural noncommodity revenues trended upward regardless of adjustment approach.
  - Cyclical adjustment (conventional approach) is almost indistinguishable from actual revenue ratios in most cases because estimated income elasticities of revenue are close to 1 (range 0.8–1.36; most clustered between 0.95 and 1.11) and output gaps are modest (order of 0–4 percent above potential).
  - Residual (noncyclical and unaccounted) revenue increases in 2007 are materially large (in the order of 0.6–2.9 percentage points of GDP) in five of the eight countries.
- Interpretation of residuals:
  - For most countries, KPSS tests suggest residuals are stationary; therefore residual revenue buoyancy should be treated as temporary rather than structural.
  - Argentina is a main exception where tests and time pattern suggest the revenue shift is plausibly structural.

### Results — Structural commodity revenues
- Procedure:
  - Construct country-specific export-share-weighted price indices.
  - Compute structural commodity revenues by multiplying actual commodity revenues by ratio of five-year-ahead expected to current prices, using IMF and World Bank projections.
- Main patterns:
  - Structural commodity revenues generally exhibit upward trends similar to actual revenues, but divergence exists.
  - Chile: large gap between actual and structural commodity revenues (order of 4–5 percent of GDP) irrespective of IMF or World Bank forecasts.
  - Results sensitive to forecast source:
    - IMF projections (futures-market-based) imply higher persistence of recent commodity prices → structural commodity revenues rise significantly and remain high.
    - World Bank model-based projections envisage greater medium-term declines (including energy), producing more pessimistic structural revenue estimates.
  - Large oil producers (Ecuador, Mexico, Trinidad and Tobago, Venezuela):
    - IMF forecasts imply current revenue levels can be sustained.
    - World Bank projections imply structural commodity revenues are below actual by:
      - Venezuela and Trinidad and Tobago: 3–4 percent of GDP
      - Ecuador and Mexico: around 2 percent of GDP

### Results — Structural primary balances
- Construction:
  - Combine structural revenue estimates (both noncommodity and commodity alternatives) with noninterest expenditure data to obtain structural public sector primary balances.
- Working assumptions:
  - For commodity revenues accruing to public enterprises, focus on overall public sector structural balances; treat noncentral-government noncommodity revenues as conventionally adjusted (noncyclical changes structural).
  - For countries without detailed noncommodity analysis, use average income elasticity ≈ 1.1 for cyclical adjustment.
- Main findings:
  - Structural primary balances improved significantly relative to early 2000s; many countries showed structural primary surpluses in 2007 though peaks occurred in 2006.
  - Cross-country dispersion in 2007 structural primary balances large: e.g., Chile has a large public sector primary surplus; Venezuela shows a substantial deficit.
  - Ranges of structural balance estimates can be wide due to alternative commodity price projections and alternative treatments of noncommodity residuals.
  - Statistical criteria favor the more pessimistic approach (treat residuals as temporary) except for Argentina.
- Risk factors:
  - Uncertainty from commodity price forecasts and residual attribution leads to significant margin of error in structural balance estimates.
  - Rapid expenditure growth in recent years (real growth rates in the last two years “namely, 8-10 percent on average in real terms”) threatens fiscal gains.

