Fossil Fuel Policies and Firm Innovations: Evidence from a Developing Countries Database

Presented at: The 10th International Conference on Sustainable Urban Development (ICSUD), October 03, 2024

Authors: Dr. Le Van Ha - Supervisor, Le Dang Trung Duc - Speaker/Author, Nguyen Quoc Minh - Co-Author

Institution: Vietnamese-German University (VGU) - Global Finance and Economics (GFE)

Conference Evidence & Media:

• ICSUD General Information (Website)

• ICSUD Event Announcement (Facebook Post)

• ICSUD Event Announcement (Facebook Post)

(Note: The detailed speaker schedule page is no longer maintained by the organizers, but the presentation slides and related materials are available in this repository).

---

📌 Abstract & Motivation

Institution: Vietnamese-German University (VGU) - Global Finance and Economics (GFE)

---

📌 Abstract & Motivation

Innovation is a key driver of economic growth and competitiveness, particularly in developing countries characterized by rapid economic transitions and market volatility. In such environments, fuel prices significantly impact production costs, operational strategies, and market dynamics.

This project explores the empirical relationship between national fossil fuel price regulations (taxes, subsidies, and price volatility) and firm-level innovation (product and process) across developing nations. By understanding this relationship, we aim to provide actionable insights for policymakers to design fuel regulations that foster technological adoption and resilience while maintaining economic stability.

📊 Data Infrastructure

The empirical analysis is constructed by integrating four distinct datasets via firm-level identifiers and temporal alignment:

1. World Bank Enterprise Survey (WBES) & Innovation Follow-up Survey (IFS): Standardized survey data encompassing 321,500 firm-level observations across 47 developing countries (2011–2015).
2. Global Gasoline Price Data: Monthly retail gasoline prices and international benchmark prices to calculate national net taxes and subsidies.
3. Quality of Government (QoG): National-level macroeconomic and institutional controls (e.g., GDP per capita, Corruption Index).
4. Total Factor Productivity (TFP): Firm-level productivity estimates utilizing the Cobb-Douglas production function framework.

🔬 Methodology

Due to the hierarchical and non-panel structure of the data (firms nested within industries, nested within countries over time), we employ robust econometric frameworks to account for unobserved heterogeneity and within-cluster correlations.

1. Firm Innovation (Binary Outcomes)

To evaluate the propensity of a firm to introduce a new product or process, we utilize Probit Regression and Multilevel/GEE Logistic Regression Models:

$$logit(P(Y_{ij}=1))=\beta_{0}+\sum_{k=1}^{K}\beta_{k}X_{ijk}+u_{j}$$

For interaction effects (e.g., Price Gap $\times$ Price Volatility):

$$logit(P(Y_{ij}=1))=\beta_{0}+\beta_{1}X_{1ij}+\beta_{2}X_{2j}+\beta_{3}(X_{1ij}\times X_{2ij})+u_{j}$$

Where $u_{j}\sim N(0,\sigma_{y}^{2})$ represents the random effect for country $j$.

2. Firm Productivity (Continuous Outcomes)

To evaluate the impact on Total Factor Productivity (tfprVAKL), we utilize Ordinary Least Squares (OLS) and GEE Models with population-averaged estimates and an Exchangeable correlation structure.

💡 Key Findings

• The Dual Effect of Taxation: High taxation (a positive average price gap) sends a signal to reduce consumption or shift towards alternatives, incentivizing product innovation. Firms facing higher fuel costs are driven to innovate as a strategic response to operational pressures.
• The U-Shaped Productivity Curve: The squared mean price exhibits a statistically significant impact on productivity, suggesting that while moderate price increases may initially hinder productivity, sustained high prices eventually force firms to find efficiencies and adapt.
• Volatility Undermines Policy: The interaction term between average_price_gap and price_volatility is significantly negative. High market volatility creates uncertainty, making consumers and firms risk-averse and hesitant to invest in energy-efficient technologies, thereby neutralizing the intended behavioral shifts of both tax and subsidy policies.

🏛 Policy Implications

1. Prioritize Market Stability: For fuel taxation or subsidy frameworks to successfully drive innovation, governments and regulators must first focus on reducing extreme market volatility. A predictable pricing environment allows economic agents to better assess and react to policy signals.
2. Targeted Innovation Support: Governments should couple fuel taxation with direct tax incentives or subsidies for R&D to help firms navigate cost pressures while upgrading their technological capabilities.

🚀 Repository Structure & Usage

• /data/: Contains the raw .dta files (WBES, QoG, Gasoline Prices, TFP). (Note: Proprietary datasets may require access permissions).
• /scripts/: Python scripts for data cleaning, merging, rolling average calculations (e.g., 1-year vs 3-year fiscal lag definitions), and econometric modeling using statsmodels.
• /output/: Generated analytical datasets (firm_innovation.xlsx, firm_productivity.xlsx) and regression summary logs.
• /docs/: Conference slide deck and the technical programming language report.

📝 Important Notes for Researchers & Reviewers

1. Data Availability & Reproducibility

• Raw Data Access: Due to the proprietary nature and terms of use of the World Bank Enterprise Survey (WBES) and Quality of Government (QoG) datasets, the raw .dta files are not included in this public repository.
• Replication: To fully replicate this study, researchers must independently obtain the raw datasets from their respective official sources (World Bank and QoG Institute) and place them within the data/raw/ directory before executing the analysis scripts.
• Sample Data (Optional): A small sample dataset (sample_data.csv) is provided for demonstration purposes only, allowing users to test the execution of the modeling scripts without the full dataset.

2. Methodological Clarifications

• Oil Price Lag Structures: * For Firm Productivity models, fossil fuel prices and emissions are calculated using a 1-year lag (based on the fiscal year prior to the survey).

  • For Firm Innovation models, a 3-year rolling average is applied to align with the survey questions regarding innovation activities "over the last three years."

• Handling of Multicollinearity: During the analysis, to prevent perfect multicollinearity issues caused by the unbalanced panel data, dummy variables for specific years with insufficient observations were systematically excluded and treated as the reference group.

• Model Selection: While the mathematical formulas presented in the documentation include a random effect term ($u_j$) characteristic of Multilevel Logistic Regression, the empirical analysis in Python utilizes Generalized Estimating Equations (GEE) with an Exchangeable correlation structure to account for within-country correlation (population-averaged effects).

• Questions or Feedback: If you have any questions regarding the methodology, code implementation, or findings, please feel free to open an issue in this repository.

• Citation: If you use this code or reference these findings in your own work, please cite this project as presented at the 10th International Conference on Sustainable Urban Development (ICSUD), 2024.