## 8. Volatility

## Source details

**Canonical URL:** [8. Volatility](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0649.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0649.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0649.pdf.json)

---

### Introduction and objectives
- Coverage: four large Latin American economies — Argentina, Brazil, Chile, and Mexico — which together accounted for "some 70 percent" of the region’s GDP over the past half century.
- Sample span: 1870–2004.
- Four main questions:
  - How volatile has Latin America been relative to other countries and across sub-periods (e.g., pre-1930 gold standard, post-1980s)?
  - How persistent have macroeconomic fluctuations been?
  - Do stylized facts documented for other economies hold for these countries?
  - Is there an identifiable regional business cycle?

### Methodology: dynamic common factor backcasting
- Constructed indices of economic activity using a dynamic common factor methodology applied to a broad set of macroeconomic and sectoral variables.
- Dataset composition:
  - Between 20 to 25 time series per country in the final panel.
  - Variables include agriculture, manufacturing, mining and cement production, fiscal expenditures and revenues, terms of trade, real exchange rate, import and export volume, interest rates, monetary aggregates, and external variables (foreign interest rates, real income in advanced countries).
- Detrending and stationarity:
  - Hodrick-Prescott (HP) filter with smoothing factor set to 100.
  - Baxter and King band-pass with symmetric moving average parameter set to three and bandwidth 2 to 20 years.
- Factor estimation approaches:
  - Stock and Watson static principal components (SW).
  - Forni, Hallin, Lippi and Reichlin dynamic/generalized principal components (FHLR).
  - Two-step weighted principal components (Boivin and Ng style) with estimated Ω.
- Backcasting equation and estimation:
  - Real GDP cyclical component γt modeled as yct = α + b′F t + εt.
  - Backcasting parameters estimated over recent sample (e.g., 1950–2004 or 1915–2004 depending on country) and applied to factor series over 1870–2004.

### New business cycle indices and robustness
- Coverage and construction notes:
  - Full national accounts available only from the mid-1930s (Argentina) or 1940s (Brazil, Chile, Mexico) — extended backcasting relies on broader panel indicators.
  - Mexico backcast index starts in 1878 due to data availability.
  - In some countries (e.g., Brazil, Argentina) key sectoral series only become available from 1900 or 1875; proxy series used (e.g., cement consumption for construction).
- Empirical fit (1950–2004):
  - Argentina: R−values vary from 0.89 (all-regressor) to around 0.80 (two-factor SW/FHLR approaches).
  - Chile and Mexico: first two factors explain "75–85 percent" of the variance of the real GDP cycle.
  - Brazil: fit improves above 70 percent when agricultural and manufacturing series (available from 1900) are included.
- Robustness checks:
  - Sensitivity analysis with 72 specifications for Argentina, 36 for Brazil, and 12 for Chile and Mexico varying estimation sample (1915–2004 and 1950–2004), factor specifications (SW r = 1,2,3; FHLR (q,r) = (1,2), (3,1), (2,2), (2,3)), sample breaks, and inclusion/exclusion of external variables.
  - Results: narrow ranges across specifications except for isolated outliers (Brazil 1890–91, 1986, 1989; Chile 1929–32; Mexico 1916), with all estimates pointing in the same direction.
  - Recursive estimation of backcasting regression coefficients shows reasonable stability for 1930–60 (1940–60 for Chile); some instability in Mexico 1921–1925 likely due to measurement problems in official series pre-1925.
- Gains from the extended dataset:
  - Factors built from sectoral + fiscal + financial + external variables capture deeper crises (e.g., Argentina 1918 and 1990; Brazil and Mexico post-World War I; Chile after the Great Depression) that narrow sectoral-only indices miss.
- U.S. backcast validation:
  - Applied same methodology to U.S. pre-war cycle using a restricted set of 18 variables.
  - Post-war fit: R−values around 0.80 for factor models.
  - Backcast closely tracks Balke and Gordon (1989) revised U.S. real GDP cycle (largest discrepancy at World War II boom).

### Stylized business cycle facts for the four countries
- Volatility and persistence (1870–2004):
  - Average business cycle volatility in all four countries was considerably higher than in advanced economies over the full sample, with important sub-period differences.
  - Volatility pattern by sub-period:
    - Relatively high pre-1930.
    - Dropped sharply in 1930–70.
    - Rose again in the 1970s–1980s (more than twice advanced country average).
    - Declined sharply more recently.
  - Cyclical persistence has been high throughout, producing long business cycle durations relative to advanced country standards.
- Cycle duration and spectral properties:
  - Pre-1930 dominant cyclical length around 14 to 16 years (Kuznets-type swings).
  - Post-war dominant cycle around 10–12 years.
  - Bry and Boschan dating reported using two-year and six-year windows; longer window emphasizes Kuznets swings.
- Variable-specific stylized facts:
  - Terms of trade: strongly procyclical; large amplitude.
  - Trade balance: counter-cyclical.
  - Fixed investment: several times more volatile than output.
  - Fiscal variables:
    - Government spending volatility higher than output volatility.
    - Real government expenditure cycle and public expenditure/revenues ratio show large magnitudes.
    - Fiscal policy (Kaminsky, Reinhart, and Végh (2004) gauge) found to be strongly procyclical.
  - Monetary aggregates: generally less volatile than fiscal variables, except Argentina and Brazil over the past two decades and Chile in the 1970s (periods of high/hyper-inflation).
  - Inflation: historically counter-cyclical in all four countries.
  - Real wages: broadly procyclical.
- Cross-country synchronization and regional factor:
  - Concordance indices (Harding and Pagan 2002) indicate a reasonably high degree of synchronization across the four countries for 1870–2004.
  - Pooled dynamic factor extraction identifies a sizeable common regional factor.
  - Correlation of the regional factor with the country procyclical factor (F1) ranges between 0.6 and 0.75.
  - Despite small intra-regional trade historically and limited capital account linkages, global shocks (foreign interest rates, advanced countries' real income, commodity terms of trade) emerge as key drivers of the common regional cycle.

