Вы находитесь на странице: 1из 26

The current issue and full text archive of this journal is available on Emerald Insight at:


Predicting corporate failure: Predicting

a systematic literature review of failure
methodological issues
Kingsley Opoku Appiah 461
Department of Accounting and Finance,
Received 17 April 2014
Kwame Nkrumah University of Science and Technology, Kumasi, Ghana Revised 26 May 2014
Accepted 24 July 2014
Amon Chizema
Department of Accounting and Finance, Loughborough University,
Loughborough, UK, and
Joseph Arthur
Department of Accounting and Finance,
Kwame Nkrumah University of Science and Technology, Kumasi, Ghana

Purpose – This paper aims to review the existing literature systematically so as to contribute towards
a better understanding of methodological problems of the classical statistical techniques, artificially
intelligent expert systems and theoretical approaches to solve the corporate failure syndrome.
Design/methodology/approach – This paper presented a systematic review of 83 articles reporting
137 prediction failure models published within 1966-2012 in scholarly reviewed journals in four main
disciplines, namely, accounting, finance, banking and economics. The authors performed the
systematic literature review with five main sources, namely, Science Direct, Google Scholar, Wiley
Interscience, Metalib, Web of Science and Business Source Complete of the Social Sciences. The review
modified the approaches used by Aziz and Dar (2006), Ravi and Ravi (2007) and Balcaen and Ooghe
Findings – The results indicate significant body of prior literature on prediction of corporate failure,
but a theoretically sound, highly accurate, simple and widely used corporate failure prediction model for
stakeholders has yet to be developed.
Originality/value – This paper contributes towards a systematic understanding of the
methodological problems associated with the statistical, artificially intelligent expert systems and
theoretical approaches to solve the corporate failure prediction problems faced by firms in 11 countries.
Keywords Systematic literature review, Artificially intelligent expert systems,
Corporate failure prediction, Multivariate discriminant analysis, Statistical techniques
Paper type Literature review