### Output gap estimates (methodology and 2007 estimates)
- Method:
  - Output gaps estimated by statistical filtering of annual (log) real GDP series for each country, 1970–2010, using IMF World Economic Outlook projections for 2008–2010.
  - Filters compared: HP(a) with lambda = 6.25; HP(b) with lambda = 100; Baxter-King (BK); Ouliaris-Corbae (OC).
  - Main text uses HP(a) (smoothing parameter = 6.25); robustness checks indicate conclusions are not driven by filter choice.
- Main findings (2007 output gaps, in percent of potential GDP):
  - Most countries had positive, albeit small, output gaps in 2007 (output above potential), in the range of 0–3 percent.
  - Standard deviation across filters is between about 0.2 and 1 percent of GDP.
- Select filter-based 2007 output gap estimates (HP(a); HP(b); BK; OC; Mean; Stan. Dev.):
  - Jamaica: -0.4 0.0 -0.2 0.8 0.04 0.46
  - Trinidad & Tobago: 0.9 3.3 1.2 0.9 1.57 1.01
  - Costa Rica: 1.2 1.4 1.0 1.0 1.16 0.18
  - El Salvador: 0.2 -0.6 0.0 -0.9 -0.32 0.47
  - Guatemala: 0.4 0.1 0.2 0.9 0.42 0.33
  - Honduras: 1.3 2.4 1.1 0.2 1.27 0.78
  - Nicaragua: -0.3 -0.7 -0.5 -0.8 -0.56 0.21
  - Panama: 2.1 0.9 0.4 1.9 1.33 0.69
  - Argentina: 2.7 4.1 1.7 2.2 2.69 0.89
  - Bolivia: -0.4 -1.0 -0.7 -0.3 -0.62 0.28
  - Brazil: 0.4 0.6 0.2 0.5 0.44 0.14
  - Chile: 0.3 -0.2 0.0 -0.3 -0.03 0.23
  - Colombia: 1.7 1.9 1.0 0.8 1.33 0.45
  - Ecuador: -0.2 1.0 -0.3 0.2 0.18 0.53
  - Mexico: 0.6 0.4 0.5 3.0 1.13 1.08
  - Paraguay: 1.2 1.3 0.8 1.0 1.04 0.19
  - Peru: 1.5 1.8 0.9 -0.8 0.87 1.03
  - Uruguay: 2.2 3.3 1.4 0.6 1.86 1.01
  - Venezuela: 3.5 5.8 2.3 2.1 3.43 1.46

### Conclusions and policy recommendations
- Overall assessment:
  - Structural fiscal balances in Latin America improved between 2002 and 2007, though less than headline balances; improvements were significant in many cases.
  - Much of the headline improvement stems from commodity price increases and changes in taxation/tax administration rather than cyclical factors.
  - Structural balances remain vulnerable to commodity price reversals and to unclarified residual revenue buoyancy in some countries.
- Policy implications and recommendations:
  - Curtail expenditure growth to avoid reverting to primary deficits and renewed debt problems.
  - Consider institutional or structural reforms to improve expenditure control and fiscal resilience, such as:
    - Reducing budgetary rigidities,
    - Increasing expenditure efficiency and flexibility,
    - Strengthening public financial management systems.
- Caveats emphasized:
  - Some portion of revenue improvement may reflect speculative elements in commodity price increases (bubble risk).
  - Structural balance estimates do not fully capture parameter uncertainty and are sensitive to alternative commodity price projections (IMF vs. World Bank) and to the interpretation of noncommodity revenue residuals.

*Source: _wp08137 - Appendix Tables*

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

### _wp08137 - References................................................................................................................................20

### Tables
- 1.   Latin America: Fiscal Developments...................................................................................3
- 2.   Long-Run Income Elasticity of Central Government Tax Revenues ................................10

### Figures
- 1.    Actual and Structural Noncommodity Revenues..............................................................12
- 2.    Actual and Structural Commodity Revenues....................................................................15
- 3.    Actual and Estimated Structural Primary Balances ..........................................................17

### Appendices
- 1.    Revenue Data and Changes in Tax Structure ...................................................................22
- 2.    Output Gap Estimates .......................................................................................................28
- 3.    Commodity Price Indices and Projections ........................................................................29

*Source: _wp08137 - References*

### Appendix Tables

### Appendix Tables

### Introduction and context
- Latin America’s average fiscal deficit hovered around 5 percent in the last decade, declined beginning in 2003, and reached a surplus in 2006.
- Fiscal developments (unweighted averages for listed countries, in percent of GDP):  
  - Public sector revenue: 24.6 (1995-01), 25.0 (2002), 25.5 (2003), 26.0 (2004), 27.1 (2005), 28.5 (2006), 28.9 (2007)  
  - Public sector expenditures: 27.6, 29.6, 28.7, 27.9, 27.9, 28.2, 28.7  
  - Noninterest: 24.2, 25.2, 25.0, 24.6, 24.9, 25.4, 26.2  
  - Interest: 3.4, 4.3, 3.7, 3.3, 3.0, 2.8, 2.5  
  - Public sector overall balance: -2.6, -4.6, -3.2, -1.8, -0.7, 0.3, 0.2  
  - Public sector primary balance: 0.8, -0.3, 0.5, 1.5, 2.2, 3.1, 2.7
- Key question: how much of recent revenue growth is “permanent” versus cyclical or commodity-driven?

