Analyzing the Economic Complexity of the Western Roman Empire

Written by Michele Coscia, this piece explores the intersection of historical data and modern economic metrics. The author admits to a persistent preoccupation with the Roman Empire—a passion likely fueled by an addiction to Crusader Kings III and similar grand strategy games.

The "Eureka" Moment

The project began when the author's expertise in economic complexity (a metric used to track development trajectories) collided with a presentation by Petra Heřmánková. This intellectual spark led to a realization that could be described as a "data science version of Young Frankenstein."

"It can be done!"

This exclamation marked the beginning of a collaborative effort between Petra, Matteo Mazzamurro, Tom Brughmans, and Michele Coscia. Their findings were eventually published in the journal npj Complexity under the title "The economic complexity of the Roman Empire as inferred from Latin inscriptions."

The Data Source: LIST

The foundation of this research is the LIST dataset (Latin Inscriptions in Space and Time), which represents the most comprehensive collection of Roman-era Latin inscriptions available. Each entry provides:

1. The precise geographic coordinates of the find.
2. A digital transcription of the text.

From Death to Data

The researchers leveraged a morbid but effective link: the connection between death and economic record-keeping. Most inscriptions were created to immortalize an individual, often boasting about their achievements and, crucially, their profession. This allowed the team to map specific economic activities to exact latitudes and longitudes.

Addressing Data Biases

The team recognized that the data could not be taken at face value. They identified several critical biases that needed correction:

To mitigate these, Petra researched correction methods, and Matteo implemented the necessary analytical frameworks.

The Results and the ECI

The analysis produced an estimate of occupational distribution across the empire's provinces.

The resulting heatmap showed a "nested pattern"—a hallmark of economic complexity—where Latium (Rome) exhibited the highest diversity of occupations, while farming remained the most ubiquitous. This led to the calculation of the Economic Complexity Index (ECI).

The Research Workflow

A Bridge Across Millennia

Because the data included precise coordinates, the team could calculate the Roman ECI using the borders of modern-day countries. By comparing these ancient values with contemporary ECI data for approximately 23 countries, they calculated the Spearman rank correlation:

$\rho = 1 - \frac{6 \sum d_i^2}{n(n^2 - 1)}$

Despite a gap of 1,500 years, the correlation was significant. While this doesn't prove a direct causal link, it suggests that economic complexity is "sticky"—meaning regions that were complex in the classical era often share a signal with regions that are complex today.

# Conceptual logic for the correlation test
import scipy.stats as stats

roman_eci = [val for val in western_empire_data]
modern_eci = [val for val in current_global_data]

correlation, p_value = stats.spearmanr(roman_eci, modern_eci)
print(f"Correlation: {correlation}, P-value: {p_value}")

Final Thoughts and Future Steps

The project encountered a few hurdles and left some "to-do" items:

• The Eastern Empire: The sister dataset, GIST (Greek inscriptions), is currently not digitized to a sufficient standard, leading to unreliable results.
• The Western Empire: Successfully mapped and correlated.
• Geopolitical Necessity:

"Carthage must be destroyed."

(The author notes that more on the latter point will follow later).