Predicting Default More Accurately: To Proxy or Not to Proxy for Default?

Published date01 December 2019
AuthorKoresh Galil,Neta Gilat
Date01 December 2019
Predicting Default More
Accurately: To Proxy or Not
to Proxy for Default?*
Department of Economics, Ben-Gurion University, Beersheba, Israel
Previous studies targeting accuracy improvement of default models mainly
focused on the choice of the explanatory variables and the statistical
approach. We alter the focus to the choice of the dependent variable. We
particularly explore whether the common practice (in the literature) of using
proxies for default events (bankruptcy or delisting) to increase sample size
indeed improves accuracy. We examine four denitions of nancial distress
and show that each denition carries considerably different characteristics.
We discover that rating agencies effort to measure correctly the timing of
default is valuable. Our main conclusion is that one cannot improve default
prediction by making use of other distress events.
JEL Codes: G17; G33
Accepted: 17 April 2018
Previous studies targeting accuracy improvements of default models have
focused on nding the optimal set of explanatory variables or specifying the
optimal methodology used to estimate the likelihood of failure. A third aspect
concerning the prediction of nancial distress, which has not received much
attention, is the denition of nancial distress, which seems to vary signi-
cantly among different studies. Altman (1968) and Ohlson (1980) attempted to
forecast bankruptcy, which they identied with a rmsling of a bankruptcy
petition. Dichev (1998) outlined a broader denition of distress, by addressing
rms that were delisted because of poor performance as his sample for failed
rms. Shumway (2001) used bankruptcy and delisting events as an indicator of
nancial distress. Campbell et al. (2008) used a failure indicator that included
bankruptcy lings, delisting for nancial reasons, or receiving a D rating.
* The authors would like to thank Zvika Ak, Banita Bissoondoyal-Bheenick, Doron Kliger, Jan
Pieter Krahnen and Gal Zahavi, and seminar participants in Operational Research International Con-
ference (ORIC) 2012 meeting in Jerusalem, International Finance and Banking Society (IFABS) 2012
meeting in Valencia, and Multinational Financial Society (MFS) meeting in Prague for fruitful
© 2018 International Review of Finance Ltd. 2018
International Review of Finance, 19:4, 2019: pp. 731758
DOI: 10.1111/ir.12197
Bharath and Shumway (2008) dened distress as default; they obtained their
default data from the database of rm default maintained by Edward Altman
and by using the list of defaults published by Moodys. In this paper, we alter
the focus to the choice of the dependent variable and explore whether the com-
mon practice of using proxies for default events to increase sample size indeed
improves accuracy.
There are several possible reasons for researcherspractice of using proxies
for default events. First, a standard dataset of default events among US public
companies is nonexistent, and therefore, researchers rely on diverse sources for
the construction of their events lists. Second, the number of default events is
relatively small. Moreover, once such a list is intersected with other data
(e.g. accounting or market data), the nal set becomes even smaller. Under
these terms, it is tempting to use alternative distress denitions (proxies for
defaults) in order to expand the set of failure events.
The proximity of the default events to other types of negative events assists
in identifying such proxies. A nancial default is a state in which a debtor is
unable or unwilling to fulll the terms of a debt contract or a debt instrument.
Such an event may come after the occurrence of other negative events, such as
a rating downgrade and a major drop in the value of the equity. A default may
also precede other types of nancial distress events, such as bankruptcy ling
and delisting. Rating agencies exert effort in identifying default events and their
exact timing. Moodysdenition of default includes three types of credit
events: (i) a missed or delayed disbursement of interest and/or principal;
(ii) bankruptcy ling or legal receivership; and (iii) a distressed exchange.
time of default is set (by the rating agency) to be the earliest of the above events
because it is then that the major loss is recognized. Missed or delayed payments
and distressed exchanges normally precede bankruptcy lings, and therefore,
default events (as dened by rating agencies) normally precede bankruptcy
Moodys (2000) emphasizes that the alternative denitions of default
are not intended to broaden the central idea of nonpayment or bankruptcy, but
simply to get the timing right. Yet, Moodysdenition of default is also slightly
broader than the denition of bankruptcy because it also includes delayed pay-
ments. Such events do not necessarily lead to bankruptcies as debtors could be
repaid later on; however, rating agencies still consider them as default events
because of the meaningful opportunity costs they load on investors.
1 This denition appears in various default studies by Moodys. See, for example, Moodys
(2011, p. 61). S&P denition of default is similar. See, for example, Standard and Poors
(2011, p. 65).
2 A distressed exchange is an event in which the issuer offers bondholders a new security or
package of securities that amount to a diminished nancial obligation (such as preferred or
common stock, or debt with a lower coupon or par-amount) helping the borrower to avoid
the other types of default.
3 Brunner and Krahnen (2008) showed in the context of bank debt that private workout activi-
ties usually commence well before formal bankruptcy proceedings are initiated, and therefore,
a bankruptcy ling may be a late indicator of nancial distress.
© 2018 International Review of Finance Ltd. 2018732
International Review of Finance

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT