Econometrics Project

Project Overview

This econometrics project explores the determinants of Preference Utilisation Rates (PUR) under the UK-EU Trade and Cooperation Agreement (TCA). Preference utilisation is a key metric in trade policy, indicating how effectively firms are taking advantage of zero-tariff provisions under the TCA.

The core research question is: “What are the relative effects of tariff margins, Rules of Origin (RoO), and trade scale on PUR between the UK and EU?”

Econometric Strategy

Guided by established literature—particularly Nilsson (2015) and Ayele et al. (2023)—the determinants of PUR are grouped into three key effects:

• Margin Effect: The incentive created by tariff differentials between MFN and preferential rates.
• RoO Effect: The restrictiveness or complexity of the rules firms must meet to qualify for preferences.
• Scale Effect: The size of trade flows, which can affect firms' ability to absorb compliance costs.

I ran two model specifications using Python (in Visual Studio Code):

• A standard Ordinary Least Squares (OLS) model
• A Fractional Logit Model—appropriate for a dependent variable bounded between 0 and 1

Tools and Methods

• Software: Python, VS Code, Overleaf (for academic write-up)
• Libraries: statsmodels, pandas, matplotlib, linearmodels
• Data Prep: Feature engineering to derive log-scale variables, create dummy variables for HS codes and countries, and normalize restrictiveness indices.
• Model Diagnostics: Residual plots, heteroskedasticity checks, and sensitivity testing (e.g., excluding outliers, log-transforms)

Key Findings

• The preferential margin had a statistically significant positive effect on PUR—suggesting tariff incentives matter.
• Restrictiveness of Rules of Origin negatively impacted PUR, particularly in sensitive agricultural sectors.
• Trade scale (log of preferential imports) correlated positively with PUR, supporting economies of scale in compliance.

These results align with broader literature and provide actionable insights into the barriers and incentives that shape trade behaviour under free trade agreements.

Visual Summary

What I Learned

• How to select the right econometric model for bounded continuous outcomes
• The importance of thoughtful variable transformation and clustering in cross-sectional trade data
• How to communicate technical findings clearly in an academic-style report and in policy briefings