1. Introduction
Following Beaver (1966, 1968a, 1968b) and Altman’s (1968) seminal work, numerous
studies suggest that corporate failure prediction is theoretically explainable and
International Journal of Law and
empirically feasible (Scott, 1980). There is also a near consensus among stakeholders Management
that a corporate failure prediction model devoid of methodological issues, to some Vol. 57 No. 5, 2015
pp. 461-485
extent, may predict corporate failure with a high degree of accuracy. Moreover, the © Emerald Group Publishing Limited
recent financial scandals and failures of ENRON, WorldCom, Parmalat, Lehman DOI 10.1108/IJLMA-04-2014-0032
IJLMA Brothers and others worldwide have renewed interest on several methodological issues
57,5 which affect the predictive ability of existing models.
The literature on methodological issues associated with corporate failure prediction
models has been well-documented in a number of reviews. These include Scott (1980),
Zmijewski (1984), Taffler (1984), Altman (1984), O’Leary (1998), Keasey and Watson
(1991a, 1991b), Altman and Narayanan (1996), Dimitras et al. (1996), Morris (1997),
462 O’Leary (1998), Tay and Shen (2002), Aziz and Dar (2006), Balcaen and Ooghe (2006) and
Ravi and Ravi (2007).
Scott (1980), for example, reviews and integrates empirical predictions and
theoretical models, and finds that the overlap between these two strands of research is
substantial but not complete. Taffler (1984) evaluates published UK Z-score models and
points out the need for true validation samples in assessing model predictive
performance. Zmijewski (1984) examines the problem of choices-based sample and
sample selection biases, and concludes that the results from these techniques are similar
to non-random sample techniques. Altman’s (1984) international survey highlights
study designs, innovations and outcomes that are relevant to researchers and
practitioners. Altman (1984), however, stresses that the quality and reliability of models
reviewed would be improved as researchers and practitioners become more aware of the
methodological problems associated with these models (see also Altman and
Narayanan, 1996).
Similarly, O’Leary’s (1998) meta-analysis suggests that training proportion of
bankrupt firms in the data influenced the quality of the results, implying lack of
“upward” generalization. The deviation from a single hidden layer also exhibits a
negative impact on the relative quality of the models. Finally, O’Leary reports that time
negatively influences the relative quality of the neural network models. Keasey and
Watson (1991a, 1991b) highlight the uses and limitations related to the adoption of
prediction models in the management context. Dimitras et al. (1996) present a
comprehensive review of the articles published during 1968-2005, in the application of
statistical and intelligent techniques to solve the bankruptcy prediction problem faced
by banks and firms. Dimitras et al. (1996) observe that various methods appeared
mainly to overcome the limitations of multivariate discriminant analysis. Morris’s
(1997) review ignores artificially intelligent expert system (AIES) models.
Recently, Tay and Shen (2002) review the literature related to bankruptcy prediction
using rough sets model and, in the light of this review, suggest that the rough set model
is a promising alternative to the conventional methods. Aziz and Dar (2006) review 46
articles of corporate failure prediction and, in the light of this evaluation, provide a
ranking system to solve the problem of model choice in empirical application of
bankruptcy models. Ravi and Ravi (2007) review 128 articles published during
1968-2005, in the application of statistical and intelligent techniques to solve the
bankruptcy prediction problem faced by banks and firms. Ravi and Ravi’s (2007) review
framework highlights the data source, variables used, country of origin, sample periods
and the comparative performance of techniques. They conclude that statistical
techniques in a stand-alone mode are no longer used.
Finally, Balcaen and Ooghe (2006), the closest study to ours, contribute towards a
better understanding of methodological problems of the classical statistical failure
prediction models. In this respect, Balcaen and Ooghe (2006) identify the classical
paradigm, neglect of the time dimension of failure and the application focus as the main
related problems of the classical statistical techniques. Balcaen and Ooghe (2006) also Predicting
observe the use of financial data and a linear classification rule as well as the neglect of corporate
the multidimensional nature of failure as other related problems of the statistical
The above analysis suggests that early reviews of this literature are limited in scope
(Tay and Shen, 2002; Aziz and Dar, 2006; Balcaen and Ooghe, 2006; Ravi and Ravi, 2007),
out of date (Scott, 1980; Altman, 1984; Keasey and Watson, 1991a, 1991b) or both 463
(Taffler, 1984; O’Leary, 1998; Dimitras et al. 1996; Morris, 1997). In addition, prior
reviews are largely unconnected, which, in turn, is counterproductive. Thus, we invoke
O’Leary’s (1989) notion that there is little direct cross-fertilization of different reviews
and streams of research work in the field of corporate failure prediction. More
importantly, to date, there is no systematic literature review (SLR) that contributes
towards a better understanding of methodological issues that integrates corporate
failure prediction models from the three strands, namely, classical statistical techniques,
AIESs and theoretical approaches. Our study contributes to the existing body of
accounting and finance literature by filling this gap.
Following the approach by Biolchini et al. (2005), the aim of this review is
accomplished through a SLR. According to Nicolas and Toval (2009, p. 1292):
SLR is a research technique used to analyse the state of the art in a specialised field of
knowledge by formally defining the problem statement, the sources of information, the search
strings, the criteria for inclusion and exclusion of the papers found in the searches, the
quantitative analysis to be undertaken (if necessary), and the templates for ordering the
information collected from the papers.
The paper proceeds as follows: Section 2 summarizes the design of SLR. Section 3 deals
with the results and discussions. Section 4 outlines the limitations of the SLR. Section 5
concludes the SLR with recommendations for further research.

2. Planning the SLR

Following Biolchini et al. (2005) and Nicolas and Toval (2009), the review protocol
proceeds as follows. Section 2.1 contains the scope of the SLR. Section 2.2 outlines the
research questions of this SLR. Section 2.3 highlights the planning of the literature
search process. Section 2.4 summarizes the exclusion and inclusion criteria. Section 2.5
describes the quality assessment of the SLR. Section 2.6 reports the data collected from
the selected studies. Finally, the data analysis is the focus of Section 2.7.

