Banking Crisis and Gravity Model Analysis

type

Post

status

Published

date

May 1, 2024

slug

dissertation

summary

My undergraduate dissertation

📝 Abstract

This paper examines the factors influencing bilateral international trade flows among RCEP member countries. Using bilateral trade panel data from 15 RCEP member countries between 1990 and 2022, the study conducts empirical analysis through the gravity model. The analysis builds upon Mou Yifei's 2016 research methodology, adopting similar model construction and data selection approaches to investigate the impact and lagged effects of banking crises during the Asian economic crisis on regional economies.

The primary aim is to analyze the specific relationship between bilateral trade and banking crises in RCEP countries by exploring whether stable patterns exist between banking crises and trade fluctuations. Empirical findings reveal that countries with banking crisis experience demonstrate higher trade volumes. This observation strengthens the established connection between banking crises and financial development levels, supporting the hypothesis that financial development significantly influences trading activities. Specifically, when one trading partner faces a banking crisis, their trade volume notably decreases in subsequent periods. However, if their trading partner later experiences a crisis, this second crisis does not substantially worsen existing trade disruptions. This suggests that banking crises do not create simple cumulative effects.

Through ongoing refinement of research methods and models, this study aims to provide valuable insights for trade policy development and economic planning in RCEP member countries.

KEYWORDS: fixed effects method, trade volume, banking crisis, RCEP member countries

Data Source

📎

For my undergraduate dissertation, I conducted an empirical study on bank crises by applying a gravity equation framework. The goal was to explore how international trade factors and distance influence the risk of bank crises, focusing specifically on countries within the Regional Comprehensive Economic Partnership (RCEP).

I chose this topic for two main reasons. First, I wanted to apply a quantitative method to a real-world issue that combines both economic theory and financial stability—a direction that aligns with my growing interest in data-driven research. Second, I saw a research gap: although the gravity model is widely used in trade studies, it is rarely applied to financial crisis prediction, especially in the context of newer trade blocs like RCEP. By exploring the relationship between trade connectivity and systemic risk, I hoped to reveal insights relevant to both international economic integration and financial policy design.

The first major challenge was collecting and combining data from three separate sources, each with different formats and variables. I compiled these datasets in Excel, created new variables such as country IDs and dummy variables to standardize the information, and converted numeric indicators into categorical variables to simplify interpretation in the model. Data cleaning turned out to be the most time-consuming part, as I had to ensure consistency across multiple layers of data, including country-level attributes, trade figures, and geographical distance.

After preparing the dataset, I applied the gravity equation model, which is commonly used in global datasets but had rarely been applied to RCEP countries with more recent data beyond 2017. Although traditionally used to predict trade volume, I adapted the model to examine financial contagion risk and crisis likelihood, treating trade and proximity as transmission channels. The analysis involved multi-layer regression models, which were technically challenging due to the complex interactions between variables. However, with the guidance of my tutor and through regular updates, I was able to refine the model and successfully generate meaningful results. In the end, the findings were more insightful than I initially expected, offering valuable perspectives on trade dynamics and financial stability within the selected countries.

This dissertation greatly improved my data handling and analytical skills. I learned how to merge complex datasets into a unified structure, manage data cleaning for multi-source information, and apply multi-level regression techniques. While the modeling was difficult at first, it taught me how to work with complex statistical methods and interpret results carefully. Moreover, I gained experience in independent research, learned when to seek help, and improved my ability to communicate progress with my supervisor.