## _wp0674

## Source details

**Canonical URL:** [_wp0674](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0674.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0674.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0674.pdf.json)

---

### Introduction and objectives
- Main goal: disentangle the effects of oil prices from other factors underlying Venezuela’s equilibrium real exchange rate (REER) to assess the appropriateness of the official peg and likely consequences of liberalization.
- Secondary goal: examine the role of foreign exchange controls and parallel market exchange rates by reestimating the VEC model with parallel market REERs.
- Approach: estimate a vector error correction model (VEC) using annual data (1950–2004) on REER and determinants, perform robustness tests, and compare results using official and parallel market rates.

### Framework and data
- Equilibrium REER modeled as time-varying and driven by:
  - LOIL: U.K. Brent spot price deflated by the unit value of manufacturing exports (MUV), in logs.
  - LRGDPC: PPP-based real GDP per capita differential (Venezuela vs. trading partners), in logs.
  - GEXP: central government expenditure as a share of GDP.
  - IRATE: real interest rate differentials vis-à-vis trading partners.
- VEC specification: REER = f(loil, lrgdpc, gexp, irate).
- Data:
  - Annual series for 1950–2004.
  - CPI-based REER from IMF IFS for 1980–2004, extended back to 1950 with constant trade weights.
  - Parallel market REER constructed by adjusting official REER with exchange rate premia (Reinhart and Rogoff, 2002).
  - Trade weights (largest weights accounting for 88 percent of trade): United States 37 percent; Germany 10 percent; Italy 8 percent; Japan 8 percent; France 6 percent; United Kingdom 5 percent; Brazil 4 percent; Canada 4 percent; Spain 3 percent; Belgium 3 percent.
  - Series smoothed for visualization using a Hodrick-Prescott (HP) filter (suffix HP denotes filtered series).

### Stylized data patterns
- Official vs. parallel REER:
  - Many instances of sharp differences between official and parallel market rates and an increase in volatility since 1980.
- Interest rate differentials:
  - Increased volatility and sharp swings since the mid-1980s.
- Oil prices:
  - Sharp variation since the 1970s.
- Government expenditure:
  - Mirrors oil price developments to some degree; pro-cyclical behavior noted in the 1970s and 1990s.
- Productivity differentials:
  - Venezuela’s income per capita was about 50 percent of U.S. per capita income in the early 1950s; this has declined to 12½ percent.
  - Per capita growth rates since the 1970s: (i) 2 percent annual decline in 1970-89; (ii) 1 percent annual decline during the 1990s; and (iii) ¼ percent annual decline in 2000-04 (negative per capita growth since the 1970s).

### Unit root tests (Augmented Dickey-Fuller) — main result
- Main macroeconomic series:
  - Nonstationary in levels (have unit roots).
  - Stationary in first differences.
- Table 1. Augmented Dickey Fuller Unit Root Tests (1950–2004; annual data):
  - LREER OFF MKT (REER official market rate; in logs): level t-ADF = -1.16 ; first difference t-ADF = -6.71 ***
  - LREER PAR MKT (REER parallel market rate; in logs): level t-ADF = -1.55 ; first difference t-ADF = -6.04 ***
  - LOIL (Real oil prices; in logs): level t-ADF = -1.24 ; first difference t-ADF = -7.55 ***
  - LRGDPC (Real GDP per capita differentials; in logs): level t-ADF = 0.64 ; first difference t-ADF = -5.63 ***
  - GEXP (Government expenditures as a share of GDP): level t-ADF = -2.27 ; first difference t-ADF = -6.99 ***
  - IRATE (Real interest rate differentials): level t-ADF = -2.80 * ; first difference t-ADF = -7.20 ***
- Notes: * and *** denote rejection at, respectively, 10 and 1 percent level. Lags chosen using Schwarz criterion.