### Methodology (overview)
- Main innovations relative to standard structural balance methodology:
  - Distinguish noncommodity and commodity revenues; estimate structural level separately.
  - Adjust noncommodity revenue series for identifiable tax regime changes before cyclical regressions; residuals treated under two alternative views (structural vs. temporary).
- Noncommodity structural revenue:
  - Standard formulation: Rnc_t = A_t Y_t^{ε}. Structural estimate via cyclical correction using potential output and estimated ε (equation (2)).
  - Alternative: adjust revenue series for tax policy changes, test for cointegration between adjusted revenue and GDP; structural revenue = fitted value from regression evaluated at cyclically neutral output (equation (3)).
- Commodity structural revenue:
  - Assume commodity revenues proportional to commodity prices (γ = 1) and compute structural commodity revenue by valuing actual commodity revenues at medium-term expected prices (five-year ahead expectations), using two forecast sources: IMF and World Bank (equation (4)).
- Structural balances:
  - Structural balance = structural revenues − structural expenditures; assume expenditures are structural (no automatic cyclical link).
  - Structural public sector primary balance computed using either the conventional (equation (6)) or alternative (equation (7)) treatment of noncommodity residuals, combined with IMF or World Bank commodity price projections.

### Key parameter estimates and statistical tests
- Sample with detailed tax-adjusted analysis limited to eight countries: Argentina, Brazil, Colombia, Costa Rica, Chile, El Salvador, Panama, and Peru.
- Dynamic OLS estimated long-run income elasticities of central government tax revenues (selected estimates from Table 2; standard errors in parentheses):
  - Argentina: 1.00 (0.01)
  - Brazil: 1.03 (0.01)
  - Chile: 0.96 (0.03)
  - Colombia: 1.10 (0.04)
  - Costa Rica: 1.11 (0.02)
  - El Salvador: 1.36 (0.17)
  - Panama: 0.80 (0.16)
  - Peru: 1.11 (0.07)
- Statistical test outcomes (summary):
  - For most countries there is at least weak evidence of cointegration between adjusted revenues and GDP (ADF and KPSS test columns summarized in Table 2).
  - Exceptions and mixed results: Peru (ADF cannot reject nonstationary residuals; KPSS rejects stationarity), Argentina and Costa Rica (contradictory test indications).

### Results — Structural noncommodity revenues
- Empirical findings:
  - In 5 out of 8 countries, both actual and structural noncommodity revenues trended upward regardless of adjustment approach.
  - Cyclical adjustment (conventional approach) is almost indistinguishable from actual revenue ratios in most cases because estimated income elasticities of revenue are close to 1 (range 0.8–1.36; most clustered between 0.95 and 1.11) and output gaps are modest (order of 0–4 percent above potential).
  - Residual (noncyclical and unaccounted) revenue increases in 2007 are materially large (in the order of 0.6–2.9 percentage points of GDP) in five of the eight countries.
- Interpretation of residuals:
  - For most countries, KPSS tests suggest residuals are stationary; therefore residual revenue buoyancy should be treated as temporary rather than structural.
  - Argentina is a main exception where tests and time pattern suggest the revenue shift is plausibly structural.

### Results — Structural commodity revenues
- Procedure: construct country-specific export-share-weighted price indices; compute structural commodity revenues by multiplying actual commodity revenues by ratio of five-year-ahead expected to current prices, using IMF and World Bank projections.
- Main patterns:
  - Structural commodity revenues generally exhibit upward trends similar to actual revenues, but divergence exists.
  - Chile: large gap between actual and structural commodity revenues (order of 4–5 percent of GDP) irrespective of IMF or World Bank forecasts.
  - Results are sensitive to choice of forecast source:
    - IMF projections (futures-market-based) imply higher persistence of recent commodity prices → structural commodity revenues rise significantly and remain high.
    - World Bank model-based projections envisage greater medium-term declines (including energy), producing more pessimistic structural revenue estimates.
  - Large oil producers (Ecuador, Mexico, Trinidad and Tobago, Venezuela): IMF forecasts imply current revenue levels can be sustained; World Bank projections imply structural commodity revenues are below actual by:
    - Venezuela and Trinidad and Tobago: 3–4 percent of GDP
    - Ecuador and Mexico: around 2 percent of GDP

