Groupe d’études et de recherche en analyse des décisions

G-2021-11

Quasi-maximum likelihood for estimating structural models

, , et

The structural model of Merton (1974) gives rise to multiple applications and extensions in corporate credit-risk analysis. The estimation of this framework poses a major challenge as its underlying state variable (the firm's asset value) is not directly observable. Since Duan (1994), maximum likelihood has become the benchmark to estimate structural models where corporate securities are valued in closed form. We propose a quasi-maximum likelihood (QML) approach that remains appropriate even when the explicit approach is unachievable. QML is highly flexible and effective. To assess our construction, we conduct an empirical investigation, highlight the credit-spread puzzle, and discuss a remedy via bankruptcy costs.

, 11 pages