Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Beyond the Numbers: Causal Effects of Financial Report Sentiment on Bank Profitability

Krishna Neupane, Prem Sapkota, Ujjwal Prajapati

Abstract

This study establishes the causal effects of market sentiment on firm profitability, moving beyond traditional correlational analyses. It leverages a causal forest machine learning methodology to control for numerous confounding variables, enabling systematic analysis of heterogeneity and non-linearities often overlooked. A key innovation is the use of a pre-trained FinancialBERT to generate sentiment scores from quarterly reports, which are then treated as causal interventions impacting profitability dynamics like returns and volatilities. Utilizing a comprehensive dataset from NEPSE, NRB, and individual financial institutions, the research employs SHAP analysis to identify influential profit predictors. A two-pronged causal analysis further explores how sentiment's impact is conditioned by Loan Portfolio/Asset Composition and Balance Sheet Strength/Leverage. Average Treatment Effect analyses, combined with SHAP insights, reveal statistically significant causal associations between certain balance sheet and expense management variables and profitability. This advanced causal machine learning framework significantly extends existing literature, providing a more robust understanding of how financial sentiment truly impacts firm performance.

Beyond the Numbers: Causal Effects of Financial Report Sentiment on Bank Profitability

Abstract

This study establishes the causal effects of market sentiment on firm profitability, moving beyond traditional correlational analyses. It leverages a causal forest machine learning methodology to control for numerous confounding variables, enabling systematic analysis of heterogeneity and non-linearities often overlooked. A key innovation is the use of a pre-trained FinancialBERT to generate sentiment scores from quarterly reports, which are then treated as causal interventions impacting profitability dynamics like returns and volatilities. Utilizing a comprehensive dataset from NEPSE, NRB, and individual financial institutions, the research employs SHAP analysis to identify influential profit predictors. A two-pronged causal analysis further explores how sentiment's impact is conditioned by Loan Portfolio/Asset Composition and Balance Sheet Strength/Leverage. Average Treatment Effect analyses, combined with SHAP insights, reveal statistically significant causal associations between certain balance sheet and expense management variables and profitability. This advanced causal machine learning framework significantly extends existing literature, providing a more robust understanding of how financial sentiment truly impacts firm performance.
Paper Structure (17 sections, 6 equations, 5 figures)

This paper contains 17 sections, 6 equations, 5 figures.

Table of Contents

  1. Introduction
  2. Method
  3. extreme Gradient Boosting (XGBoost)
  4. Feature Importance
  5. Addressing the the Causal Question with Causal Forests
  6. The Data
  7. Control Covariates (X)
  8. Outcome Covariate (Y)
  9. Treatment Covariate (T)
  10. Analysis
  11. Correlation Analysis: Heatmap and Dendrogram Insights
  12. SHAP Analysis for Feature Importance
  13. Estimation of Causal Effects
  14. Loan Portfolio / Asset Composition
  15. Balance Sheet Strength / Leverage
  16. ...and 2 more sections

Figures (5)

  • Figure 1: Heatmap: Correlation matrix showing the pairwise correlation coefficients for all features. The color scale represents the correlation value, with dark purple indicating a perfect positive correlation (+1) and dark orange indicating a perfect negative correlation (-1)
  • Figure 2: Dendogram: Hierarchical clustering dendrogram of financial features. The clustering, performed using Ward's linkage and a distance based on absolute correlation, visualizes the relationships and underlying dimensions among the features.
  • Figure 3: SHAP Plot: SHAP beeswarm plot for the XGBoostRegressor model predicting net profit. Displays feature importance and influence on prediction, with color indicating feature value (low (blue) to high (red)).
  • Figure 4: Average Treatment Effect plot for features in Loan Portfolio / Asset Composition Analysis. The plot shows the estimated ATE and 95 percent confidence intervals. Features are ranked as in the SHAP plot. Asterisks indicate statistical significance (* p < 0.05, ** p < 0.01 ,*** p < 0.001 ).
  • Figure 5: Average Treatment Effect plot for features in Balance Sheet Strength / Leverage Analysis. The plot shows the estimated ATE and 95 percent confidence intervals. Features are ranked as in the SHAP plot. Asterisks indicate statistical significance (* p < 0.05, ** p < 0.01 ,*** p < 0.001)