### Results — Structural primary balances
- Construction: combine structural revenue estimates (both noncommodity and commodity alternatives) with noninterest expenditure data to obtain structural public sector primary balances.
- Two additional working assumptions:
  - For commodity revenues accruing to public enterprises, focus on overall public sector structural balances; treat noncentral-government noncommodity revenues as conventionally adjusted (noncyclical changes structural).
  - For countries without detailed noncommodity analysis, use average income elasticity ≈ 1.1 for cyclical adjustment.
- Main findings (Figure 3 summary):
  - Structural primary balances improved significantly relative to early 2000s; many countries showed structural primary surpluses in 2007 though peaks occurred in 2006.
  - Cross-country dispersion in 2007 structural primary balances large: e.g., Chile has a large public sector primary surplus; Venezuela shows a substantial deficit.
  - Ranges of structural balance estimates can be wide due to alternative commodity price projections and alternative treatments of noncommodity residuals.
  - Statistical criteria favor the more pessimistic approach (treat residuals as temporary) except for Argentina.
- Risk factors:
  - Uncertainty from commodity price forecasts and residual attribution leads to significant margin of error in structural balance estimates.
  - Rapid expenditure growth in recent years (real growth rates in the last two years “namely, 8-10 percent on average in real terms”) threatens fiscal gains.

### Conclusions and policy recommendations
- Overall assessment:
  - Structural fiscal balances in Latin America improved between 2002 and 2007, though less than headline balances; improvements were significant in many cases.
  - Much of the headline improvement stems from commodity price increases and changes in taxation/tax administration rather than cyclical factors.
  - Structural balances remain vulnerable to commodity price reversals and to unclarified residual revenue buoyancy in some countries.
- Policy implications and recommendations:
  - Curtail expenditure growth to avoid reverting to primary deficits and renewed debt problems.
  - Consider institutional or structural reforms to improve expenditure control and fiscal resilience, such as:
    - Reducing budgetary rigidities,
    - Increasing expenditure efficiency and flexibility,
    - Strengthening public financial management systems.
- Caveats emphasized:
  - Some portion of revenue improvement may reflect speculative elements in commodity price increases (bubble risk).
  - Structural balance estimates do not fully capture parameter uncertainty and are sensitive to alternative commodity price projections (IMF vs. World Bank) and to the interpretation of noncommodity revenue residuals.

