Universität zu Köln

Modeling and Estimating the Loss Given Default of Leasing Contracts

Miller, Patrick Michel (2017) Modeling and Estimating the Loss Given Default of Leasing Contracts. PhD thesis, Universität zu Köln.

PDF - Published Version
Available under License Creative Commons Attribution Non-commercial No Derivatives.

Download (4Mb) | Preview


    This thesis consists of three essays dealing with the modeling and estimation of the LGD of leases. The first essay (Hartmann-Wendels, Miller, and Töws, 2014, Loss given default for leasing: Parametric and nonparametric estimations) focuses on the methodological aspects of estimating the LGD and compares different estimation approaches. The results expose that it is indispensable to interpret the results of different validation techniques in order to reliably evaluate which methods are suitable to predict the LGD of leases. Specifically, it its exposed that in-sample results might be misleading when estimating out-of-sample LGDs. It turns out that finite mixture models are capable of reproducing the shape of the LGD density, but the generated LGD predictions are quite poor. In fact, the results show that the model tree M5’ is particularly suited to predict leasing LGDs. Motivated by the insights gained in the first study (Miller, 2015, Does the Economic Situation Affect the Loss Given Default of Leases?) analyzes the factors driving the LGD of leases. The results show that the factors identified as drivers of the LGD are to some extent different for the considered companies. Nevertheless, the results affirm that the LGD of leases generally depends in particular on factors that are related to the leases asset. Based on the findings of the first two studies, the third essay (Miller and Töws, 2016, Loss Given Default-Adjusted Workout Processes for Leases) introduces an LGD estimation approach explicitly for leases. This approach bases on the economic consideration that the revenues received during the workout process of a defaulted lease come from two payment sources. The results show that the developed multi-step model generates precise and sturdy predictions of leasing LGDs. Moreover, this multi-step LGD estimation approach provides valuable interim results that can be used as a decision support for actions to be taken during the workout process of a defaulted lease.

    Item Type: Thesis (PhD thesis)
    Miller, Patrick MichelP-Miller@gmx.de
    URN: urn:nbn:de:hbz:38-71214
    Subjects: General statistics
    Management and auxiliary services
    Uncontrolled Keywords:
    Loss Given Default, Leasing, Workout Process, Regression and Model Trees, Finite Mixture Models, Idiosyncratic and Macroeconomic Factors, Economic Model, Forecasting, ValidationEnglish
    Faculty: Wirtschafts- u. Sozialwissenschaftliche Fakultät
    Divisions: Wirtschafts- u. Sozialwissenschaftliche Fakultät > Seminar für Allg. BWL und Bankbetriebslehre
    Language: English
    Date: 03 February 2017
    Date Type: Publication
    Date of oral exam: 11 January 2017
    Full Text Status: Public
    Date Deposited: 01 Feb 2017 10:38:03
    NameAcademic Title
    Hartmann-Wendels, ThomasUniv.-Prof. Dr.
    Hess, DieterUniv.-Prof. Dr.
    Schradin, Heinrich R.Univ.-Prof. Dr.
    URI: http://kups.ub.uni-koeln.de/id/eprint/7121

    Actions (login required)

    View Item