### Conclusions and policy-relevant implications
- Long-run characterization:
  - Latin America historically displayed high cyclical volatility relative to advanced economies; volatility was time-varying across sub-periods (high pre-1930, low 1930–70, high 1970s–1980s, lower recently).
  - High cyclical persistence amplified welfare costs associated with volatility.
- Sources and amplification:
  - External factors (terms of trade, foreign interest rates, advanced-country income) are important drivers of country cycles and the regional common cycle.
  - Domestic amplification mechanisms appear important: large fiscal swings (procyclical government spending), volatile public finances, and, to a lesser extent, financial and monetary variability.
- Policy implications suggested:
  - Stabilizing fiscal policy and smoothing public expenditure cycles could materially reduce output volatility given fiscal procyclicality and the large magnitudes of fiscal swings.
  - Managing exposure to terms-of-trade shocks and designing institutions to better absorb external shocks matter for reducing volatility and persistence.
  - Because global shocks drive much of the regional co-movement, regional risk-sharing mechanisms historically had limited scope; policies that improve shock absorption and reduce domestic amplification can be valuable.

### Appendix I — Data construction and sources (selected series and coverage by country)
- Argentina (selected series and coverage highlights):
  - Agricultural Output: 1900–1960 (Banco Central de Argentina (1976)); 1961–2004 (World Bank, World Development Database).
  - Industrial Output: 1875–1960 (della Paolera and Taylor (2003)); 1961–2004 (World Bank, World Development Database).
  - Cement Consumption: 1870–1913 (total cement imports from the UK and US); 1913–2000 (Oxford Latin American Economic History database); 2001–2004 (Instituto Nacional de Estadisticas y Census de la Republica Argentina).
  - Fixed Investment: 1870–1884 (capital goods imports from the UK and US); 1885–1960 (della Paolera and Taylor (2003)); 1961–2004 (IFS, IMF).
  - Central Government Expenditures and Revenues: 1870–1960 (Brian M. Mitchell, 1998); 1961–2004 (Luis A.V. Catão and Marco E. Terrones, 2005), deflated by CPI.
  - Narrow (M0) and Broad Money (M2): 1870–1960 (Mitchell); 1961–2004 (IFS), deflated by CPI.
  - CPI: 1870–1913 (Catão and Solomou (2005)); 1914–1960 (della Paolera and Taylor (2003)); 1961–2004 (IFS).
  - Average Interest Rate on Domestic Public Bonds: 1870–1913 (Monetary and Banking Experiments in Argentina: 1861–1930); 1914–1993 (della Paolera and Javier Ortiz, 1995); 1993–2004 (IFS line 60p). Real interest rate series deflated by current period CPI inflation.
  - Export and Import volumes and Net Barter Terms of Trade: 1870–1913 (Catão and Solomou (2005)); 1914–1960 (Oxford Latin American Economic History database); 1961–2004 (IFS).
  - Real Effective Exchange Rate: 1870–1913 (Catão and Solomou (2005)); 1914–2004 (CPI-based geometric weighted averages with eight largest trading partners). Fisher ideal indices for 1914–1946 and 1946–2004, spliced at 1946. Nominal exchange rates source: Carmen M. and Kenneth S. Rogoff, 2004.
  - Net Foreign Capital Inflows and other series: detailed splicing and deflation procedures described in source.
- Brazil (selected series and coverage highlights):
  - Agricultural and Manufacturing Output: 1900–1946 (Haddad, 1978); 1947–2004 (IBGE).
  - Transport, Communications, Cement Consumption, Machinery Investment: series span variously from 1870 onwards with sources Haddad (1978), Villela and Suzigan (1975), Wilson Suzigan (1986), IBGE, IPEA.
  - Central Government Expenditures and Revenues: 1870–1960 (IBGE); 1961–2004 (Catão and Terrones, 2005), deflated by GDP deflator.
  - Narrow (M1) and Broad Money (M2): 1870–1960 (IBGE); 1961–2004 (IFS), deflated by GDP deflator.
  - Real Effective Exchange Rate: 1870–1913 (Catão and Solomou (2005)); 1914–2004 (GDP deflator-based geometric weighted averages with eight largest trading partners covering between 64 and 75 percent of visible trade). Fisher ideal indices spliced at 1946.
  - Domestic Interest Rate: various historical series and interpolations explained; real interest rate series deflated by annual changes in GDP deflator.
  - External Interest Rate Spread and other external series: sources and construction documented.
- Chile (selected series and coverage highlights):
  - Agriculture, Manufacturing and Mining Output: 1870–1960 (Braun et al., 2000); 1961–2004 (World Development database, World Bank, and Central Bank of Chile).
  - Machinery Investment: 1870–1900 (capital goods imports deflated by UK capital good deflator); 1901–1990 (Andre Hoffman, 2000); 1991–2004 (Central Bank of Chile).
  - Central Government Expenditures and Revenues: 1870–1960 (Brian M. Mitchell, 1998); 1961–2004 (Catão and Terrones, 2005), deflated by CPI.
  - Narrow Money (M0) and Broad Money (M2): various sources 1870–2004; real terms by deflating by CPI.
  - Real Effective Exchange Rate: 1870–1913 (Catão and Solomou (2005)); 1914–2004 (CPI-based geometric weighted averages with up to twenty largest trading partners covering between 79 and 93 percent of visible trade). Fisher ideal indices spliced at 1946.
- Mexico (selected series and coverage highlights):
  - Agricultural, Manufacturing, Mining Output: series beginning 1878 for many series; multiple sources including Colegio de México (1960), Solis (1975), World Development database, INEGI.
  - Machinery Investment: 1870–1925 (Luis Catão, 2005); 1925–1940 (Enrique Cardenas, 1987); 1941–1993 (Hoffman); 1994–2004 (INEGI).
  - Central Government Expenditures and Revenues: 1870–1960 (IBGE); 1961–2004 (Catão and Terrones, 2005), deflated by CPI.
  - Narrow (M0) and Broad Money (M2): 1870–1925 (Catão); 1926–1960 (INEGI); 1961–2004 (IFS), deflated by CPI.
  - Real Effective Exchange Rate: 1870–1913 (Catão and Solomou (2005)); 1914–2004 (CPI-based geometric weighted averages with six largest trading partners covering over 90 percent of visible trade). Fisher ideal indices spliced at 1946.
- United States (selected series and coverage highlights for validation):
  - Agricultural, Manufacturing, Transportation and other sectoral outputs: 1870–1949 (NBER historical database; Balke and Gordon (1989)); 1950–2004 (Bureau of Economic Analysis chained quantity indices).
  - Central Government Expenditures and Revenues: 1870–1958 (Mitchell); 1959–2004 (IFS).
  - Narrow (M0) and Broad Money (M2): 1870–1970 (Bureau of Economic Analysis, Historical Statistics); 1970–2004 (Haver Analytics).
  - CPI and other series: multiple historical sources spliced and documented.
  - Foreign Output: real GDP of Australia, Canada, UK, France, Germany, Italy and Japan in 1990 PPP constant dollars from Maddison (2003) extended with IFS through 2004.