*Source: _wp08137 - Appendix Tables*

### REFERENCES

### _wp08137 - REFERENCES

### References
- Alier, Max, “Measuring Budget Rigidities in Latin America,” IMF Working Paper (unpublished; Washington: International Monetary Fund).
- Alberola, Enrique, and José Manuel Montero, 2007, “Debt Sustainability and Procyclical Fiscal Policies in Latin America,” Economía, Vol. 7, No. 1 (Fall 2006 Issue), pp. 157–84.
- Baxter, Marianne, and Robert King, 1999, “Measuring Business Cycles: Approximate Band-Pass Filters for Economic Time Series,” Review of Economics and Statistics, Vol. 81, No. 4, pp. 575–93.
- Chalk, Nigel, 2002, “Structural Balances and All That: Which Indicators to Use in Assessing Fiscal Policy,” IMF Working Paper 02/101 (Washington: International Monetary Fund).
- Coelho, Isaias, Liam Ebril, and Victoria Summers, “Bank Debit Taxes in Latin America: An Analysis of Recent Trends,” IMF Working Paper WP/01/67 (Washington: International Monetary Fund).
- Contraloría General de la República de Costa Rica, 2002, El Sistema Tributario Costarricense: Contribuciones al Debate Nacional (San José: Costa Rica).
- Corbae, Dean, and Sam Ouliaris, 2002, “Band Spectral Regression with Trending Data,” Econometrica, Vol. 70, No. 3, pp. 1067–1109.
- ———, 2006, “Extracting Cycles from Nonstationary Data,” Econometric Theory and Practice: Frontiers of Analysis and Applied Research (Cambridge and New York: Cambridge University Press), pp. 167–77.
- Cubero, Rodrigo, and Rhiannon Sowerbutts, “Actual vs. Structural Fiscal Balances: Evidence from Costa Rica,” IMF Working Paper (unpublished; Washington: International Monetary Fund).
- De Mello, Luiz, and Diego Moccero, 2006, “Brazil’s Fiscal Stance During 1995–2005: The Effect of Indebtedness on Fiscal Policy over the Business Cycle,” Economics Department Working Paper No. 485 (Paris, France: OECD).
- Duarte Lledo, Victor, 2005, “Tax Systems under Fiscal Adjustment: a Dynamic CGE Analysis of the Brazilian Tax Reform,” IMF Working Paper 05/142 (Washington: International Monetary Fund).
- Hagemann, Robert, 1999, “The Structural Budget Balance: The IMF’s Methodology,” IMF Working Paper 99/95 (Washington: International Monetary Fund).
- Hodrick, Robert J., and Edward C. Prescott, 1997, “Postwar U.S. Business Cycles: An Empirical Investigation,” Journal of Money, Credit and Banking, Vol. 29 (February), pp. 1–16.
- Izquierdo, Alejandro, Pablo Ottonello, and Ernesto Talvi, “If Latin America Were Chile: A Comment on Structural Fiscal Balances and Public Debt” (unpublished; IADB Research Department and CERES).
- Inter-American Development Bank, 2008, “All That Glitters May Not Be Gold: Assessing Latin America’s Recent Macroeconomic Performance,” (Washington: Inter-American Development Bank).
- International Monetary Fund, 2006, Regional Economic Outlook: Western Hemisphere, November 2006 (Washington).
- ———, 2007, Regional Economic Outlook: Western Hemisphere, November 2007 (Washington).
- Lozano, Ignacio, and Jorge Toro, 2007, “Fiscal Policy Throughout the Cycle: the Colombian Experience,” Borradores de Economía, No. 34 (Colombia: Banco de la Republica).
- Marcel, Mario C., Marcelo Tokman, Rodrigo Valdés, and Paula Benavides, 2001, “Balance Estructural del Gobierno Central Metodología y Estimaciones para Chile: 1987–2000,” Estudios de Finanzas Publicos, September (Santiago de Chile: Direccion de Presupuestos, Ministerio de Hacienda, Government of Chile).
- Martner, Ricardo, and Varinia Tromben, 2004, “Tax Reforms and Fiscal Stabilization in Latin American Countries,” Serie Gestión Pública, No. 45, CEPAL, Santiago, Chile.
- Ravn, Morten O., and Harald Uhlig, 2002, “On Adjusting the Hodrick-Prescott Filter for the Frequency of Observations,” Review of Economics and Statistics, Vol. 84, No. 2, pp. 371–76.
- Rico Torres, Cecilia, 2006, “Recursos Fiscales Para el Restabecimiento del Orden Publico en Colombia: 1991–2006,” Direccion de Impuestos y Aduanas Nacionales, Oficina de Estudios Economicos, Documento Web 013.
- Rodríguez Arosemena, Nicolás, 2007, “Incidencia de la Política Fiscal en Panamá” (Washington: Banco Interamericano de Desarrollo).
- Swiston, Andrew, Martin Mühleisen, and Koshy Mathai, “U.S. Revenue Surprises: Are Happy Days Here to Stay?” IMF Working Paper 07/143 (Washington: International Monetary Fund).
- Werneck, Rogerio, 2004, “An Evaluation of the 2003 Tax Reform Effort in Brazil,” Discussion Paper No. 488, PUC-Rio.

### Appendix — 1. Revenue Data and Changes in Tax Structure

- General:
  - Country-level tax revenue series are reported at specified government levels and sample spans as indicated for each country.
  - Adjustments to tax revenue series were made to account for commodity-related revenues where indicated, introduction or removal of specific taxes, changes in tax rates, tax base broadening or narrowing, and notable tax administration reforms. Step dummies were used when nominal impacts were infeasible to estimate.

