Editor’s Note: Annie White, Director of Software Tools at the Growth Lab, Harvard Kennedy School, and Sebastian Bustos, Senior Research Fellow at the Center for International Development, Harvard University, recently presented their flagship project, the Growth Lab. Their lecture, titled Every Country’s Data At Your Fingertips: The Atlas of Economic Complexity, was delivered via Zoom. The session was part of the International Trade Fellowship, organised by the National Press Foundation and hosted at the Hinrich Foundation office in Singapore. This article is the second in a series that will delve into key themes and insights from the Fellowship sessions over the coming days.
by Our Diplomatic Affairs Editor
The capacity of a country to produce and export a wide range of sophisticated goods has long been regarded as a marker of its economic strength. Yet the ability to measure this capacity, and to interpret it meaningfully, has remained elusive for many. At the centre of the recent session of the International Trade Fellowship was a presentation by Annie White and Sebastian Bustos, who have worked extensively on the Harvard Growth Lab’s Atlas of Economic Complexity, a data tool designed to evaluate and visualise how countries perform and evolve in terms of productive knowledge. Their lecture introduced not only the core principles that inform the Economic Complexity Index (ECI), but also the technical architecture of the tool and its application to contemporary trade and policy questions.
White and Bustos began by introducing the foundational concepts behind the ECI, which measures the diversity and sophistication of a country’s productive output based on the products it exports. A country that can produce a wide variety of products that few others can make is considered economically complex. This form of analysis offers an alternative to traditional macroeconomic indicators by focusing on the underlying knowledge embedded in production and trade. The Atlas uses detailed trade data—particularly exports at the four-digit level of product classification—to identify the depth and breadth of economic activity. Through this approach, the tool assesses whether a country is positioned to sustain long-term development or is overly reliant on low-value or easily replicable goods.
An important feature of the Atlas is its interactivity. Users can access and explore economic data from over two decades, comparing countries’ performance, tracking changes in their export structures, and observing how certain industries emerge, decline, or stabilise over time. White demonstrated how the tool enables a user to filter by product category, sector, and time period, offering insights that are both detailed and accessible. Although primarily built for economic research and policy analysis, its open design and free availability also make it valuable to journalists, educators, and the general public.
One major question addressed during the session was whether the tool could show how the complexity of an economy changes over time—not just in absolute terms, but also in relation to other economies. White explained that although the current data visualisations include time series showing ECI rankings since 2000, the question of economic acceleration or deceleration is more complicated. The ECI is a relative measure, meaning that changes in one country’s ranking may reflect developments elsewhere as much as changes in its own economy. To understand a country’s trajectory, it is necessary to look not just at the position it holds on the index, but also at the specific products it gains or loses competitive advantage in, and how these shifts compare with global trends.
This issue brought forward the experience of countries that have moved backwards on the index. Bustos, reflecting on his background in Chile, pointed to Argentina as a case where productive capacity has deteriorated over time. Argentina was once ranked relatively high in terms of economic complexity but has declined due to economic mismanagement and structural weaknesses. The ECI captures such regressions clearly, which serves as a reminder that productive knowledge, once lost, can be difficult to recover. These examples serve to caution against viewing complexity as a guaranteed upward path; it is subject to both gains and losses depending on political, institutional, and social conditions.
Further questions were raised around the role of data accuracy. International trade data is known to suffer from several weaknesses. Countries may report incomplete or misleading information, and some flows are deliberately obscured for political or commercial reasons. In response, White and Bustos described how their methodology attempts to mitigate these limitations. They employ a form of data validation known as “mirror statistics,” in which trade flows are analysed from both the exporter’s and importer’s perspectives. If a country reports that it has exported a certain quantity of goods, but the receiving country records a different figure for imports, the discrepancy can be investigated. Using statistical models, the Growth Lab team weights the reliability of different data sources and arrives at a corrected estimate of the actual trade flow. This method allows for a more dependable reconstruction of trade activity, especially in regions where official statistics are inconsistent or unavailable.