*Source: IMF Working Paper chapter "8. Volatility" (content unit _wp0649).*

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

### _wp0649 - References................................................................................................................................23

### Tables
- 1. Factor Loadings ...................................................................................................................28
- 2. In-Sample Fit .......................................................................................................................29
- 3. In-Sample and Out-of-Sample Fit for the United States......................................................30
- 4. Dating the Cycle ..................................................................................................................31
- 5. Spectral Density Function Estimates of Cyclical Durations................................................32
- 6. Persistence and Volatility of the Cycle ................................................................................33
- 7. International Comparisons ...................................................................................................34
- 8. Cyclical Volatility Estimates of Selected Variables ............................................................35
- 9. Synchronization of the Latin American Cycles ...................................................................36

### Figures
- 1. Common Factors ..................................................................................................................37
- 2. Actual and Backcasted Values of Cyclical Growth .............................................................38
- 3. Robustness Check  ...............................................................................................................40
- 4. Recursive Parameter Estimates............................................................................................42
- 5. Estimates of the First Common Factor ................................................................................43
- 6. Actual and Backcasted Values of the US Business Cycle ...................................................44
- 7. Historical Charts ..................................................................................................................45

*Source: _wp0649 - References..............................................................................................................*

### 8. Volatility ..........................................................................................................

### 8. Volatility

### Introduction and objectives
- Paper covers four large Latin American economies: Argentina, Brazil, Chile, and Mexico, which together accounted for "some 70 percent" of the region’s GDP over the past half century.
- Sample span: 1870–2004.
- Four main questions addressed:
  - How volatile has Latin America been relative to other countries and across sub-periods (e.g., pre-1930 gold standard, post-1980s)?
  - How persistent have macroeconomic fluctuations been?
  - Do stylized facts documented for other economies hold for these countries?
  - Is there an identifiable regional business cycle?