- Argentina
  - Data: Tax revenues at the federal government level, 1980–2007.
  - Commodity proxy: Export taxes used as a proxy for commodity-related revenues.
  - Tax structure changes:
    - Financial transactions tax introduced 1988–1992; tax revenues series adjusted by nominal yield.
    - December 1989 broadening of VAT base; modeled with a step dummy starting in 1990.
    - VAT rate changes: 16 percent in 1980, 18 percent in 1991, 21 percent in 1995; nominal series adjusted to account for rate changes.
    - Financial transactions taxes re-introduced in 2001; revenue series adjusted by nominal yield.

- Brazil
  - Data: Tax revenue at the federal government level, 1992–2007.
  - Commodity-related revenue: estimated relatively small at the federal government level (about 2 percent of GDP or less over the sample period); no adjustment made for commodity-related revenues.
  - Tax structure changes:
    - Financial transactions tax (CPMF) introduced in February 1997; nominal revenue yield used to adjust series.
    - Fiscal package announced November 10, 1997; revenue measures included a personal income tax surcharge of 10 percent for all taxpayers in the upper 25 percent tax bracket, limits on deductions, a 25 percent reduction in regional fiscal incentives for 1998, an increase in IPI for specified goods, and temporary increases in prices of petroleum derivatives and alcohol. Controlled by a step dummy beginning in 1998 due to unavailable nominal impact estimates.
    - Major tax reform in 2003 with main impact beginning in 2004: conversion of Cofins from a turnover tax into a non-cumulative tax; CPMF extended. Initial ex-ante impact estimates for 2004 were released by the Ministry of Finance; nominal revenue series adjusted by the actual ex-post impact observed for that revenue item.

- Chile
  - Data: Tax revenue at the central government level, 1993–2007.
  - Commodity revenues: accrue to state enterprises and the central government; central government tax revenue series adjusted by available series on income and VAT payments by private mining companies.
  - Tax structure changes:
    - Major reforms occurred before the sample period: early 1980s VAT fluctuated between 20 and 16 percent; 1984 income tax lowering; 1990 VAT raised from 16 to 18 percent and corporate tax adjustments for 1991–1993 from 10 to 15 percent.
    - Since 1993, only changes were in income tax rates: personal income tax raised from 15 percent to 16 percent and 16.5 percent in 2002 and 2003; as of January 1, 2004, rate raised to and has remained at 17 percent.
    - Maximum marginal tax rate for Impuesto Global Complementario: fell from 45 percent in 2001, to 43 percent in January 2002, and 40 percent in 2003.
    - Given detailed information on the base, a nominal impact was calculated to adjust the tax revenue series.

- Colombia
  - Data: Tax revenue at the central government level, 1993–2007.
  - Commodity revenues: accrue to state enterprises and do not affect central government revenues substantively; no adjustment made.
  - Tax structure changes:
    - Financial transactions tax introduced in 1999; tax revenue series adjusted to exclude nominal impact.
    - Wealth tax introduced August 2002 (Decree No. 1838), assessed on net assets at a rate of 1.2 percent for payers of income tax; tax revenue series adjusted to exclude nominal impact. Estimates presented in Rico Torres (2006) differ from collections reported to IMF staff; IMF staff reported series used for adjustments.
    - VAT rate increased from 15 to 16 percent in January 2001, together with additional tax measures; adjustments: nominal adjustment of tax revenue series to account for rate change (keeping VAT structure unchanged at 15 percent) and a step dummy for 2001 to capture potential additional tax structure changes.
    - Broadening of VAT base in 2003; controlled with a step dummy for the 2003–2007 period.

- Costa Rica
  - Data: Tax revenue at the central government level, 1991–2007.
  - Commodity-related revenues: economically insignificant.
  - Tax structure changes:
    - 1995: 1 percent tax on companies’ assets created (creditable to corporate income tax); declared unconstitutional and repealed in 1999; tax revenue series adjusted by nominal yield.
    - Sales tax rate changes in the 1990s: increased from 10 to 13 percent in 1991, reduced by 1 pp each year to 10 percent in 1994, increased to 15 percent for 18 months in September 1995, then returned to 13 percent; sales tax series adjusted for estimated nominal effects.
    - Tax administration measures phased in starting July 2005: automated customs reporting system (TICA) phased in; modeled as a step dummy starting in 2005 due to notable revenue increases at customs points.

