Commodity Prices and Fiscal Policy in Latin America and the Caribbean EMILY SINNOTT

Commodity Prices and Fiscal Policy in Latin America and the Caribbean EMILY SINNOTT

Context Examine recent fiscal dependency on commodities How dependent is the region vs. other regions? Evolution of commodity revenues (hydrocarbon and mining) Tax regimes Focus on commodity production Focus on revenues (hydrocarbon and mining) Consumption dependence also critical to fiscal story (role of subsidies in LCR, see Olade, 2007) Link between reliance on commodity revenues and procyclical fiscal policy in LCR Compliment case studies

Content 1. Dependency of fiscal revenues on commodity production in LCR 2. Evolution over time of taxation regimes for commodity-dependent countries 3. Fiscal cycles and commodity prices 4. Expenditure and revenue response during recent boom 5. Preliminary results on fiscal response to commodity price movements 6. Future work

1. Commodity dependence Following IMF (2007): Countries are considered dependent on mineral resources on the basis of the following criteria i. Average share of mineral fiscal revenues in total fiscal revenues is at least 25 percent on average over 2000-2005, or ii. Average share of mineral export proceeds in total export proceeds is at least 25 percent on average over 2000-2005.

Six hydrocarbon-dependent countries in region Country/region Average Annual Hydrocarbon Revenues 2000-05 In percent of total fiscal revenue 1/ In percent of GDP Average Annual Hydrocarbon Exports (Goods) 2000-05 In percent of total exports (Goods) In percent of GDP LCR 29.2 8.0 42.7 13.1 Bolivia 20.9 5.6 23.0 5.0 Colombia 10.0 3.0 26.7 4.4 Ecuador 26.0 6.6 46.9 11.8 M exico 33.3 7.5 17.2 3.0 Trinidad and Tobago 36.4 9.3 59.9 28.4 Venezuela 48.8 15.8 82.5 25.8

Two mining-dependent countries in region Country/ Region Mineral Average Annual M ineral Revenues 2000-05 In percent of total fiscal revenue In percent of GDP Average Annual M ineral Exports 2000-05 In percent of total exports (goods) In percent of GDP LCR 6.4 1.9 45.0 9.9 Chile Copper 9.4 2.2 39.1 11.7 Peru Gold, copper, silver 3.3 1.5 50.8 8.1

What can we say about dependence in the region The natural resource a country produces matters. Oil is associated with far greater fiscal revenues than gas or metals activities. Profitability of gas projects is less robust to changes in production cost parameters compared to oil (Kellas, 2008). Mining tax pressure worldwide tends to be lower relative to that for the oil sector reflecting lower economic rents (Baunsgaard, 2001) Some evidence that mining tax regimes are lower still for Latin America (Otto, 2004). Large diversity in dependency among hydrocarbon producers in the region. Venezuela and Trinidad and Tobago high fiscal, output and export dependence Mexico high fiscal, lower exports and low output dependence Colombia low fiscal and output dependence, high but high export dependence. Mineral producers (Chile, Peru) high export dependence, but lower output and low fiscal dependence.

Fiscal dependence: measurement problems Quasi-fiscal activities of SOEs and government institutions not included (provision of social services and other public good). Employment over and above the norm in hydrocarbon SOEs not reflected. Cost to SOE of using its leverage to borrow for the government. Use of production-sharing contracts or export bans/taxes to provide a subsidy for domestic consumption of natural resources excluded. Mineral/hydrocarbon levels NB at a subnational but not at a national level.

Regional comparison hydrocarbon dependency and abundance measures (annual average, 2000-2005)

Since the late 1990s hydrocarbon/mining revenues have grown substantially in the region both as a share of GDP 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Change 1998-2008 Natural resource revenues as a share of GDP (in percent) Average 3.7 4.4 6.0 5.1 5.1 5.9 6.4 8.6 10.5 9.7 10.3 6.7 Bolivia 5.4 5.3 5.1 5.1 4.6 4.6 5.0 9.0 11.6 11.3.. 5.9 Chile 0.4 0.4 0.9 0.5 0.5 0.9 3.0 5.1 8.8 8.6 6.0 5.6 Ecuador 3.9 6.3 9.2 6.4 5.6 5.8 6.5 5.9 7.8 7.2 16.6 12.7 M exico 6.1 6.2 7.0 6.7 6.5 7.1 7.4 7.9 8.3 7.9 8.7 2.7 Peru 1.4 1.3 1.4 1.5 1.7 1.8 1.7 2.2 3.3 3.9 2.8 1.3 Trinidad and Tobago 3.0 4.7 8.6 6.4 6.5 9.7 9.8 16.6 17.5 14.2 15.6 12.6 Venezuela 5.8 6.7 10.0 9.4 10.5 11.6 11.2 13.4 15.8 14.6 12.0 6.2

