Economic Fitness 2

Definition

Economic Fitness (EF) is both a measure of a country’s diversification and ability to produce complex goods on a globally competitive basis. Countries with the highest levels of EF have capabilities to produce a diverse portfolio of products, ability to upgrade into ever-increasing complex goods, tend to have more predictable long-term growth, and to attain good competitive position relative to other countries. Countries with low EF levels tend to suffer from poverty, low capabilities, less predictable growth, low value-addition, and trouble upgrading and diversifying faster than other countries. The starting data is the COMTRADE list of products exported by each country. This data defines a bipartite network of countries and products. A suitably designed mathematical algorithm applied to this network leads to the Economic Fitness of all countries and the Complexity of all products. The comparison of the Fitness to the GDP reveals hidden information for the development and the growth of the countries.

Source

World Bank, Economic Fitness project. For more details, please visit
The Fitness and Complexity algorithm has been introduced in: https://www.nature.com/articles/srep00723
Details about the cleaning procedure and the predictive performance of EF are described in: https://www.nature.com/articles/s41567-018-0204-y/
and http://documents.worldbank.org/curated/en/632611498503242103/On-the-predictability-of-growth
The convergence criterion of the Fitness and Complexity algorithm is discussed in: https://link.springer.com/article/10.1140/epjst/e2015-50118-1

Development Relevance

Development encompasses many factors - economic, environmental, cultural, educational, and institutional -
which are difficult to measure, compare, and assemble in a unified picture. The Economic Fitness approach assumes that such factors are summed up in the possible export competitiveness of countries, and algorithmically extracts this information directly from data.

Methodology

The new literature of Economic Fitness uses techniques which, differently from traditional index construction approaches, do not try to average out the complexity of the system, but embraces it by explicitly building on the heterogeneity of individual actors, activities and interactions to extract relevant parameters to characterize the system. In this way, information about production capabilities may be extracted from trade in goods. The interaction among products traded, and the relatively unique combinations are a precursor to future competitiveness and long-term growth. A basic characteristic of Economic Fitness is being parameter free. The standard methods of analysis consider many elements and sum them up in some suitable way. This sum of incommensurate elements leads to a major problem of controlling noise while increasing signal. The Fitness approach starts by considering a single dataset to control noise problems. Other data can then be added later in a controlled hierarchical framework (e.g, services, technologies). The algorithm is designed on simple and transparent economical concepts which have a clear meaning and have been extensively tested. The evolution of each country is defined in the GDPper capita-Fitness space which shows a strong heterogeneity in the dynamics. There is zone characterized by regular flow and another one which is more chaotic. This implies that growth forecasting should consider this heterogeneity and go beyond standard regressions. This novel approach to the analysis and long-term forecasting has been shown to outperform the standard methods even if it requires much less data.

Limitations and Exceptions

The trade data are necessary to define a coherent network for all countries and all products. This may have some limitations for countries in which the exported products are not a good proxy of its industrial competitiveness. Also export refers generally to manufacturing. Services can be included but the corresponding database trade in services are less granular. In principle the approach could use other data like the labor statistics which automatically include all services. A basic concept of the algorithm is the importance of diversification. This is correct at the level of countries but it becomes gradually problematic if one moves to smaller scales like regions, cities up to individual firms where specialization becomes dominant. In these cases suitable modifications should be considered. The COMTRADE dataset comes at different levels of granularity. Each level has advantages and disadvantages which should be considered in relation to the problem addressed.
In order to define the countries-products network we perform several stages of data cleaning and regularization over the COMTRADE dataset. It is important to notice that some of these regularization procedures involve the use of the whole time-frame of the dataset.