### VEC estimation results and interpretation (official exchange rates)
- Cointegration:
  - Trace and maximum eigenvalue statistics confirm the existence of at most one cointegrating vector between REER and its determinants.
  - Trace statistic: 98.7 *** 75.1 ***
  - Maximum eigenvalue statistic: 59.4 *** 54.0 ***
- Estimates of the cointegrating relationship (official market):
  - Dependent variable: LREER OFF MKT (-1)
  - LOIL (-1) -1.298  ***  -1.022  ***
    - t-statistics: (-3.80) (-4.26)
  - LRGDPC (-1) -1.572  ***  -1.407  ***
    - t-statistics: (-5.34) (-5.19)
  - GEXP (-1) 0.040
    - t-statistic: (0.68)
  - IRATE (-1) -0.144  ***  -0.138  ***
    - t-statistics: (-8.85) (-8.76)
  - C 0.613 0.233
- Long-run elasticities and effects (text summary):
  - A 1 percent increase in oil prices → a 1¼ percent appreciation in the REER.
  - A 1 percent worsening in Venezuela’s productivity differential → a 1½ percent depreciation effect on the REER.
  - A 1 percent increase in real interest rate differentials → real appreciation pressures of just over 10 percent.
- Speed of adjustment and half-life (official market):
  - Estimates of the speed of adjustment: 0.049 *** 0.054 ***
    - t-statistics: (3.80) (2.68)
  - Time required to adjust by 50 percent any deviation from the equilibrium REER: 13.8 12.5
  - Interpretation: convergence is slow — it takes a decade to reduce by half any deviation between the official and the equilibrium rate, implying persistent misalignment is possible.

### Parallel market VEC model and results
- Motivation: official exchange rate shows slow adjustment; Venezuela exhibited prolonged large differences between official and parallel market exchange rates.
- Number of cointegrating vectors and normality:
  - Trace statistic 86.5 ***
  - Maximum eigenvalue statistic 60.2 ***
  - Normality Test: Skewness dfProb.40.018; Kurtosis dfProb.40.001; Normality 80.000
- Estimates of the cointegrating relationship (parallel market):
  - Dependent variable: LREER PARALLEL MKT (-1)
  - LOIL (-1) -0.442 ***
    - t-statistic: (-3.95)
  - LRGDPC (-1) -1.159 ***
    - t-statistic: (-7.85)
  - IRATE (-1) -0.077 ***
    - t-statistic: (-13.54)
  - C -2.059
- VEC lag exclusion Wald tests:
  - Dlag1 Statistic 58.0 Prob. 0.000
  - Dlag2 Statistic 67.1 Prob. 0.000
  - Dlag3 Statistic 49.6 Prob. 0.001
  - df 25
- Speed of adjustment and half-life (parallel market):
  - Estimate of the speed of adjustment: 0.243 **
    - t-statistic: (2.31)
  - Time required to adjust by 50 percent any deviation from the equilibrium REER: 2.5
  - Interpretation: much faster adjustment than official-rate model (about 2.5 years versus over a decade).

### Interpretation of parallel-market results and comparison
- Coefficient magnitudes are smaller than in the official-rate model but qualitative conclusions remain:
  - Oil price increases cause appreciation of the equilibrium exchange rate (by ½ percent for every 1 percent rise in oil prices).
  - Increases in productivity and real interest rate differentials are associated with appreciation pressures (slightly smaller than in the official-rate model).
- Parallel-market equilibrium series shows less abrupt swings in response to oil price changes but still indicates a steady decline in REER driven by deteriorating productivity differentials.
- Equilibrium vs. actual REER and valuation assessment:
  - Both parallel and official market equilibrium rates at end-2004 lie above their corresponding actual rates, implying some under-valuation at end-2004.
  - The parallel market and equilibrium rates are a closer match than official and equilibrium rates, reflecting quicker adjustment in the parallel market.
  - The parallel market rate is 15 percent more appreciated than its equilibrium rate (but still below the official rate).
  - Developments since 2004:
    - Non-oil activity remained buoyant and oil prices were 42 percent higher, on average, than in 2004.
    - Real appreciation pressures likely built up further during 2005.
    - The parallel market rate at end-2005 has a smaller premium than in 2004.