### Methodology: dynamic common factor backcasting
- Construct new indices of economic activity for each country using a dynamic common factor methodology applied to a broad set of macroeconomic and sectoral variables.
- Dataset composition:
  - Between 20 to 25 time series per country in the final panel.
  - Variables include: agriculture, manufacturing, mining and cement production, fiscal expenditures and revenues, terms of trade, real exchange rate, import and export volume, interest rates, monetary aggregates, and external variables (foreign interest rates, real income in advanced countries).
- Detrending and stationarity:
  - Two detrending approaches used: Hodrick-Prescott (HP) filter with smoothing factor set to 100; Baxter and King band-pass with symmetric moving average parameter set to three and bandwidth 2 to 20 years.
- Factor estimation approaches:
  - Stock and Watson static principal components (SW).
  - Forni, Hallin, Lippi and Reichlin dynamic/generalized principal components (FHLR).
  - Two-step weighted principal components (Boivin and Ng style) with estimated Ω.
- Backcasting equation:
  - Real GDP cyclical component γt is modeled as driven by common factors ft: yct = α + b′F t + εt (notation consistent with the paper).
  - Backcasting parameters estimated over recent sample (e.g., 1950–2004 or 1915–2004 depending on country data availability) and applied to factor series over the longer 1870–2004 span.

### New business cycle indices and robustness
- Coverage and data construction:
  - Full national accounts available only from the mid-1930s (Argentina) or 1940s (Brazil, Chile, Mexico) — extended backcasting relies on the broader panel of indicators.
  - Mexico backcast index starts in 1878 due to data availability.
  - In some countries (e.g., Brazil, Argentina) key sectoral series only become available from 1900 or 1875; proxy series used (e.g., cement consumption for construction).
- Empirical fit (1950–2004):
  - Argentina: R−values vary from 0.89 (all-regressor) to around 0.80 (two-factor SW/FHLR approaches).
  - Chile and Mexico: first two factors explain "75–85 percent" of the variance of the real GDP cycle.
  - Brazil: fit improves above 70 percent when agricultural and manufacturing series (available from 1900) are included.
- Robustness checks and sensitivity:
  - Sensitivity analysis considered 72 specifications for Argentina, 36 for Brazil, and 12 for Chile and Mexico, varying: backcasting estimation sample (1915–2004 and 1950–2004), factor specifications (SW with r = 1,2,3; FHLR with (q,r) = (1,2), (3,1), (2,2), (2,3)), factor estimation sample breaks, inclusion/exclusion of external variables.
  - Result: narrow ranges across specifications except for isolated outliers (Brazil 1890–91, 1986, 1989; Chile 1929–32; Mexico 1916), with all estimates pointing in the same direction.
  - Recursive estimation of backcasting regression coefficients (e.g., re-estimating over 1961–2004 and rolling back) shows reasonable stability for 1930–60 (1940–60 for Chile); some instability in Mexico 1921–1925 likely due to measurement problems in official series pre-1925.
- Gains from extended dataset:
  - Factors built from the wider set (sectoral + fiscal + financial + external) capture deeper crises (e.g., Argentina 1918 and 1990; Brazil and Mexico post-World War I; Chile after the Great Depression) that narrow sectoral-only indices miss.
- U.S. backcast validation:
  - Applied same methodology to U.S. pre-war cycle using a restricted set of 18 variables.
  - Post-war fit: R−values around 0.80 for factor models.
  - Backcast closely tracks Balke and Gordon (1989) revised U.S. real GDP cycle (largest discrepancy at World War II boom).

### Stylized business cycle facts for the four countries
- Volatility and persistence (1870–2004):
  - Average business cycle volatility in all four countries was considerably higher than in advanced economies over the full sample, with important sub-period differences.
  - Volatility was relatively high pre-1930; dropped sharply in 1930–70; rose again in the 1970s–1980s (more than twice advanced country average), then declined sharply more recently.
  - Cyclical persistence has been high throughout, yielding a combination of high volatility and long business cycle durations relative to advanced country standards.
- Cycle duration and spectral properties:
  - Pre-1930 dominant cyclical length around 14 to 16 years (Kuznets-type swings); post-war dominant cycle around 10–12 years.
  - Bry and Boschan dating reported using two-year and six-year windows; longer window emphasizes Kuznets swings.
- Variable-specific stylized facts:
  - Terms of trade: strongly procyclical; large amplitude.
  - Trade balance: counter-cyclical.
  - Fixed investment: several times more volatile than output.
  - Fiscal variables: government spending volatility higher than output volatility; real government expenditure cycle and public expenditure/revenues ratio show large magnitudes.
  - Fiscal policy: using the gauge of Kaminsky, Reinhart, and Végh (2004), fiscal policy found to be strongly procyclical in these countries.
  - Monetary aggregates: generally less volatile than fiscal variables, except Argentina and Brazil over the past two decades and Chile in the 1970s (periods of high/hyper-inflation).
  - Inflation: historically counter-cyclical in all four countries (contrast with Phillips-curve style trade-offs in many advanced economies).
  - Real wages: broadly procyclical.
- Cross-country synchronization and regional factor:
  - Concordance indices (Harding and Pagan 2002) indicate a reasonably high degree of synchronization across the four countries for 1870–2004.
  - Pooled dynamic factor extraction identifies a sizeable common regional factor.
  - Correlation of the regional factor with the country procyclical factor (F1) ranges between 0.6 and 0.75.
  - Despite small intra-regional trade historically and limited capital account linkages, global shocks (foreign interest rates, advanced countries' real income, commodity terms of trade) emerge as key drivers of the common regional cycle.