and as a share of overall revenues 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Change 1998-2008 Natural resource revenues as a share of total revenues (in percent) Average 16.5 20.0 26.5 24.0 21.5 25.2 27.2 31.9 34.5 37.8 37.0 21.4 Bolivia 28.3 26.4 26.0 26.0 20.9 20.0 20.5 34.4 29.8 43.5.. 15.2 Chile 1.8 1.6 3.9 2.2 2.0 3.9 13.7 15.5 22.1 29.2 20.9 19.1 Ecuador 6.6 21.3 30.5 19.9 15.4 15.9 20.0 23.7 28.4 25.0 32.8 26.3 M exico 29.8 29.9 33.1 30.4 29.5 33.3 35.9 37.4 37.9 35.6 36.8 7.0 Peru 7.8 7.6 7.8 8.8 10.6 10.5 10.2 12.1 17.3 19.4 15.3 7.5 Trinidad and Tobago 6.8 11.5 27.2 34.9 23.6 38.5 41.1 52.7 61.9 56.6 57.8 51.0 Venezuela 34.5 41.6 57.1 45.7 48.5 54.5 48.6 47.4 44.0 55.4 58.3 23.8

2. Features of commodity tax regimes in the region Hydrocarbon/mining tax regime a combination of royalties, corporate profit taxes, dividend withholding taxes for nonresidents. Majority of taxes (close to 80 percent) come from corporate profits levied on producers. Strong presence of SOEs in the minerals sector mean a large share of rents accrue directly to the state sector. In 1990s a number of countries in the region reduced tax rates to encourage investment in the minerals sector (Argentina, Chile and Peru). Over recent boom reversal of this trend: increase in tax pressure (e.g. Bolivia, Ecuador) and introduction of royalties (Chile in 2005, Peru in 2004).

Countries that tax exports of selected agricultural products, 1998-2008 Jales (2008). RAW HIDES & SKINS COFFEE VEGETABLE OILS COCOA LIVESTOCK Argentina Argentina Argentina Argentina Argentina Brazil Cameroon China Benin Belize Burundi Colombia Gambia Cameroon Burkina Faso Chad Costa Rica Indonesia Côte d'ivoire Cambodia Egypt Côte d'ivoire Malaysia Gambia Central African Rep. Gambia Ecuador Nepal Ghana Chad India Gambia Niger Niger Gambia Indonesia Guatemala Russia Nigeria Malaysia Kenya Guinea Solomon Islands Sierra Leone Niger Mongolia Niger Sri Lanka Solomon Islands Saint Kitts & Nevis Niger Sierra Leone Togo Togo Togo Pakistan Togo Ukraine Paraguay Uganda Saudi Arabia Sri Lanka Tanzania Thailand Togo Turkey Uganda Ukraine Uruguay Vietnam

Dependence on commodity revenues While on average, it may appear that abundant and dependent countries in the region do not appear to be more fiscally dependent compare to other regions (but more to be done on this topic). but large difference within the region and fiscal dependence is growing as a share of GDP and overall revenues. A substantial share of the region s GDP is generated in countries that are defined as hydrocarbon/mining dependent: Five out of the LAC-7 economies are classified as commodity dependent. Little evidence for agricultural product export tax dependence (except Argentina).

3. Fiscal cycles and commodity prices Procyclical fiscal policy response to business cycle movements in LA (Gavin and Perotti, 1997; Alesina, Callante and Tabellini, 2008). Two critical channels whereby commodities to trigger procylical response. i. Political economy explanations for procyclical expenditure responses to booms; and ii. Problems of revenues particular to dependence on fiscal revenues from commodity production.

Evidence that commodity revenue dependency leads to higher variability of the tax base Structure of taxation in LCR skewed towards consumption and natural resource taxation vs. personal and social security taxation in industrialized countries Commodity revenues more volatile than non-natural resource revenues Evidence of natural resource revenues offsetting other revenues (Bornhorst, Gupta and Thornton, 2008) and reduction in tax effort (Knack, 2008)

What do we mean by procyclical? Kaminsky, Reinhart and Végh (2004) defintion i. Procyclical if tax rates fall in booms and rise in recessions, and real government spending increases (decreases) during good (bad) times. ii. Acyclical if constant tax rates and real government spending over the cycle. iii. Countercyclical if tax rates rise (fall) in good (bad) times and real government spending decreases (increases) during good (bad) times.

