Page 418 - Special Topic Session (STS) - Volume 3
P. 418
STS552 Natalia Nehrebecka
The Basel Committee on Banking Supervision (2005) points to the importance of
adequate estimates for economic downturns and unexpected losses. The
1
purpose of this research is forecasting the recovery rate of non-financial
corporations with particular emphasis on sectorial analysis. The company’s
industry and its characteristics have an impact on the loss given default. In
addition, a research hypothesis has been put forward that assumes that
enterprises operating in industries where the market is small have higher losses
in the event of the company’s insolvency due to the lack of active bidders’
market. If the market is not liquid, it is more difficult for creditors to recover the
amounts due, and the time may be increased until they are collected.
Research on the Loss Given Default began to gain momentum only in the
21st century. Most empirical work on LGD for loans began to arise after the
introduction of the New Capital Accord in 2004. In the first works on modelling
losses due to default [Altman, Gande and Saunders 2003; Arner, Cantor, Emery
2004; Cantora and Varmy 2010] linear regression was used. The Basel Committee
on Banking Supervision (2005) points to the importance of adequate estimates
for economic downturns and unexpected losses. Board of Governors of the
Federal Reserve System (2006) proposes the computation of Downturn LGD
measures by a linear transformation of means [Dowturn LGD = 0,08 + 0,092 * E
(LGD)]. Most academic and practical credit risk models focus on mean LGD
predictions. However if we consider two loans with different distributions (a
uniform and a beta) but the same means values then we have real quantiles and
downturns as well as unexpected losses differ. A relatively new method for
modelling the loss given default, which was also used in this research, is quantile
regression [Somersa and Whittaker 2007; Krüger and Rösch 2017]. While other
methods only allow estimating the mean or variance of LGD, quantile regression
allows modelling of all quantiles of the dependent variable. In this way, it is easy
to obtain measures in the event of a downturn and unexpected losses.
The remainder of this paper is organized as follows. Section 2 presents the
methodology. Section 3 presents and discusses the empirical results, while
Section 4 concludes the paper.
2. Methodology
In order to calculate the LGD parameter, the Recovery Rate (RR) should be
initially estimated. RR is defined as one minus any impairment loss that has
occurred on assets dedicated to that contract (see IAS 36, Impairment of
Assets) / Exposure at Default. Most LGDs are nearly total losses or total
1 Recovery Rate given default is the part of the loan liabilities that a creditor can recover from
the debtor in the event that the debtor defaults.
407 | I S I W S C 2 0 1 9