### Conclusions and policy-relevant implications
- Long-run characterization:
  - Latin America historically displayed high cyclical volatility relative to advanced economies; volatility was time-varying across sub-periods (high pre-1930, low 1930–70, high 1970s–1980s, lower recently).
  - High cyclical persistence amplified welfare costs associated with volatility.
- Sources and amplification:
  - External factors (terms of trade, foreign interest rates, advanced-country income) are important drivers of country cycles and the regional common cycle.
  - Domestic amplification mechanisms appear important: large fiscal swings (procyclical government spending), volatile public finances, and, to a lesser extent, financial and monetary variability.
- Policy implications suggested by findings:
  - Stabilizing fiscal policy and smoothing public expenditure cycles could materially reduce output volatility given fiscal procyclicality and the large magnitudes of fiscal swings.
  - Managing exposure to terms-of-trade shocks and designing institutions to better absorb external shocks matter for reducing volatility and persistence.
  - Because global shocks drive much of the regional co-movement, regional risk-sharing mechanisms historically had limited scope; policies that improve shock absorption and reduce domestic amplification can be valuable.

*Source: IMF Working Paper chapter "8. Volatility" (content unit _wp0649).*

### REFERENCES

### _wp0649 - REFERENCES

### Major referenced literature (themes)
- Institutional determinants and long-run growth: Acemoglu et al., 2003; Acemoglu, Johnson and Robinson, 2006.
- Business cycles, historical estimates, and measurement methods: Backus and Kehoe, 1992; Balke and Gordon, 1989; Basu and Taylor, 1999; Baxter and King, 1999; Burns and Mitchell, 1946; Kuznets, 1958; Romer, 1989.
- Factor models, diffusion indexes, and forecasting methods: Forni et al., 2000; Forni et al., 2003; Stock and Watson, 1989, 1991, 1999, 2002, 2005; Bernanke, Boivin, and Eliasz, 2005; Boivin and Ng, 2003.
- Monetary policy measurement and real-time analysis: Giannone, Reichlin, and Sala, 2004; Bernanke et al., 2005.
- Trade, terms of trade, and external shocks: Mendoza, 1995; Kose and Reizman, 2001; Kose, Otrok, and Whiteman, 2003; Kose, Prasad, and Terrones, 2006.
- Capital flows, external finance, and crisis dynamics: Calvo, Leiderman, and Reinhart, 1993; Fernandez-Arias, 1994; Kaminsky, Reinhart, and Végh, 2004; Van Wincoop, 1999.
- Historical and country-specific economic histories and datasets: Maddison, 1995, 2003; della Paolera and Taylor, 2003; Braun et al., 2000; Haddad, 1978; Goldsmith, 1986; Mitchell, 1998; various country monographs and archival sources.

### Appendix I — Data construction and sources: Argentina (selected series and coverage)
- Agricultural Output:
  - 1900–1960: Banco Central de Argentina (1976)
  - 1961–2004: World Bank, World Development Database.
- Industrial Output:
  - 1875–1960: della Paolera and Taylor (2003)
  - 1961–2004: World Bank, World Development Database.
- Transport Output:
  - 1870–1960: Geometric weighted average of passengers and tons of freight per kilometers times total railway road extension; spliced in 1913 with index in Carlos F. Diaz-Alejandro, 1970.
- Cement Consumption:
  - 1870–1913: Total cement imports from the United Kingdom and the United States (UK Board of Trade; US Foreign Commerce, Navigation and Tonnage).
  - 1913–2000: Oxford Latin American Economic History database.
  - 2001–2004: Instituto Nacional de Estadisticas y Census de la Republica Argentina.
- Fixed Investment:
  - 1870–1884: Capital goods imports from the United Kingdom and the United States, deflated by UK capital good deflator (Feinstein, 1972).
  - 1885–1960: della Paolera and Taylor (2003).
  - 1961–2004: International Financial Statistics (IFS), IMF.
- Central Government Expenditures and Revenues:
  - 1870–1960: Brian M. Mitchell, 1998.
  - 1961–2004: Luis A.V. Catão and Marco E. Terrones, 2005; series expressed in real terms by deflating by CPI.
- Narrow (M0) and Broad Money (M2):
  - 1870–1960: Mitchell, op. cit.
  - 1961–2004: IFS; both series expressed in real terms by deflating by the CPI.
- Consumer Price Index (CPI):
  - 1870–1913: Catão and Solomou (2005).
  - 1914–1960: della Paolera and Taylor (2003).
  - 1961–2004: IFS.
- Average Interest Rate on Domestic Public Bonds:
  - 1870–1913: `Monetary and Banking Experiments in Argentina: 1861–1930’, Paper presented March, 1992.
  - 1914–1993: della Paolera and Javier Ortiz, 1995.
  - 1993–2004: IFS (line 60p). Real interest rate series obtained by deflating annual nominal yields by current period CPI inflation.
- Export and Import volumes and Net Barter Terms of Trade:
  - 1870–1913: Catão and Solomou (2005).
  - 1914–1960: Oxford Latin American Economic History database.
  - 1961–2004: IFS.
- Real Effective Exchange Rate:
  - 1870–1913: Catão and Solomou (2005).
  - 1914–2004: CPI-based geometric weighted averages with eight largest trading partners (covering between 67 and 80 percent of visible trade). Fisher ideal indices for 1914–1946 and 1946–2004, spliced at 1946.
  - Nominal exchange rates source: Carmen M. and Kenneth S. Rogoff, 2004.
- Net Foreign Capital Inflows:
  - 1870–1960: Spliced series using Stone (1999) UK capital flows and della Paolera (1988) and Cavallo-Mundlak series; series deflated by US WPI (Global Financial Database for 1870–1947; IFS for 1948–2004).
  - 1961–2004: Difference between changes in international reserves and current account balance from IFS; deflation by US WPI.
- Wages:
  - 1870–1913: Jeffrey G. Williamson, 1995.
  - 1914–1980: della Paolera and Taylor (2003).
  - 1981–2004: IMF's WEO database; series deflated by CPI for real wage index.
- Foreign 3-month bill rate:
  - 1870–1920: Annual average yields of 3-month bills on the London market (Holmer and Sillas, 1996).
  - 1921–2004: Annual average yields of the US 3-month Treasury Bill (Holmer and Sillas, 1996).
  - Both series deflated by respective countries' CPI inflation (Catão and Solomou 1870–1913; Mitchell 1914–1960; IFS 1961–2004).
- Foreign Output:
  - Sum of French, German, U.K., and U.S. GDPs expressed in 1990 PPP constant dollars from Maddison (2003).
- Population at mid-year:
  - 1870–1960: Maddison (2003).
  - 1961–2004: IFS (2003).