- El Salvador
  - Data: Tax revenue at the central government level, 1992–2007.
  - Commodity-related revenues: economically insignificant.
  - Tax structure changes:
    - November 1994 and July 1995: income tax and VAT changes. November 1994 modifications increased withholding rates on professionals from 2 to 10 percent, on corporations and individually owned enterprises from 1.0 to 1.5 percent, and on coffee producers (scale up to 15 percent according to international price of coffee). VAT rate raised from 10 to 15 percent in July 1995. Modeled with a step dummy starting in 1995.
    - Second tax reform package approved Q3 1999 and H1 2000: partial package in September 1999 modified dividend taxation, limited income tax deductions, discontinued VAT credits for certain expenses, and established new anti-evasion procedures. Remaining VAT and income base-broadening measures approved mid-2000. Overall impact estimated at 0.8 percent of GDP; modeled with a step dummy starting in 2000.
    - 2005 reforms: broadened income and VAT bases, raised excise taxes, reinforced tax office. Authorities’ preliminary estimates attribute about 0.5 percentage points of the increase in tax revenue in 2005 to the reform, with a further 0.1 percent of GDP in 2006. Tax revenue series adjusted to reflect nominal impacts.

- Panama
  - Data: Tax revenue at the central government level, 1992–2007. Commodity-related revenues economically insignificant. Tax revenues exclude revenue from canal operations.
  - Tax structure changes:
    - June 1995 Tax Incentives Harmonization Law: phased out/scaled back tax breaks and subsidies, reduced maximum corporate income tax from 34 to 30 percent, increased income tax for exporters in Colon Free Zone from 2.5–8.5 to 15 percent, generalized a tax credit up to 25 percent of income tax liabilities for investment until 2000, extended a preferential import tariff of 3 percent on raw materials to the whole manufacturing sector, and reduced home mortgage subsidies via better targeting. Modeled with a step dummy starting in 1996.
    - Rapid trade liberalization starting 1997 with lowering of import tariffs and phasing out of export taxes; effects on trade tax revenue notable starting in 2000. Trade tax revenue series was excluded from the overall tax revenue series to account for this.
    - December 2002 reforms: increased annual exemption under personal income tax, introduced a minimum tax on banks’ income, phased out some fiscal incentives, scheduled corporate income tax reductions from 30 percent to 29 percent in 2005 and 27 percent in 2007, widened VAT base to include services, extended consumption tax to luxury goods, and provided more autonomy for tax administration. Modeled with a step dummy starting in 2003 due to lack of nominal yield estimates.
    - February 2005 measures: postponement of scheduled corporate income tax reduction, maximum income tax rate on individuals reduced from 30 percent to 27 percent, introduction of minimum income tax on companies and individuals, elimination of tax incentives for non-traditional exports, changes to CFZ tax regime, increased sanctions for noncompliance, and changes to petroleum taxation. Nominal estimates available and used to adjust series.
    - 2006: Tax revenue series adjusted for one-off revenue associated with purchase of Banistmo by HSBC.

- Peru
  - Data: Tax revenue at the central government level, 1992–2007.
  - Commodity-related revenue: mining and petroleum amounting to about 5 ¼ percent of GDP in 2007 were removed from the tax revenue series.
  - Tax structure changes:
    - 1994: reduction in tax evasion and broadening of tax base led to substantial increase in central government tax revenue; modeled with a step dummy starting in 1994 due to absence of nominal yield estimate.
    - 1997 onward: several tax rate reductions starting in 1997:
      - Payroll tax rate reduced from 9 to 5 percent of gross wages in September 1997.
      - Excise rates on petroleum products reduced in March and August 1997; excise rate on fuel oil eliminated in July 1998.
      - Corporate minimum income tax rate lowered from 2 percent to 0.5 percent in May 1997, and further to 0.2 percent in January 1999.
      - Import tariffs lowered in April 1997, reducing average weighted tariff rate from 15.2 to 13.5 percent.
      - Fiscal cost of 1997–1999 measures roughly estimated at 1–1.5 percentage points of GDP, only partially compensated by tax administration improvements.
    - Effects of 1997–1999 rate reductions modeled with a step dummy starting in 1997.
    - December 2002: several income tax exemptions allowed to expire; modeled with a step dummy starting in 2003.
    - 2004: financial transactions tax (FTT) introduced on credit and debit operations with exemptions on transfers between accounts of same account holder; tax rate initially set at 0.1%, reduced to 0.08 percent in 2005. Tax revenue series adjusted for the nominal yield of the FTT.

