Using AI in the informal currency market: evidence from Cuba

This article explains the various steps of a novel methodology designed to analyse the informal currency market and calculate a parallel exchange rate (ER). We utilize artificial intelligence algorithms for Natural Language Processing to extract data from social network groups and classified websites. The methodology has been implemented since 2021 amidst Cuba’s complex financial and political landscape. Given that the official exchange rate remains fixed and overvalued, the calculated ER has become vital for assessing the real-time effects of regulatory changes, economic shocks, and policies. We conduct a statistical and econometric analysis using vector autoregressive models and spillover indexes to validate the economic and financial consistency of the informal market’s time series.