### Appendix I — Data construction and sources: Brazil (selected series and coverage)
- Agricultural and Manufacturing Output:
  - 1900–1946: Haddad (1978).
  - 1947–2004: Instituto Brasileiro de Geografia e Estatistica (IBGE), http://www.IBGE.gov.br.
- Transport Output:
  - 1870–1907: Average of freight and passenger transported in railways, using 1908 weights (Haddad, 1978).
  - 1908–1946: Haddad (1978).
  - 1947–2004: IBGE.
- Communications Output:
  - 1870–1907: Average of mail and telegraph traffic weighted by 1889 values (IBGE, 1987).
  - 1908–1946: Haddad (1978).
  - 1947–2004: IBGE.
- Cement Consumption:
  - 1870–1901: Cement imports from France, Germany, the United Kingdom, and the United States.
  - 1902–1945: Villela and Suzigan, 1975.
  - 1945–2004: IBGE; IPEA database.
- Machinery Investment:
  - 1870–1913: Wilson Suzigan, Industria Brasileira, 1986.
  - 1913–2004: IBGE.
- Central Government Expenditures and Revenues:
  - 1870–1960: IBGE.
  - 1961–2004: Catão and Terrones, 2005; series deflated by the GDP deflator.
- Narrow (M1) and Broad Money (M2):
  - 1870–1960: IBGE.
  - 1961–2004: IFS; real terms by deflating by the GDP deflator.
- GDP deflator:
  - 1870–1913: Catão and Solomou (2005).
  - 1914–1960: IBGE.
  - 1961–2004: IFS.
- Domestic Interest Rate:
  - 1870–1961: Yields on government perpetuities (“apolices”); pre-1930 data from research by Bill Summerhill and Gail Trinner; post-1930 from Claudio Contador, 1975.
  - 1965–1980: Equivalent nominal yield on inflation indexed public bonds (ORTNs) from Goldsmith.
  - Gap bridged by linear interpolation; 1981–2004: Money market interest rate from IFS.
  - Real interest rate series obtained by deflating annual nominal yields by current annual percentage changes in the GDP deflator.
- External Interest Rate Spread:
  - 1870–1913: Mauro, Sussman and Yafeh, 2002.
  - 1914–1956: Global Financial Data.
  - 1968–1986: Cardoso and Fishlow, in Sachs (ed.).
  - 1987–1993: Estimated as the one-year libor interest rate plus a 400 basis points spread minus the US 10-year bond interest rate.
  - 1993–2004: IMF's global data source database.
- Export and Import volumes and Net Barter Terms of Trade:
  - 1870–1913: Catão and Solomou (2005).
  - 1914–1960: IBGE.
  - 1961–2004: IFS.
- Real Effective Exchange Rate:
  - 1870–1913: Catão and Solomou (2005).
  - 1914–2004: GDP deflator-based geometric weighted averages with eight largest trading partners (covering between 64 and 75 percent of visible trade). Fisher ideal indices for 1914–1946 and 1946–2004, spliced at 1946.
  - Nominal exchange rate source: Carmen M. and Kenneth S. Rogoff, 2004.
- Wages:
  - 1870–1940: Jeffrey G. Williamson, 1995.
  - 1940–1955: IBGE.
  - 1955–1976: Raymond Goldsmith, 1986.
  - 1977–2004: IBGE.
- Foreign 3-month bill rate and Foreign Output:
  - The same as for Argentina.
- Population at mid-year:
  - 1870–1960: Maddison (2003).
  - 1961–2004: IFS (2003).