*Content derived from _wp08137 - REFERENCES*

### 2. Output Gap Estimates

### 2. Output Gap Estimates

### Methodology
- Output gaps estimated by statistical filtering of annual (log) real GDP series for each country, ranging from 1970 until 2010 (source: IMF World Economic Outlook database).
- The last three values of each series are projections reflecting IMF staff’s views of GDP growth convergence to potential growth in each country.
- Annual-data filtering used due to lack of reliable quarterly GDP data and/or lack of capital stock data in many countries.
- To address the “end-point problem”, series were filtered over the full 1970–2010 span.

### Filters compared
- Hodrick-Prescott filter with smoothing parameter equal to 6.25 (HP(a)), as recommended by Ravner and Uhlig (2002).
- Hodrick-Prescott filter with smoothing parameter equal to 100 (HP(b)) (Hodrick and Prescott, 1997).
- Baxter-King filter (BK) (Baxter and King, 1999).
- Frequency domain filter due to Corbae and Ouliaris (OC) (2002, 2006).
- Main text uses HP(a) (smoothing parameter = 6.25) for familiarity; robustness checks indicate conclusions are not driven by choice of filter.

### Main findings (2007 output gaps, in percent of potential GDP)
- Most countries analyzed had positive, albeit small, output gaps in 2007 (output above potential), in the range of 0–3 percent.
- The standard deviation across filters is between about 0.2 and 1 percent of GDP, implying robustness of conclusions to filter choice.
- Output gaps defined as actual GDP in excess of potential, in percent. Estimates based on annual data, 1970-2010, using IMF World Economic Outlook projections for 2008-2010.

### Filter-based output gap estimates for 2007 (HP(a); HP(b); BK; OC; Mean; Stan. Dev.)
- Caribbean — Jamaica: -0.4 0.0 -0.2 0.8 0.04 0.46
- Caribbean — Trinidad & Tobago: 0.9 3.3 1.2 0.9 1.57 1.01

- Central America — Costa Rica: 1.2 1.4 1.0 1.0 1.16 0.18
- Central America — El Salvador: 0.2 -0.6 0.0 -0.9 -0.32 0.47
- Central America — Guatemala: 0.4 0.1 0.2 0.9 0.42 0.33
- Central America — Honduras: 1.3 2.4 1.1 0.2 1.27 0.78
- Central America — Nicaragua: -0.3 -0.7 -0.5 -0.8 -0.56 0.21
- Central America — Panama: 2.1 0.9 0.4 1.9 1.33 0.69

- South America and Mexico — Argentina: 2.7 4.1 1.7 2.2 2.69 0.89
- South America and Mexico — Bolivia: -0.4 -1.0 -0.7 -0.3 -0.62 0.28
- South America and Mexico — Brazil: 0.4 0.6 0.2 0.5 0.44 0.14
- South America and Mexico — Chile: 0.3 -0.2 0.0 -0.3 -0.03 0.23
- South America and Mexico — Colombia: 1.7 1.9 1.0 0.8 1.33 0.45
- South America and Mexico — Ecuador: -0.2 1.0 -0.3 0.2 0.18 0.53
- South America and Mexico — Mexico: 0.6 0.4 0.5 3.0 1.13 1.08
- South America and Mexico — Paraguay: 1.2 1.3 0.8 1.0 1.04 0.19
- South America and Mexico — Peru: 1.5 1.8 0.9 -0.8 0.87 1.03
- South America and Mexico — Uruguay: 2.2 3.3 1.4 0.6 1.86 1.01
- South America and Mexico — Venezuela: 3.5 5.8 2.3 2.1 3.43 1.46

### Notes
- Abbreviations: HP(a) = Hodrick-Prescott with lambda = 6.25; HP(b) = Hodrick-Prescott with lambda = 100; BK = Baxter-King; OC = Ouliaris-Corbae.
- Estimates use annual data, 1970-2010, with IMF World Economic Outlook projections for 2008-2010.

*Source: _wp08137 - 2. Output Gap Estimates*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08137.pdf_