The Theory of Economic Complexity

We provide a mechanistic foundation for economic complexity methods. In our model, an economy’s ability to produce an activity depends on the joint presence of required factors. We analytically derive the Economic Complexity Index for this model and show that it is a monotonic function of the probability an economy holds many factors, validating it as an agnostic measure of productive capabilities. We also show that this model explains differences in the shape of networks of related activities, such as the product space or research space. These findings solve long standing puzzles in the literature on economic complexity.

💡 Research Summary

The paper provides a rigorous theoretical foundation for the Economic Complexity Index (ECI), a widely used metric that has so far lacked a clear mechanistic interpretation. The authors start by formalizing the economy‑activity bipartite network as a system of capabilities (factors) and activities that require specific subsets of these capabilities. Production is assumed to be multiplicative: an activity can be produced only if all its required capabilities are present in the economy. This assumption mirrors the classic O‑Ring and Kremer‑Shockley models but is generalized to allow any number of economies, activities, and heterogeneous capabilities.

In the simplest case—a single capability required by all activities—the authors show analytically that the leading eigenvector of the binary participation matrix (the matrix that records whether an economy possesses the required capability for an activity) coincides with the ECI. Moreover, they prove that this eigenvector is a monotonic function of the average probability that an economy holds the capability, regardless of the underlying distribution of capabilities across economies. Hence, the ECI can be interpreted as an agnostic estimator of the probability of multi‑capability endowment.

Extending the analysis to multiple, possibly correlated capabilities, the authors conduct extensive numerical simulations. Even when more than half of the capability assignments are random, the ECI remains a monotonic function of the average number of capabilities an economy possesses. This robustness demonstrates that the ECI captures the underlying “capability richness” of an economy without needing to identify the capabilities explicitly.

The paper further embeds the capability‑based production function into a short‑run general equilibrium framework. By specifying a shifted Cobb‑Douglas‑type production function (Y_{cp}=B+f_c g_p), where (f_c) reflects an economy’s capability endowment and (g_p) an activity’s capability intensity, the authors derive equilibrium wages, prices, and consumption. Wages are proportional to capability endowment, while product prices follow a convex function of capability requirements, implying a premium for complex goods. This micro‑foundation explains the empirically observed positive correlation between ECI and future economic growth.

A major contribution is the explanation of differing topologies in networks of related activities. When the capability endowment matrix exhibits strong correlations (block‑structured), the resulting product space displays a dense core of high‑complexity products surrounded by a peripheral low‑complexity region—a “core‑periphery” structure documented in trade data. Conversely, when capabilities are arranged according to a Toeplitz circulant matrix (i.e., each capability is correlated with its neighbors in a circular fashion), the resulting research space forms a ring where each field connects primarily to adjacent fields, matching observations from citation and co‑authorship networks. Thus, the model links the statistical properties of capability distributions to the macroscopic geometry of knowledge and production networks.

The authors discuss several implications. First, the ECI is not merely a proxy for diversification; it measures the probability of possessing multiple complementary capabilities. Consequently, economies with the highest ECI may actually be less diversified than moderately complex economies, reconciling the non‑monotonic relationship between diversity and complexity observed in data. Second, the model clarifies why the ECI remains a strong predictor of growth despite being derived from normalized, binary RCA matrices: the underlying multiplicative production process preserves information about capability endowments through these transformations. Third, policy makers should recognize that the shape of the related‑activity network (core‑periphery vs. ring) reflects the structure of capability correlations, suggesting tailored industrial or research strategies.

Finally, the paper situates its contributions within the broader complex‑systems literature, showing that a simple generative model with unobserved, heterogeneous, complementary constraints can give rise to the low‑dimensional spectral signatures used across fields such as ecology, health, and scientometrics. By bridging the gap between empirical economic‑complexity tools and a first‑principles production theory, the work solidifies the theoretical legitimacy of the ECI and opens avenues for its application in other complex‑system domains.