Research

Abstract Informality represents a pervasive feature of many emerging and developing economies, yet standard macroeconomic models often ignore its effects, potentially biasing the analysis of shocks and the design of monetary policy. This paper studies the macroeconomic and policy implications of informality using a structural VAR for Colombia and a two-agent New Keynesian model with formal and informal sectors, featuring heterogeneous households including hand-to-mouth consumers. I show that informal labor supply shocks generate sectoral reallocation: informal activity absorbs part of the shock, sustaining aggregate output while altering wages, hours, and capital allocation. In contrast, monetary policy shocks propagate more strongly when informality is present, amplifying distributional and capital-reallocation effects. Critically, the presence of informality alters equilibrium determinacy: standard Taylor rules may fail to ensure uniqueness, with stability depending on the share of Ricardian households, the size of the informal sector, and the monetary policy stance. My findings highlight that accounting for informal production is essential for understanding transmission mechanisms and designing effective policy in economies with significant informality.

JEL E52; E62; E26; H26; O17; O54.

Keywords: Informal economy; Tax evasion; Monetary policy transmission;  Capital reallocation; Colombia.

Abstract: This paper revisits the relationship between Artificial Intelligence (AI) and green productivity by explicitly modeling AI adoption as an endogenous and directed process. Existing literature typically treats AI as a homogeneous and exogenous increase in automation, documenting a “green productivity paradox” whereby digital investment does not necessarily translate into immediate productivity gains, particularly in energy-intensive economies. We propose a framework in which firms optimally choose the direction of AI innovation between green and non-green applications. Using firm-level patent data (PATSTAT/OECD REGPAT), we construct a measure of green-directed AI as the share of AI patents associated with climate mitigation technologies (Y02). This is combined with balance-sheet data (ORBIS/Amadeus) and country-level measures of energy prices and carbon regulation. We first estimate a policy function for the direction of AI adoption, and then assess its impact on green total factor productivity accounting for endogeneity. The results show that the effects of AI on environmental performance critically depend on its direction of use. The so-called green productivity paradox largely reflects the assumption of exogenous and undirected AI adoption. Once AI is modeled as endogenous and directed, its impact is shaped by firms’ characteristics and policy incentives. 

JEL: O33, O30, Q55, Q58, L25

Keywords:Artificial Intelligence, Green Productivity, Directed Technological Change, Firm-Level Innovation, Environmental Regulation, Productivity Paradox