One of the more complex aspects of the discussion related to the classification of economies that specialise in assembly rather than full-scale manufacturing. Vietnam was presented as a case in point. While the country is a major exporter of electronics and computers, it does not produce many of the high-value components used in these products. The question was raised as to whether such economies are being given more credit for complexity than they deserve. Bustos clarified that the Growth Lab’s analysis depends on whether a country exports a product with what is called “revealed comparative advantage” (RCA). This is a measure of whether a country exports a product at a higher share than the global average, suggesting it does so competitively. In other words, not all exports are treated equally. A country that exports a trivial amount of a complex product will not be considered a genuine producer in that area. However, if it meets the RCA threshold, it is assumed that the country possesses or is developing the necessary capabilities, even if these initially involve only partial stages of production.
In the case of Vietnam, Bustos described how garment manufacturing often serves as a gateway to more complex industries. Workers trained in textiles may transition to electronics assembly due to the overlapping demands of quality control, precision, and dexterity. Over time, as firms develop local supply chains and invest in training and infrastructure, the economy moves from assembly to more integrated forms of production. The data confirms this: Vietnam’s exports have diversified and increased in complexity, suggesting an upward shift in the structure of its economy.
Another key contribution of the Atlas is its geographic component. Users can visualise trade flows between countries or regions, not only by product but also by destination. One participant asked whether it is possible to view changes in these flows over time, particularly the emergence of new trade relationships or the decline of older ones. The answer is yes. By using the Atlas’s three-map visualisation, users can examine how the composition and direction of exports have evolved. For instance, one can observe how the UK’s exports to Eastern Africa have changed in value and composition between 2010 and 2023, or how China has shifted its sourcing strategies by increasing investment in Southeast Asian economies. These features make the Atlas especially useful for those studying global supply chains and their geographic reorientation.
The session concluded with questions about informal or illicit trade, such as money laundering or unrecorded flows. While the Growth Lab’s tools do not directly measure illegal activity, their methodological transparency makes them suitable for identifying anomalies. For example, if reported imports from a country are disproportionately low compared to its known production, it may suggest hidden flows or manipulation of data. Bustos noted that many African economies have few direct records of trade but can still be analysed using data from their trading partners. In this way, the Atlas provides partial visibility even in difficult data environments.
The core strength of the Atlas lies not only in its methodological sophistication, but also in its accessibility. The platform does not require specialised training in econometrics or data science to use effectively. With clear visuals and interactive features, it allows users to answer specific questions—such as which products a country is gaining competitiveness in, or how trade with a given region has changed—without needing to extract or manipulate raw data. This makes it especially relevant for journalists and policy researchers who need reliable information quickly, without relying on government press releases or proprietary databases.
Yet perhaps the most significant aspect of the Growth Lab’s approach is its framing of economic development as a process of acquiring productive knowledge. Rather than focusing exclusively on inputs such as capital or labour, the Atlas attempts to capture the knowledge required to make complex goods. This view aligns with a broader tradition in development economics that treats production capabilities—not just endowments—as the central factor in long-term progress. In doing so, it brings attention to the importance of education, institutions, infrastructure, and industrial policy, all of which play a role in enabling countries to move into more demanding forms of production.
The relevance of this work is amplified in a period when global trade patterns are in flux. Supply chains are being restructured due to political tensions, technological changes, and environmental concerns. Countries are reconsidering their dependencies and seeking to build domestic or regional production bases. In this context, a tool that provides insight into how economies build and lose productive capabilities is not only informative but necessary.
White and Bustos’s presentation at the Fellowship session provided a clear and thoughtful overview of how the Atlas of Economic Complexity functions, what questions it can help answer, and where its limitations lie. Their emphasis on the relative nature of the ECI, the attention to data quality, and the ability to visualise geographic and temporal trends offers users a wide range of possibilities. The discussion also served as a reminder that while quantitative tools can inform our understanding, they must be interpreted in light of historical, political, and institutional contexts.
The Atlas does not claim to predict the future with certainty. It offers a structured view of what is already happening beneath the surface of trade statistics and provides clues about where an economy might go next. By focusing on actual production rather than abstract indicators, it presents a more grounded view of development. For those involved in trade reporting, policy analysis, or economic research, it is a resource of substantial value—one that combines technical reliability with practical usability.
