This work aims to analyze input-output matrices to understand the structural relationships in an economy and how the system evolves. We first place our study in the intersection of two streams in the I/O literature. The first one defines measures or indicators based on matrices of inter-industrial transactions (Kolokontes et al., 2019; Miller and Blair, 2009). This is largely built on I/O matrices defined by national accounting and general equilibrium macroeconomic models. The second one pursues a strategy based on complexity approaches (Bosma et al., 2005; Meyers, 2009). This means that the matrix of inter-industrial transactions is described as a complex network whose dynamics motivates the emergence of properties in the economic system, e.g. level of output, imports/exports… Within this framework, the analysis becomes similar to another complex approaches used in ecology, physics, etc. We will use available databases that have been standardized over the last years. Specifically, we focus our analysis on using non-parametric estimation (Hastie et al., 2001) to describe I/O matrices through selected indexes, including the Complexity Index (which is based on structural clustering). Once these indexes are identified, we identify changes in economic structures over time. From this point of view, we will check whether those measures can be used as proxies of structural aspects of an economy to design economic policies.
