亚开行-2021年亚洲中小企业监测：第四卷中小企业发展试点指数：应用概率主成分分析（英）-2022.4-44页_1mb

Summary of ASIA SMALL AND MEDIUM-SIZED ENTERPRISE MONITOR 2021, Volume IV

Core Content

The Asia Small and Medium-Sized Enterprise Monitor (ASM) 2021, Volume IV presents a new pilot of the Small and Medium-Sized Enterprise Development Index (SME-DI). This index is designed to measure the development of MSMEs across Asia using a multivariate analytical approach based on real country data. The report introduces a probabilistic principal component analysis (PCA) model to address data limitations and improve the accuracy of the SME-DI.

The study focuses on both regional (Southeast Asia and South Asia) and country-level (Viet Nam) MSME development. It emphasizes the importance of finance in MSME growth and recovery, especially after the 2008-2009 global financial crisis (GFC). The SME-DI aims to support evidence-based policymaking and access to finance for MSMEs.

Main Points

• SME-DI Purpose: To measure MSME development across Asia using real data and advanced econometrics.
• Data Challenges: MSME data is often sparse, incomplete, and not comparable across countries.
• Methodology: A probabilistic PCA model is introduced to handle missing data and provide a more accurate index.
• Regional Analysis:
  • From 2009 to 2020, the GFC slowed MSME development in Southeast Asia and South Asia.
  • A recovery began between 2013 and 2015 with improvements in capital markets and MSME financing.
• Country Analysis (Viet Nam):
  • MSMEs showed continuous sales growth in manufacturing, trade, and transport.
  • Employment increased in trade and agri-food services.
  • Profitability improved briefly in construction before a slight decline.
  • New infrastructure needs and a national procurement law supported MSME development.
• Key Findings:
  • Equity finance was critical for recovery after the GFC.
  • Bank credit and nonbank finance supported further development once the sector recovered.
  • Finance remains central to MSME growth and resilience.
• Future Steps:
  • Expand data collection, especially granular firm-level data.
  • Conduct test runs across different countries and levels.
  • Use the SME-DI to predict important outcomes, such as MSME contributions to GDP.

Key Information

• SME-DI Construction: Based on real data from the ADB 2021 ASM database.
• Probabilistic PCA: A method that allows for missing data and provides a more robust index.
• Data Sources:
  • Aggregate MSME data for 15 countries in Asia.
  • Firm-level data from Viet Nam's Agency for Business Registration (ABR).
• Challenges:
  • Data limitations and incomparability across countries.
  • Missing data entries and short time series.
• Collaboration: ADB and UTokyo Economic Consulting Inc. (UTEcon) worked together to develop the index.
• License: The document is available under the Creative Commons Attribution 3.0 IGO license.

Methodology

• Probabilistic PCA Model: Assumes that observed data corresponds to a latent variable and is generated by a linear model.
• Expectation-Maximization (EM) Algorithm: Used to estimate parameters iteratively, ensuring the likelihood does not decrease in each step.
• Regularization: Data is normalized to mean 0 and variance 1 for each variable before analysis.
• Results Interpretation: The model produces principal components that reflect MSME development across various dimensions.

Conclusion

The SME-DI is an important tool for evaluating MSME development in Asia. While the index is still in development, the probabilistic PCA approach provides a more robust method for dealing with data limitations. The study highlights the role of finance in MSME recovery and growth, and emphasizes the need for more comprehensive and comparable data to improve the accuracy and usefulness of the index. Future work will focus on expanding the dataset, refining the model, and using the SME-DI for policy prediction.