
Imagine you have a bathroom scale that reads 170 lbs when you stand in the kitchen but 155 lbs when you move to the bedroom. You wouldn't trust either number — you'd fix the scale. That's the core problem Du and Zaccaria identify in how we measure economic complexity at subnational levels. The same product gets a different complexity score depending on whether you're looking at countries, states, or provinces. The ruler changes when you zoom in. The committed claim: network-based economic complexity indices — the tools policymakers and development economists increasingly rely on to rank regions and predict growth — produce inconsistent results when applied below the national level. The estimated complexity of the same product shifts with the geographical scale of analysis. This isn't a minor calibration issue; it means the rankings themselves are artifacts of methodological choices, not stable properties of the economies being measured. The field fight here is between endogenous and exogenous approaches to measuring product complexity. The dominant methods — Economic Complexity Index (ECI) from Hidalgo and Hausmann, and Fitness-Complexity from Tacchella, Cristelli, Caldarelli, and Zaccaria himself — compute product complexity from the same bipartite network of countries-and-products that they use to compute country complexity. Elegant, but circular: the complexity of a product depends on who makes it, and who makes it depends on the complexity of their basket. When you swap out countries for provinces, the network topology changes, and the circular computation settles on different fixed points. The product didn't change — the mirror did. Du and Zaccaria's proposed fix is structurally simple: use an exogenous measure of product complexity — one computed independently of the territorial units being ranked — and combine it with an extensive (additive) aggregation rather than the intensive (average-like) aggregation most methods use. This means a region's complexity score reflects both the sophistication of its products AND the breadth of its productive capabilities. The choice of extensive over intensive is doing real load-bearing work here: it prevents a tiny region that makes one complex product from outranking a diversified industrial economy. The validation is correlation-based: the proposed measure aligns more strongly with GDP per capita and employment than the existing network-based indices when applied subnationally. That's a meaningful signal, but it's worth noting the circularity risk — GDP per capita is itself a standard economic indicator, and showing your complexity measure correlates with it doesn't prove the measure captures something GDP doesn't already tell you. The real test would be predictive power: does the new measure forecast future GDP growth better than the old ones, or better than GDP itself? The practical upshot matters because subnational complexity analysis is booming. Regional development agencies, EU cohesion funds, and national industrial strategies increasingly use complexity metrics to allocate resources and target interventions. If those metrics are scale-dependent artifacts, then billions in development spending may be chasing phantoms. Du and Zaccaria aren't just fixing a methodological wrinkle — they're questioning whether an entire applied literature has been building on unstable ground. The honest limitation: this is an abstract-level read. The specific magnitude of inconsistency, the number of products and regions tested, and the exact correlation improvements are not available from the abstract alone. The claim is directionally important but the strength of the evidence requires reading the full paper.
The analogy
Imagine a bathroom scale that reads different weights depending on which room you stand in. You wouldn't blame your body — you'd blame the scale. Du and Zaccaria show that economic complexity indices are that scale: the same product gets different complexity scores depending on whether you measure at the country or province level. Their fix is to calibrate the scale once, externally, and then use it everywhere.
Broken scale
A measurement device whose output changes based on where you stand, not what you weigh
Complexity index
A network metric whose output changes based on geographical resolution, not the product's actual sophistication
Honest assessment
This paper asks the right question — do complexity metrics survive a change of zoom level? — and the answer appears to be no, which matters because subnational applications are proliferating in policy. But the validation is correlation-based (not predictive), the abstract provides no specific numbers, and the insider dynamic (Zaccaria critiquing his own prior method) makes this either admirably honest or strategically positioned. Worth reading if you use complexity indices for regional analysis; worth watching for the predictive validation sequel.
What Is CLAIMS?
53
/ 100
BedrockNews applies a structured analytical framework called CLAIMS to academic papers and technical research releases.
How CLAIMS worksClaim
3/5
Ladder
3/5
Architecture
3/5
Integrity
2/5
Milestone
2/5
Successor
3/5
What is this paper actually committing to?
Diagnostic, not constructive. The real claim is that existing methods break at subnational scale — important if true, but exposing a flaw in others' tools is a different category than building something new.
The field fight
Should economic complexity be computed endogenously from the country-product network, or should product complexity be fixed exogenously?
Side A
Endogenous computation captures emergent structure; the network IS the information
Held by: ECI (Hidalgo-Hausmann), Fitness-Complexity (Tacchella et al.)
Side B
Endogenous methods are scale-dependent artifacts; exogenous product complexity is more robust
Held by: This paper (Du, Zaccaria)
Milestone tracker
Unlocks: Reliable subnational complexity rankings for EU cohesion policy and World Bank regional diagnostics
Current
2
methods shown inconsistent
this paper
Next
5+
country datasets
cross-country validation
~1-2 years for independent replication across diverse economies
Verified Facts (5)
Several network-based measures have been proposed to assess the economic complexity of countries
Such applications lead to inconsistent results, in the sense that the estimated complexity of the same product appears to depend on methodological details such as the geographical scale of analysis
We propose a measure of territorial economic complexity based on an exogenous and extensive computation
These methodological choices yield estimates that are more consistent and more strongly aligned with standard economic indicators, such as GDP per capita and employment
Authors: Wenli Du, Andrea Zaccaria
Practical Implications (3)
If you're using ECI or Fitness-Complexity at the regional/state/province level, your rankings may be artifacts of scale, not real economic signals
Regional development agencies and EU cohesion funds using subnational complexity metrics should audit for scale-consistency before allocating resources
The exogenous-extensive approach offers a drop-in alternative that may be more robust for subnational policy targeting
More latest news
Read all