### Appendix I — Data construction and sources: Chile (selected series and coverage)
- Agriculture, Manufacturing and Mining Output:
  - 1870–1960: Braun et al., 2000, Documento de Trabajo No. 187.
  - 1961–2004: World Development database, World Bank, and Central Bank of Chile.
- Machinery Investment:
  - 1870–1900: Capital goods imports from the United Kingdom and the United States, deflated by UK capital good deflator (Feinstein, 1972).
  - 1901–1990: Andre Hoffman, 2000.
  - 1991–2004: Central Bank of Chile.
- Central Government Expenditures and Revenues:
  - 1870–1960: Brian M. Mitchell, 1998.
  - 1961–2004: Catão and Terrones, 2005; series deflated by CPI.
- Narrow Money (M0) and Broad Money (M2):
  - 1870–1878: M0 from Llona Rodriguez (calculated as paper money issued minus banks' cash-in-vault); M2 from Braun et al.
  - 1879–1960: M0 from Mitchell; M2 from Braun et al.
  - 1961–2004: IFS; series expressed in real terms by deflating by CPI.
- Mortgage Credit:
  - 1870–1995: Braun et al. (1995).
  - 1995–2004: Central Bank of Chile.
- Consumer Price Index (CPI):
  - 1870–1913: Catão and Solomou (2005).
  - 1914–1960: Mitchell.
  - 1961–2004: IFS.
- Domestic Interest Rate:
  - 1870–1993: Bank lending rate from Braun et al.
  - 1993–2004: IFS (line 60p); real interest rate series obtained by deflating annual nominal yields by current period CPI inflation.
- Export and Import volumes and Net Barter Terms of Trade:
  - 1870–1913: Catão and Solomou (2005).
  - 1914–1960: Braun et al.
  - 1961–2004: IFS.
- Real Effective Exchange Rate:
  - 1870–1913: Catão and Solomou (2005).
  - 1914–2004: CPI-based geometric weighted averages with up to twenty largest trading partners (covering between 79 and 93 percent of visible trade). Fisher ideal indices for 1914–1946 and 1946–2004, spliced at 1946.
  - Nominal exchange rates source: Carmen M. and Kenneth S. Rogoff, 2004.
- Wages:
  - 1870–1995: Real wage index from Braun et al.
  - 1995–2004: Average nominal wage index from IMF's WEO database, deflated by CPI.
- Population at mid-year:
  - 1870–1960: Maddison (2003).
  - 1961–2004: IFS (2003).

### Appendix I — Data construction and sources: Mexico (selected series and coverage)
- Agricultural Output:
  - 1878–1910: Colegio de México, 1960, Estadísticas Economicas del Porfiriato, p.61 (export crop sub-index; converted from fiscal to calendar year by averaging two successive years).
  - 1911–1921: Index constructed as weighted average of output of ten main crops (beans, corn (maiz), cotton, coffee, garbanzo, rice, sisal, sugar, and tomatoes) weighted by their 1900 (normalized) share.
  - 1921–1960: Leopoldo Solis, 1975.
  - 1961–2004: World Development database, World Bank, and INEGI.
- Manufacturing Output:
  - 1878–1910: Colegio de México, op. cit., p.105; prior to 1892 reflects domestic textile production index; gaps in 1879–1882, 1884–87, and 1890 interpolated linearly.
  - 1910–1921: INDEC.
  - 1921–1960: Solis, 1975.
  - 1961–2004: World Development database, World Bank, and INEGI.
- Mining Output:
  - 1878–1910: Colegio de México, op. cit., p.135.
  - 1911–1921: Weighted average of ten main domestically produced metals and oil.
  - 1921–1960: Solis, 1975.
  - 1961–2004: World Development database, World Bank, and INEGI.
- Transportation and Communications Output:
  - 1870–1921: Weighted average of railway freight and passenger traffic (John Coatsworth) and postal service traffic (Mitchell).
  - 1921–1960: Solis, 1975.
  - 1961–2004: World Development database, World Bank, and INEGI.
- Cement Consumption:
  - 1870–1931: Cement imports from the United Kingdom and the United States; from 1906 domestic production added.
  - 1932–2000: Oxford Latin American Economic History database.
  - 2001–2004: INEGI.
- Machinery Investment:
  - 1870–1925: Luis Catão, 2005 (unpublished).
  - 1925–1940: Enrique Cardenas, 1987 (index based on volume of imported capital goods).
  - 1941–1993: Hoffman, op. cit.
  - 1994–2004: INEGI.
- Central Government Expenditures and Revenues:
  - 1870–1960: IBGE, op. cit.
  - 1961–2004: Catão and Terrones, 2005; both series expressed in real terms by deflating by the CPI deflator.
- Narrow (M0) and Broad Money (M2):
  - 1870–1925: Catão, op. cit.
  - 1926–1960: INEGI.
  - 1961–2004: IFS; resulting real series obtained by CPI deflation.
- CPI:
  - 1870–1913: Catão and Solomou (2005).
  - 1914–1917: Interpolated assuming relative PPP.
  - 1918–1940: Williamson.
  - 1941–1960: Mitchell.
  - 1961–2004: IFS.
- Export and Import volumes and Net Barter Terms of Trade:
  - 1870–1925: Catão (2005).
  - 1926–1940: Cardenas.
  - 1941–1960: INEGI.
  - 1961–2004: IFS.
- Real Effective Exchange Rate:
  - 1870–1913: Catão and Solomou (2005).
  - 1914–2004: CPI-based geometric weighted averages with six largest trading partners (covering over 90 percent of visible trade). Fisher ideal indices for 1914–1946 and 1946–2004, spliced at 1946. Nominal exchange rates source: Carmen M. and Kenneth S. Rogoff, 2004.
- Wages:
  - 1870–1940: Jeffrey G. Williamson, 1995.
  - 1940–1974: Mitchell.
  - 1974–2004: WEO and IFS databases.
- Foreign 3-month bill rate:
  - 1870–1920: London market yields (Holmer and Sillas, 1996).
  - 1921–2004: US 3-month Treasury Bill yields (Holmer and Sillas, 1996).
  - Deflation by respective CPI sources: Catão and Solomou (1870–1913); Mitchell (1914–1960); IFS (1961–2004).
- Foreign Output:
  - US real GDP in 1990 PPP constant dollars from Maddison (2003).
- Population at mid-year:
  - 1870–1960: Maddison (2003).
  - 1961–2004: IFS (2003).

