Where does recourse collide?

In stochastic optimization, recourse is the corrective second-stage decision a planner makes after uncertain data are revealed: a first-stage choice is committed under uncertainty, then once outcomes are known, recourse actions adjust the plan at some cost. Dantzig's 1955 Linear programming under uncertainty founded this two-stage "programming with recourse" framework. The referent is a modeled adaptive decision inside an optimization problem, measured by the expected cost of the recourse actions, and it is a different construct from a lender's claim on a borrower beyond the shared idea of a fallback.[1]

Unit

The second-stage corrective decision and its expected cost

Recovery means

Not applicable

Measured by

Quantified by the expected recourse cost in a two-stage stochastic program

In lending, recourse describes how far a lender can pursue a defaulting borrower: with recourse debt the lender can claim the borrower's other assets or income beyond the pledged collateral, while non-recourse debt limits the lender to the collateral itself. Ghent and Kudlyak's 2011 Recourse and residential mortgage default found that borrowers in US states permitting recourse default less readily. The referent is the scope of a creditor's claim on default, measured by whether and how a deficiency can be recovered, and it is unrelated to a corrective decision in an optimization model beyond the shared meaning of a remedy to fall back on.[2]

Unit

The scope of the lender's claim beyond the pledged collateral

Recovery means

Not applicable

Measured by

Classified as recourse versus non-recourse, and by the deficiency recoverable after collateral is seized