Research
Research
WORKING PAPERS
ABSTRACT: According to French law, employers have to pay at least six months salary to employees whose seniority exceeds two years in case of unfair dismissal. We show, relying on data, that this regulation entails a hike in severance payments at two-year seniority which induces a significant rise in the job separation rate before the two-year threshold and a drop just after. The layoff costs and its procedural component are evaluated thanks to the estimation of a search and matching model which reproduces the shape of the job separation rate. We find that total layoff costs increase with seniority and are about four times higher than the expected severance payments at two years of seniority. Counterfactual exercises show that the fragility of low-seniority jobs implies that layoff costs reduce the average job duration and increase unemployment for a wide set of empirically relevant parameters.
ABSTRACT: We study the propagation of financial shocks in decentralized finance (DeFi) using a novel dataset from Compound, a prominent DeFi lending application on Ethereum. Unlike traditional interbank networks, Compound exhibits a bipartite structure in which users lend and borrow via lending pools implemented as smart contracts. We construct daily balance sheets for users and pools from January 2020 to June 2024, model the liability network, and apply the DebtRank algorithm to simulate distress cascades following tokenspecific price shocks. Our findings show that the network topology is the most robust predictor of contagion, outperforming standard financial indicators. We further show that the structure of systemic risk varies over time and across asset types, with stablecoin pools exhibiting more concentrated and persistent vulnerabilities than crypto-asset pools. These results underscore the need for topology-aware risk monitoring in algorithmic credit systems.
ABSTRACT: Delegated staking protocols allow operators to validate blockchain networks using external capital. This improves capital efficiency but creates a moral hazard problem: when an operator is slashed, the penalty falls largely on investors' assets rather than her own. We show that without reputational concerns, a fully delegated operator exploits every attack opportunity regardless of slashing severity. We then develop a dynamic model in which operators choose both their stake and the share they fund with their native capital, while the market learns about operator types from their track records. Native capital and reputation act as substitutes. New operators have no reputation to lose and are disciplined by putting their native capital at risk. As their track record grows, they replace this costly collateral with cheaper delegated capital. Reputation then acts as synthetic equity, disciplining operators through the threat of losing future reward. Stronger slashing improves security monotonically, both by deterring attacks and by reducing the re-entry of attackers under new identities. However, operators partly offset this improvement by economizing on their native capital. Our results provide a theoretical foundation for the design of security mechanisms in delegated staking protocols.
PAPERS BY TOPICS
SELECTED PUBLICATIONS
OTHER PUBLICATIONS