Working Papers

Automation, Gender Tipping and Occupational Segregation, with Sergi Basco and Marc Teignier

Abstract: This paper examines how industrial automation reshapes gender composition within occupations. Automation disproportionately displaces the minority gender within an occupation, causing occupations to become more concentrated in their historically dominant gender. This pattern increases within-occupation segregation, even as aggregate gender gaps remain unchanged. The effect is strongest in routine-intensive occupations and remains robust across alternative specifications. These findings suggest that automation influences gender inequality through occupational sorting rather than differential impacts on wages or overall employment opportunities.

International Propagation of the "AI Bubbles", with Sergi Basco and Marc Teignier

Abstract: We examine how AI valuation bubbles propagate internationally through multinational production (MP) networks.  We identify three transmission channels: a financial channel via cross‑border equity holdings, a downstream channel through globally traded AI services that inflate foreign effective labor, and an upstream channel via bubble‑financed investment pulling foreign intermediates. When the bubble bursts, the bubble‑financed fraction of installed AI capital is destroyed, triggering a global real contraction amplified by production linkages. The analysis quantifies the international exposure to speculative dynamics of AI bubble, a risk absent under traditional multinational firms.

Who Issues Programmable Money: Screening and Credit Creation, with Yidan Yan.

Abstract: Central bank digital currencies (CBDCs) embedded with smart contracts can act as a programmable lending technology that restrict the use of borrowed funds, extending their functionality beyond payments. This paper develops a theoretical model of programmable CBDC lending under asymmetric information, identifying a core trade-off: restricting fund use sharpens screening and curbs inefficient lending to risky firms, but the full-reserve backing requirement suppresses banks’ endogenous credit-creation capacity.We characterize when banks optimally adoptCBDClending, showing that adoption hinges on borrower composition and balance-sheet strength—severe adverse selection favors CBDC substitution, while an abundance of productive borrowers preserves conventional intermediation. We further show that CBDC lending can expand credit access for constrained small firms yet lower aggregate welfare, as private lending decisions do not internalize the social value of credit allocation. The consequences of programmable money thus depend fundamentally on how it reshapes the microeconomic foundations of financial intermediation.

Accepted and Published Papers

To be released.