### Policy implications and conclusions
- Venezuela has a time-varying equilibrium real exchange rate; oil price increases generate appreciation pressures and productivity differential declines exert depreciating pressures (a reverse Balassa-Samuelson effect).
- VEC modeling is more informative when applied to parallel market exchange rates because of faster adjustment.
- The slow adjustment of official rates suggests that foreign exchange restrictions enabled persistent departures from equilibrium.
- If higher oil prices are persistent, the current peg regime may:
  - Need reassessment because oil-driven appreciation pressures are a long-run determinant of equilibrium REER.
  - Rely excessively on monetary policy to restrain inflationary pressures and may ultimately fail to prevent real appreciation under a continued peg.

*Source: IMF staff analysis as presented in the provided content unit.*

### 1. Augmented Dickey Fuller Unit Root Tests .........................................................................10

### 1. Augmented Dickey Fuller Unit Root Tests

### Introduction and objectives
- Main goal: disentangle the effects of oil prices from other factors underlying Venezuela’s equilibrium real exchange rate (REER) to assess the appropriateness of the official peg and likely consequences of liberalization.
- Secondary goal: examine the role of foreign exchange controls and parallel market exchange rates by reestimating the VEC model with parallel market REERs.
- Approach: estimate a vector error correction model (VEC) using annual data (1950–2004) on REER and determinants, perform robustness tests, and compare results using official and parallel market rates.

### Framework: determinants of a time-varying equilibrium REER
- The equilibrium REER is modeled as a time-varying path influenced by:
  - Oil prices (loil): U.K. Brent spot price deflated by the unit value of manufacturing exports (MUV), in logs.
  - Productivity differentials (lrgdpc): PPP-based real GDP per capita differential (Venezuela vs. trading partners), in logs.
  - Macro-policy variables: central government expenditure as a share of GDP (gexp), and real interest rate differentials (irate) vis-à-vis trading partners.
- VEC model specification: REER = f(loil, lrgdpc, gexp, irate)
- Data:
  - Annual series for 1950–2004.
  - CPI-based REER from IMF IFS for 1980–2004, extended back to 1950 with constant trade weights.
  - Parallel market REER constructed by adjusting official REER with exchange rate premia (Reinhart and Rogoff, 2002).
  - Trade weights used for REER and for constructing differential series (largest weights accounting for 88 percent of trade: United States 37 percent; Germany 10 percent; Italy 8 percent; Japan 8 percent; France 6 percent; United Kingdom 5 percent; Brazil 4 percent; Canada 4 percent; Spain 3 percent; Belgium 3 percent).
  - Series were smoothed using a Hodrick-Prescott (HP) filter for visualization (suffix HP denotes filtered series).

### Stylized data patterns highlighted in the paper
- Official vs. parallel REER:
  - Many instances of sharp differences between official and parallel market rates and an increase in volatility since 1980.
- Interest rate differentials:
  - Increased volatility and sharp swings since the mid-1980s.
- Oil prices:
  - Sharp variation since the 1970s.
- Government expenditure:
  - Mirrors oil price developments to some degree; pro-cyclical behavior noted in the 1970s and 1990s.
- Productivity differentials:
  - Striking and steady decline: Venezuela’s income per capita was about 50 percent of U.S. per capita income in the early 1950s; this has declined to 12½ percent.
  - Per capita growth rates since the 1970s: (i) 2 percent annual decline in 1970-89; (ii) 1 percent annual decline during the 1990s; and (iii) ¼ percent annual decline in 2000-04 (negative per capita growth since the 1970s).

