Rethinking Country Asset Allocation: Big Data and Cluster Analyses

Countries are increasingly bucketed as “developed” or emerging, even though most countries fall along a continuum of socioeconomic development. The Capstone team built upon existing research by Loomis, Sayles & Co. to group 90 emerging markets in algorithmically determined clusters based on their performance in two categories of indicators: financial/economic, including data on currency valuation, market capitalization, reserves coverage, and credit rating; and social/institutional, including data on demographic patterns, political rights and civil liberties, and labor force participation. Cluster analysis is a method that classifies multivariate data based on proximity to maximize homogeneity within each cluster (internal cohesion) as well as heterogeneity between different clusters (external isolation).

The team observed significant stability in the top-ranked cluster, consisting of small wealthy economies such as Singapore and Luxembourg, as well as in the bottom- ranked clusters consisting of low-income countries with weak institutions such as Venezuela, Pakistan, and Honduras. On the other hand, the team also observed movements of both developed and developing countries across the various clusters. These movements include, for example, the visible impact of the sovereign debt crisis on countries such as Greece, Portugal, Spain and Italy.

The implications of finding a data-driven way to organize similar countries together into truly informative and dynamic groupings are enormous: A low correlation between returns and level of market development confirms the benefits of diversification. The research also showed the stark contrast in how different clusters perform on the S&P Global Equity Index: while the highest-ranked clusters tend to have moderate median returns with relatively low spread, lower ranked clusters, featuring countries such as Rwanda, Bangladesh, and Zambia, have poor returns. The results of this research will help Loomis Sayles & Co identify the specific factors that investors should consider in making investment allocation decisions.