2017年-IMF国际货币组织全球_A_Toolkit_to_Assess_the_Consistency_Between_Real_Sector_and_Financial_Sector_Forecasts_30页_2mb

Summary of "A Toolkit to Assess the Consistency Between Real Sector and Financial Sector Forecasts"

Core Content

This document presents a toolkit designed to assess the consistency between real sector and financial sector forecasts. The toolkit is based on empirical analysis of macroeconomic and financial data from 182 economies over the period 1980–2015. It aims to improve the alignment of forecasts across sectors by leveraging historical patterns and empirical distributions of real and financial variables.

Main Points

• Purpose: To develop a method for evaluating the consistency between real sector (e.g., GDP, consumption, investment, employment) and financial sector (e.g., credit growth, house prices, financial account balance) forecasts.
• Empirical Basis: The toolkit is grounded in empirical regularities observed across different country groups (advanced economies, emerging markets, low-income countries) and regions.
• Key Variables:
  • Credit Growth: Strongly correlated with real sector performance, especially during extreme credit booms or busts.
  • House Prices: Also correlated with real sector outcomes, with higher growth associated with better real performance.
  • Output Gap: Used as a second conditioning variable to capture the cyclical state of the economy.
  • Financial Account Balance: Not found to significantly affect real sector outcomes when included as a second variable.

Key Findings

• Credit and Real Sector Correlation: Credit growth is positively correlated with real sector performance. During credit booms, real GDP, consumption, and investment growth are typically higher, while during busts, they are lower.
• Inflation and Credit: The relationship between credit growth and inflation is weak. Inflation remains relatively stable regardless of credit conditions.
• Country Group Differences:
  • Advanced Economies: Show less variability in real and financial variables compared to emerging markets and low-income countries.
  • Emerging Markets and Low-Income Countries: Exhibit greater dispersion in real and financial outcomes, especially in investment and consumption growth.
• Extreme Outcomes: The toolkit flags forecasts that are inconsistent with historical patterns, highlighting the possibility of extreme joint outcomes (e.g., high real activity growth during large credit contractions).
• Model-Free Approach: The toolkit uses empirical distributions rather than model-based approaches, which can be more prone to misspecification. This makes it more flexible and applicable to a wide range of economies with limited data.

Methodology

• Data Sources: Macro and financial data are sourced from the IMF's IFS and WEO databases, as well as the OECD and BIS.
• Variables Used:
  • Real variables: GDP, consumption, investment, employment, and inflation.
  • Financial variables: Credit growth, house prices, and financial account balance.
• Approach:
  • The toolkit estimates the inverse cumulative distribution function (CDF) for real variables conditional on financial variables and the output gap.
  • It provides four threshold values (10th, 25th, 75th, and 90th percentiles) for each conditional CDF to signal deviations from historical norms.
  • It allows users to check the consistency of forecasts against these historical thresholds.

Flexibility and Application

• Customization: The toolkit is flexible and can be adapted to different country groups based on income level, region, and export earnings sources.
• Use Case: It is intended to complement traditional model-based approaches (e.g., DSGE models) and be used in conjunction with country-specific economic narratives to ensure consistency with macroeconomic frameworks.

Limitations

• Narrative Gap: The toolkit does not provide a micro-founded narrative of real-financial linkages, unlike model-based approaches.
• No Impact from Financial Account Balance: Including the financial account balance as a second variable does not improve the understanding of real sector outcomes, suggesting it may not be as important as other financial indicators.

Conclusion

The toolkit offers a practical and accessible method for assessing the consistency of real and financial forecasts, particularly useful for economies with limited data. It highlights the importance of credit and house prices in shaping real sector outcomes and provides a way to flag forecasts that deviate significantly from historical patterns. The results suggest that while credit growth is a strong indicator of real sector performance, inflation is not strongly influenced by credit conditions. The toolkit is recommended for use alongside macroeconomic narratives to ensure that forecasts are consistent with the broader economic context.