### Unit root tests (Augmented Dickey-Fuller) — main result
- The main macroeconomic series are:
  - Nonstationary in levels (have unit roots)
  - Stationary in first differences
- Table 1. Augmented Dickey Fuller Unit Root Tests (1950–2004; annual data)
  - LREER OFF MKT (REER official market rate; in logs): level t-ADF = -1.16 ; first difference t-ADF = -6.71 ***
  - LREER PAR MKT (REER parallel market rate; in logs): level t-ADF = -1.55 ; first difference t-ADF = -6.04 ***
  - LOIL (Real oil prices; in logs): level t-ADF = -1.24 ; first difference t-ADF = -7.55 ***
  - LRGDPC (Real GDP per capita differentials; in logs): level t-ADF = 0.64 ; first difference t-ADF = -5.63 ***
  - GEXP (Government expenditures as a share of GDP): level t-ADF = -2.27 ; first difference t-ADF = -6.99 ***
  - IRATE (Real interest rate differentials): level t-ADF = -2.80 * ; first difference t-ADF = -7.20 ***
  - Notes: * and *** denote rejection at, respectively, 10 and 1 percent level. Lags chosen using Schwarz criterion.

### VEC estimation results and interpretation (official exchange rates)
- Cointegration:
  - Trace and maximum eigenvalue statistics confirm the existence of at most one cointegrating vector between REER and its determinants.
- Regression highlights:
  - Regression 1 (includes gexp): coefficients indicate REER appreciates as oil prices increase; productivity differentials have a reverse Balassa-Samuelson effect (steady worsening in productivity differentials causing depreciation pressure); increases in real interest rate differentials are linked to appreciation pressures; government spending is inversely related to the REER but not statistically significant.
  - Regression 2 (excludes gexp): provides estimated elasticities summarized below.
- Estimated elasticities and effects (from VEC estimates discussed in text):
  - A 1 percent increase in oil prices → a 1¼ percent appreciation in the REER (almost one-to-one effect).
  - A 1 percent worsening in Venezuela’s productivity differential → a 1½ percent depreciation effect on the REER.
  - A 1 percent increase in real interest rate differentials → real appreciation pressures of just over 10 percent.
- Speed of adjustment:
  - The speed of convergence of the official REER to its equilibrium is slow: it takes a decade to reduce by half any deviation between the official and the equilibrium rate.
  - Implication: the official rate can remain substantially and persistently misaligned with the equilibrium REER.

### Robustness and comparison with parallel market rates (summary of implications)
- Oil prices are a significant determinant of a time-varying equilibrium REER in Venezuela.
- Productivity declines are identified as a key factor behind the long-term depreciating trend in the REER.
- Appreciation pressures were building up, intensified by oil price increases in 2005 (text notes these pressures).
- When the VEC model is estimated using parallel market exchange rates:
  - The speed of convergence of the REER to equilibrium is higher compared with the official-rate-based VEC estimates.
  - Interpretation: the government’s foreign exchange controls and the state’s dominant role in oil export earnings have allowed the official rate to remain sharply deviated from the equilibrium rate.

*Source: IMF staff analysis as presented in the provided content unit.*

### 1. Number of cointegrating vectors11

### _wp0674 - 1. Number of cointegrating vectors11

### Cointegration test results
- Trace statistic98.7 ***75.1 ***
- Maximum eigenvalue statistic59.4 ***54.0 ***

### Estimates of the cointegrating relationship (official market)
- Dependent variable: LREER OFF MKT (-1)
- LOIL (-1) -1.298  *** -1.022  ***
  - t-statistics: (-3.80) (-4.26)
- LRGDPC (-1) -1.572  *** -1.407  ***
  - t-statistics: (-5.34) (-5.19)
- GEXP (-1) 0.040
  - t-statistic: (0.68)
- IRATE (-1) -0.144  *** -0.138  ***
  - t-statistics: (-8.85) (-8.76)
- C 0.613 0.233

### Speed of adjustment and half-life (official market)
- Estimates of the speed of adjustment based on the cointegrating equation for the REER:
  - 0.049 *** 0.054 ***
  - t-statistics: (3.80) (2.68)
- Time required to adjust by 50 percent any deviation from the equilibrium REER:
  - 13.8 12.5
- Note: 1/ t-statistics between parethesis.

