Updating MRIO models – An application to the Portuguese case

Economic challenges in the post-COVID years highlighted sectorial linkages, logistics, and trade relationships. The availability of new data sources, combined with new methodological advances, allowed researchers to improve their knowledge of how shocks spread downstream and upstream of the production process and ultimately impact the macroeconomic indicators of a region/country. Given the absence of multiregional Input-Output (I-O) tables that would allow for an adequate understanding of complex sectorial/regional linkages in the Portuguese context, this project seeks to design and implement a method that generates and updates input-output matrices at the NUTS-II level (PReMMIA – Portuguese Regions Model for Multi-Impact Analysis).
PReMMIA is a Multiregional I-O (MRIO) model that depends on the sectorial information for I-O tables provided by Statistics Portugal (INE) on the National Accounts, which is made available with considerable time lags between them (typically five years). Thus, given the changes that affected the Portuguese economy after 2017 (COVID-19 pandemics and Russo-Ukrainian War), we develop and apply allocation estimation techniques to update MRIO models over time, using parsimoniously additional statistical information as soon as it becomes available. In other words, we depart from the existing matrices and use updated statistical information, made available typically as dispersed indicators in the Tables by Industry and Regional Accounts on the National Accounts, Agriculture Census, National Forest Inventory, and regional data on energy and electricity production. We will integrate such information into the model and enable an adequate updating of PReMMIA.
This method applies matrix balancing algorithms to know the core of the matrix in a subsequent period, based on the matrix for the previous period and the known updated information. In this work, we compare the different methodologies of bi-proportional matrix balancing and estimate the associated error in the case of Portuguese National Accounts. RAS, Generalised RAS, and Sign Preserving Squared Differences techniques are applied to compare the values on the estimations of the I-O tables for 2017, based on 2015, with the ones of the actual I-O tables for 2017. After this process, we will have reunited the conditions to apply the most robust technique to update PReMMIA, promoting a better assessment of the economic structure of the regions/country, obtaining a solid model applicable accurately over the years, and providing updated spillovers and feedback effects bi-regionally.
At last, the project will be extended in close cooperation with policymakers, empowering them for the (ex-ante and ex-post) economic impact assessment of public policies and of (planned or unexpected) shocks/events on one of the country’s regions, with a high level of sectorial disaggregation and with estimations on its distribution at the regional level.