4. Expenditure and revenue response during recent boom

Fiscal outcomes expressed as a share of GDP capture fluctuations in the commodity share of GDP 100% Non - commodity GDP / GDP 95% 90% 85% 80% 75% 70% 65% 60% 55% 50% 1998 1999 2000 2001 2002 2003 2004 2005 2006 Argentina Bolivia Chile Colombia Ecuador Mexico Peru Trinidad and Tobago Venezuela, RB Source: Own calculations using EIU,WDI and WITS-COMTRADE (SITC Rev 1.0) data. Note: Data to calculate Commodity Exports using WITS-COMTRADE SITC Rev 1, is unavailable for after 2006. Only for Ecuador, Mexico, and Peru we have for 2007

Therefore, examine fiscal revenues as a share of noncommodity GDP 50% Total Expenditure / non -commodity GDP, 1998-2006 (Average) 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% Argentina Bolivia Chile Colombia Ecuador Mexico Peru Trinidad and Tobago Venezuela Source: Own calculations using EIU,WDI and WITS-COMTRADE (SITC Rev 1.0) data.

Considerable growth in total expenditure over noncommodity GDP for many hydrocarbon producers in the region over 2002-2006 Trinidad and Tobago Venezuela Bolivia Argentina Peru Ecuador Mexico Chile Colombia Change in total expenditure over 2002-2006 -2% 0% 2% 4% 6% 8% 10% 12% 14% Note: The change in total expenditure over 2002-2006 [measured as the difference between the ratios of total expenditure to noncommodity GDP in 2002 and 2006]. Source: Own calculations using EIU,WDI and WITS-COMTRADE (SITC Rev 1.0) data. Note: Data to calculate Commodity Exports using WITS-COMTRADE SITC Rev 1, is unavailable for after 2006. Only for Ecuador, Mexico, and Peru we have for 2007

Similar results in primary expenditure. Interesting in the cases of Chile and Colombia Trinidad and Tobago Venezuela Bolivia Argentina Ecuador Peru Mexico Chile Colombia Change in primary expenditure over 2002-2006 -2% 0% 2% 4% 6% 8% 10% 12% 14% 16% Note: The change in primary expenditure over 2002-2006 [measured as the difference between the ratios of primary expenditure to non-commodity GDP in 2002 and 2006]. Source: Own calculations using EIU,WDI and WITS-COMTRADE (SITC Rev 1.0) data. Note: Data to calculate Commodity Exports using WITS-COMTRADE SITC Rev 1, is unavailable for after 2006. Only for Ecuador, Mexico, and Peru we have for 2007

For Chile, non-commodity nominal GDP growth was larger than nominal primary expenditure growth 44% 43% 42% 41% 40% 39% 38% 37% 36% Primary Expenditure Chile Non-commodity GDP - LCU Note: For each variable the figure indicates the growth between 2002 2006 Source: Own calculations using EIU,WDI and WITS-COMTRADE (SITC Rev 1.0) data. Note: Data to calculate Commodity Exports using WITS-COMTRADE SITC Rev 1, is unavailable for after 2006. Only for Ecuador, Mexico, and Peru we have for 2007

Real total expenditure has increased substantially in hydrocarbon-dependent countries on average over 2002-2008 Ecuador Trinidad and Tobago Venezuela Bolivia Mexico Argentina Peru Chile Colombia Real Total Expenditure (Average Annual Growth 2002-2008) 0% 2% 4% 6% 8% 10% 12% Source: Own calculations using EIU, and WDI.

Real primary expenditure shows the same pattern Ecuador Real Primary Expenditure (Average Annual growth 2002-2008) Trinidad and Tobago Venezuela Bolivia Argentina Mexico Peru Chile Colombia 0% 2% 4% 6% 8% 10% 12% 14% Source: Own calculations using EIU, and WDI.

Growth in total expenditure surpassed the growth in total revenues for Ecuador and Venezuela over 2002-2008; joined by many countries over 2006-2008 Countries Argentina Bolivia Chile Colombia Ecuador Mexico Peru Trinidad and Tobago Venezuela Change in Total Expenditure Relative to Change in Total Revenue (Percent) Sub-periods Entire period 2002-2004 2004-2006 2006-2008 2002-2008 67.6% 72.6% 41.2% 75.3% 73.3% 90.5% 100.1% 52.2% 34.4% 99.7% 78.9% 93.9% 95.3% 102.0% 127.5% 74.8% 117.1% 103.3% 88.4% 79.1% 58.6% 84.1% 102.9% 96.5% 83.3% 74.7% 100.7% 65.1% 67.0% 93.8% 91.9% 270.7% 116.6% 79.6% 93.1% 102.9% Source: Own calculations using EIU, and WDI.

Similar results for changes in primary expenditures relative to total revenues Countries Argentina Bolivia Chile Colombia Ecuador Mexico Peru Trinidad and Tobago Venezuela Change in Primary Expenditure Relative to Change in Total Revenue (Percent) Sub-periods Entire period 2002-2004 2004-2006 2006-2008 2002-2008 70% 89.1% 85.4% 81.9% 53.6% 53.3% 105.3% 79.6% 40.5% 34.7% 138.0% 61.2% 66.2% 86.8% 71.4% 75.0% 75.8% 75.7% 119.5% 104.1% 82.0% 85.2% 107.0% 92.5% 78.4% 61.8% 91.1% 77.0% 73.7% 66.4% 279.1% 93.5% 90.3% 93.2% 115.4% 100.3% Source: Own calculations using EIU, and WDI.