### Interpretation of long-run relationship (official market)
- The equilibrium real exchange rate (REER) is calculated by applying cointegrating coefficients to determinants using HP-filtered series to remove short-run fluctuations.
- Two conclusions:
  - The deterioration in productivity differentials is an important factor explaining the downward trend in Venezuela’s REER.
  - The downward trend was interrupted twice contemporaneously with sharp changes in oil prices (mid-1970s and mid-1990s), where oil price increases produced temporary upward trends in the equilibrium REER.

### Parallel market VEC model (adjusted REER)
- Motivation: official exchange rate shows slow adjustment; Venezuela exhibited prolonged large differences between official and parallel market exchange rates.
- Table 3 (history of exchange rate regimes) documents parallel market premium values across periods (examples preserved in source).

### Results from VEC Model using Parallel Market Exchange Rates (Table 4)
1. Number of cointegrating vectors and normality
- Trace statistic86.5 ***
- Maximum eigenvalue statistic60.2 ***
- Normality Test: Skewness dfProb.40.018; Kurtosis dfProb.40.001; Normality80.000

2. Estimates of the cointegrating relationship (parallel market)
- Dependent variable: LREER PARALLEL MKT (-1)
- LOIL (-1) -0.442 ***
  - t-statistic: (-3.95)
- LRGDPC (-1) -1.159 ***
  - t-statistic: (-7.85)
- IRATE (-1) -0.077 ***
  - t-statistic: (-13.54)
- C -2.059
- VEC lag exclusion Wald tests:
  - Dlag1 Statistic58.0 Prob.0.000
  - Dlag2 Statistic67.1 Prob.0.000
  - Dlag3 Statistic49.6 Prob.0.001
  - df25

3. Speed of adjustment and half-life (parallel market)
- Estimates of the speed of adjustment based on the cointegrating equation for the REER:
  - 0.243 **
  - t-statistic: (2.31)
- Time required to adjust by 50 percent any deviation from the equilibrium REER that may exist:
  - 2.5
- Note: 1/ t-statistics between parethesis.

### Interpretation of parallel-market results
- Coefficient estimates are smaller than in the official-rate model, but qualitative conclusions remain:
  - Oil price increases cause appreciation of the equilibrium exchange rate (by ½ percent for every 1 percent rise in oil prices).
  - Increases in productivity and real interest rate differentials are associated with appreciation pressures (slightly smaller than in the official-rate model).
- The speed of adjustment is higher in the parallel market model: about 2.5 years compared to over a decade for official rates.
- The parallel-market equilibrium series shows less abrupt swings in response to oil price changes but still indicates a steady decline in REER driven by deteriorating productivity differentials.

### Equilibrium vs. actual REER and valuation assessment
- Both parallel and official market equilibrium rates at end-2004 lie above their corresponding actual rates, implying some under-valuation at end-2004.
- The parallel market and equilibrium rates are a closer match than official and equilibrium rates, reflecting quicker adjustment in the parallel market.
- The parallel market results suggest:
  - The parallel market rate is 15 percent more appreciated than its equilibrium rate (but still below the official rate).
  - Speed of adjustment to equilibrium is much higher in the parallel market.
- Developments since 2004:
  - Non-oil activity remained buoyant and oil prices were 42 percent higher, on average, than in 2004.
  - Real appreciation pressures likely built up further during 2005.
  - The parallel market rate at end-2005 has a smaller premium than in 2004.

### Policy implications and conclusions
- Venezuela has a time-varying equilibrium real exchange rate; oil price increases generate appreciation pressures and productivity differential declines exert depreciating pressures (a reverse Balassa-Samuelson effect).
- VEC modeling is more informative when applied to parallel market exchange rates because of faster adjustment.
- The slow adjustment of official rates suggests that foreign exchange restrictions enabled persistent departures from equilibrium.
- If higher oil prices are persistent, the current peg regime may:
  - Need reassessment because oil-driven appreciation pressures are a long-run determinant of equilibrium REER.
  - Rely excessively on monetary policy to restrain inflationary pressures and may ultimately fail to prevent real appreciation under a continued peg.

*Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0674.pdf*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0674.pdf_