### Appendix I — Data construction and sources: United States (selected series and coverage)
- Agricultural Output:
  - 1870–1949: NBER historical database (output of main crops).
  - 1950–2004: Chained quantity index of value added in agriculture, Bureau of Economic Analysis.
- Manufacturing Output:
  - 1870–1949: Balke and Gordon (1989).
  - 1950–2004: Chained quantity index of value added in manufacturing, Bureau of Economic Analysis.
- Transportation and Communications Output:
  - 1870–1949: Balke and Gordon (1989).
  - 1950–2004: Chained quantity index of value added in transportation and information industries, Bureau of Economic Analysis.
- Construction Activity:
  - 1870–1949: Kuznets' construction series spliced with Gottlieb/BLS counterpart (Balke and Gordon, 1989).
  - 1950–2004: Chained quantity index of value added in construction, Bureau of Economic Analysis.
- Machinery Investment:
  - 1870–1949: Kuznets' durable producers' output series (with 5-year moving average unraveled), data kindly provided by Christina Romer.
  - 1950–2004: Bureau of Economic Analysis.
- Central Government Expenditures and Revenues:
  - 1870–1958: Mitchell.
  - 1959–2004: IFS.
- Narrow (M0) and Broad Money (M2):
  - 1870–1970: Bureau of Economic Analysis, Historical Statistics of the United States.
  - 1970–2004: Haver Analytics.
- CPI:
  - 1870–1913: Catão and Solomou (2005).
  - 1914–1947: Balke and Gordon (1989).
  - 1948–2004: IFS.
- Export and Import volumes and Net Barter Terms of Trade:
  - 1870–1913: Catão and Solomou (2005).
  - 1914–1959: Bureau of Economic Analysis, Historical Statistics of the United States.
  - 1960–2004: IFS.
- Current Account Balance:
  - 1870–1959: Bureau of Economic Analysis, Historical Statistics of the United States (calculated by adding balance of goods and services (series U15) plus net private and government transfers (series U16 and U17)).
  - 1960–2004: IFS.
- Real Effective Exchange Rate:
  - 1870–1913: Catão and Solomou (2005).
  - 1914–2004: CPI-based geometric weighted averages of United States' real bilateral exchange rates with 17 largest trading partners (covering over no less than two thirds of visible trade before World War II and no less than 80 percent onwards). Fisher ideal indices for 1914–1946 and 1946–2004 (based on 1913 and 1938, and 1970 and 2000 weights respectively), spliced at 1946. Nominal exchange rates source: Carmen M. and Kenneth S. Rogoff, 2004.
- Wages:
  - 1870–1944: Jeffrey G. Williamson, 1995.
  - 1945–2004: Hourly compensation in manufacturing from IFS deflated by CPI from the same source.
- Foreign 10-year bill rate:
  - 1870–1953: Holmer and Sillas, 1996.
  - 1954–2004: IFS.
- Foreign Output:
  - Real GDP of Australia, Canada, UK, France, Germany, Italy and Japan in 1990 PPP constant dollars from Maddison (2003) extended with IFS through 2004.
- Population at mid-year:
  - 1870–1960: Maddison (2003).
  - 1961–2004: IFS (2003).

*Source: _wp0649 - REFERENCES (Appendix I: Data Construction and Sources).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0649.pdf_