A low R-squared (1.3%) in your pooled effects model could be due to several reasons:

1. Omitted Variable Bias:

• Important variables influencing investment decisions may be missing. Consider including additional variables such as:

• Growth Opportunities (e.g., Market-to-Book Ratio)

• Macroeconomic Factors (e.g., GDP growth, interest rates)

• Industry-Specific Effects to account for sectoral differences in investment behavior.

2. Model Specification Issues:

• The relationship between financial leverage and investment may not be linear. Try:

• Non-linear Models: Logarithmic transformations or quadratic terms.

• Interaction Terms: For example, leverage × profitability to explore conditional effects.

3. Dynamic Effects and Endogeneity:

• Investment decisions are often influenced by past performance. Consider:

• Dynamic Panel Models (e.g., GMM) to incorporate lagged investment or leverage variables.

• Instrumental Variable (IV) Approach to address potential endogeneity issues, especially if leverage is influenced by past investment.

4. Unobserved Heterogeneity:

• If firms have unique characteristics that are not captured, the pooled model may not account for this. Fixed or random effects models are better suited for panel data as they control for unobserved heterogeneity.

5. Measurement Error or Data Quality:

• Ensure the accuracy and consistency of your data, particularly for leverage and investment measurements. Consider using alternative proxies or deflating variables to control for inflation.

6. Model Comparison and Robustness Checks:

• Compare the R-squared with fixed and random effects models, as these often provide better explanatory power in panel data.

• Conduct robustness checks by experimenting with different model specifications and variable transformations.