2002年-世界发展银行全球_Demand_for_Imports_in_Venezuela______________A_Structural_Time_Series_Approach_20页_850kb

Summary of "Demand for Imports in Venezuela: A Structural Time Series Approach"

Core Content

This working paper by Mario A. Cuevas analyzes the demand for imports in Venezuela from 1974 to 2000 using structural time series models. The paper decomposes real GDP and real imports into stochastic trends and stochastic cycles, and investigates the elasticity of imports with respect to GDP and the real exchange rate.

Main Findings

• Stochastic Trends:

  • The real imports trend drifts upward at 2.8% annually, which is almost twice the rate of GDP drift (1.6% annually).
  • The trend component of real imports is exogenously influenced by the real exchange rate, which is treated as an external variable.

• Stochastic Cycles:

  • Two sets of independent stochastic cycles are identified: a higher frequency cycle with a 5-year period and a lower frequency cycle with a 17-year period.
  • The amplitude of the higher frequency cycle is 0.11 for imports and 0.03 for GDP, while the lower frequency cycle has amplitudes of 0.08 for imports and 0.03 for GDP.
  • The frequency profile of import elasticities shows that elasticity increases with frequency, with the highest value of 4.55 at the 5-year cycle and a lower value of 1.71 at the zero frequency.

• Elasticities:

  • A 1% real exchange rate appreciation leads to a 0.4% increase in imports.
  • In the long run, a 1% real GDP growth is associated with a 1.7% real imports growth.
  • The GDP elasticity of imports is higher in the short run (higher frequency) than in the long run, suggesting that external imbalances are recurrently linked to short-run cyclical movements.

• Model Performance:

  • The model has 13 parameters and shows strong convergence after 100 iterations.
  • The goodness of fit is confirmed by R²D and ordinary R² statistics, with R²D = 0.60 for imports and R²D = 0.50 for GDP.
  • The residuals of the model display reasonable statistical properties, with low heteroskedasticity and autocorrelation, and spectral densities resembling a white noise series.

Key Information

• Methodology:

  • The paper uses a structural time series approach to decompose GDP and imports into trends and cycles.
  • The real exchange rate is treated as an exogenous variable with a direct impact on the system.
  • The drift components of both GDP and imports are fixed.

• Policy Implications:

  • The powerful structural tendency towards increasing imports in Venezuela is emphasized.
  • Short-run import responsiveness to GDP is significantly higher than in the long run, especially at the 5-year cycle.
  • The VECM framework is compared to the structural time series approach, with the former showing good performance in long-run relationships but poor performance in capturing short-run dynamics.

• Data and Series:

  • The data spans 1974 to 2000 and includes real GDP, real imports, and real exchange rate.
  • The real exchange rate is calculated as a trade-weighted index (1990 = 100), based on IMF data.

Structure and Components

• The log of real GDP is represented as:
  $$
  y_t = \mu_t^y + \psi_t^y + \xi_t^y + \varepsilon_t
  $$
• The log of real imports is represented as:
  $$
  m_t = \mu_t^m + \psi_t^m + \xi_t^m + r_t + \eta_t
  $$
• The common stochastic trends and independent stochastic cycles are estimated with cross-equation restrictions.

Conclusion

The study confirms a strong structural tendency in Venezuela towards increasing imports, even at slow GDP growth rates. The import demand is highly responsive to short-run GDP fluctuations, especially those associated with 5-year cycles, which could contribute to recurrent external imbalances. The structural time series model is shown to be more effective in capturing cyclical patterns compared to the VECM framework.

Policy Research Working Paper Series

• The paper is part of a series on macroeconomic issues in the Latin America and the Caribbean region.
• Copies are available free from the World Bank.
• The author can be contacted at mcuevas@worldbank.org.