5. Preliminary results: Fiscal response to commodity price movements As Kaminsky, Reinhart and Végh (2004), estimate fiscal response to commodity price changes is estimated using: log(e t ) = β 0 + β t log (P it ) where i indexes the country and t denotes years. E t is total expenditure and for business cycles measure use change in (log of) a country-specific commodity export price index P it, both deflated by the CPI. Unbalanced panel for 19 countries over 1964-2008.

Dependent variable: Change in log real government expenditure Independent variable: Change in log real commodity export price index (1) (2) (3) (4) (5) (6) All Sample 1964-1969 1970-1979 1980-89 1990-99 2000-2008 A. Pooled OLS Commodity price change 0.02-0.33 0.07 0.03-0.15 0.08 (0.03) (0.11)* (0.04) (0.06) (0.06)* (0.07) Observations 638 44 87 152 189 166 [# countries] [19] [6] [18] [18] [19] [19] B. Fixed effects Commodity price change 0.02-0.65 0.07 0.04-0.15 0.03 (0.04) (0.24)* (0.06) (0.07) (0.15) (0.08) Observations 638 44 87 152 189 166 [# countries] [19] [6] [18] [18] [19] [19] Robust standard errors in parentheses. * denotes significant at 5%; ** significant at 1%.

Dependent variable: Change in log real government expenditure Independent variable: Change in log real commodity export price index Sub-sample (1) (2) (3) (4) Non- mineral Hydrocarbon dependent dependent Mineral dependent Nonhydrocarbon Dependent Period 1964-2008 1964-2008 1964-2008 1964-2008 A. Pooled Regression Change in commodity prices -0.04 0.07 0.00 0.04 (0.06) (0.04) (0.05) (0.05) Observations 244 394 172 466 [# countries] [8] [11] [6] [13] B. Fixed effects Change in commodity prices -0.03 0.07 0.01 0.03 (0.05) (0.08) (0.07) (0.06) Observations 244 394 172 466 [# countries] [8] [11] [6] [13] Robust standard errors in parentheses. * denotes significant at 5%; ** significant at 1%.

Dependent variable: Change in log real government expenditure Independent variable: Change in log real commodity export price index Sub-sample Mineral dependent Non- mineral dependent Hydrocarbon dependent Nonhydrocarbon Dependent Period 2000-2008 2000-2008 2000-2008 2000-2008 A. Pooled Regression Change in commodity prices -0.00-0.03 0.16-0.00 (0.09) (0.09) (0.17) (0.06) Observations 72 94 54 112 [# countries] [8] [11] [6] [13] B. Fixed effects Change in commodity prices -0.05-0.04 0.07-0.01 (0.10) (0.15) (0.20) (0.11) Observations 72 94 54 112 [# countries] [8] [11] [6] [13] Robust standard errors in parentheses. * denotes significant at 5%; ** significant at 1%.

Dependent variable: Change in log real government revenues Independent variable: Change in log real commodity export price index (1) (2) (3) All Mineral dependent Non-mineral dependent Sub-period 2000-2008 2000-2008 2000-2008 A. Pooled Regression Change in commodity prices 0.27 0.26 0.05 (0.08)** (0.11) (0.08) Observations 166 72 94 [# countries] [19] [8] [11] R-squared 0.25 0.47 0.14 B. Fixed effects Change in commodity prices 0.23 0.24 0.04 (0.08)** (0.11)* (0.12) Observations 638 244 394 [# countries] [19] [8] [11] R-squared 0.29 0.49 0.21 Robust standard errors in parentheses. * denotes significant at 5%; ** significant at 1%.

Revenue and expenditure sensitivity to commodity price movements, 2000-2008 Expenditure coefficient -.3 -.2 -.1 0.1.2.3.4.5.6.7.8.9 Colombia Peru Mexico Chile Trinidad and Tobago Bolivia Ecuador Argentina Venezuela, RB -.1 0.1.2.3.4.5.6.7.8.9 1 1.1 1.2 Revenue coefficient

Preliminary results Expenditure sensitivity Coefficient on commodity price index insignificantly different from zero for all subperiods and subgroups examined. Exception 1990-99 period when there is an indication of an acyclical response (and for 1964-1969, but limited data coverage). Revenue sensitivity Tax rates the appropriate measure, but data unavailable. Strong co-movement found between revenues and commodity prices (as expected)

6. To be continued Build on overview country case studies Empirical analysis Alternative dependent variable Specification Thresholds effects