2.1 Scope of the SLR

SLR is limited to reviewing scholarly studies on corporate failure prediction from three
broad research paradigms, namely, statistical techniques, intelligence techniques and
theoretical foundation. Table I contains an overview of related methods of the three
research paradigms discussed in Section 4.
Table I is logically structured into four main columns. First, Column 1 outlines the
various related methods under each paradigm. In particular, the respective author(s)
who initiated the method in the corporate failure context is (are) stated as well. A brief
description, strengths and limitations of each method is the focus of Columns 2, 3 and 4,


Table I.

corporate failure
Three perspectives of
Dimension Statistical technique (1-5), AIES (6-16) and theoretical (17-19)
models/Landmark study Description Major strength Major limitation/critics

Univariate, Beaver (1966) If a firm’s value for a ratio is higher than a certain cut-off Simple and easy to apply Neglect multi-dimensional nature of failure
point, this signals strong financial health and vice versa (Zavgren, 1983; Keasey et al., 1991)
MDA, Altman (1968) Linear combination of certain discriminatory predictors Constructs a discriminant function (Z-score) by maximizing the Demanding assumptions, linear separability,
in the form of Z-score ratio of between-groups and within-groups variances (Fisher, multivariate normality and equal and within-group
1936) covariance and others (Eisenbeis and Avery, 1972)
LPM, Meyer and Pifer (1970) Expresses the probability of firm failure as a Estimates the odds of firm’s failure with probability Results in model considered complex for the average
dichotomous dependent variable person to interpret
Logit, Martin (1977) Replaces the LPM distribution with a logistic cumulative Same as LPM Results in models considered complex for the
distribution average person to interpret
Probit, Zmijewski (1984) Estimate the parameters of the linear model by the Replaces the logistic function with normal distribution function Several assumptions: dependent variable categorical,
maximum likelihood estimation procedures error term has a cumulative normal distribution and
others (Aldrich and Nelson, 1984)
CUSUM, Kahya and The time-series behaviour of the attributes variables for CUSUM model has a very “long memory” in the case of bad CUSUM model has a very “short memory” in the
Theodossiou (1999) each of the failed and non-failed firms is estimated by a performance of the firm case of past good performance
finite-order VAR model
Partial adjustment View failure as firm’s inability to pay financial Failing firms exhibit smaller absolute elasticity of cash Too narrow
processes, Laitinen and obligations as they fall due balances compared to non-failed firms
Laitinen (1998)
MDS, Neophytou and Produces graphical representations of the main Allows the inclusion of qualitative information and assesses Practitioners interested in the health of companies
Molinero (2004) characteristics of the data the reasons of the probability of specific firm’s failure not included in the original sample, must add these
Neural networks, Bell et al. Perform classification tasks in a way intended to emulate Ability to induce algorithms for recognizing patterns (Zhang NN models are characterized as “black boxes” which
(1990) Khanna (1990) brain processes et al., 1999) decision makers may be reluctant to rely upon due to
a lack of transparency
CBR, Kolodner (1993) Solves a neo-classification problem with the help of Fits conveniently in the bankruptcy context due to its four- 10.3 Is purely based on the assumption that similar
similar prior solved cases stage procedure cases are useful for predicting the outcome of the
new case
K-nearest neighbour (kNN), Distribution-free technique applicable under less A non-parametric method for classifying observations that Same as NN
Tam and Kiang (1992) restrictive conditions regarding population distribution relaxes the normality assumption as well as eliminates the
and data measurement scales functional form required in DA and logit
Genetic algorithm (GANN), GA works as a stochastic search technique to find an Optimized linear functions without restrictive statistical Same as NN
Kingdom and Feldman optimal solution to a given problem from a large number hypotheses such as normality of explanatory variables or the
(1995) of solutions equality of the variances/co-variances matrix
Rough set, Slowinski and Classify objects using imprecise information Result in a set of easily understandable decision rules which Same as NN
Zopounidis (1995) are supported by a set of real example
Survival analysis, Lane et al. Appropriate when measuring the time of event (Cox and Provides us with a possibility to model dynamic aspects of the The date of annual closing of accounting is not
(1986) Oakes, 1984) failure process necessarily a natural starting point for the failure
Dimension Statistical technique (1-5), AIES (6-16) and theoretical (17-19)
models/Landmark study Description Major strength Major limitation/critics

Data envelopment analysis Designed to assess the efficiency of decision-making Can study the frontier shift over time horizon, which in turn Same as NN
(DEA), Premachandra et al. units (DMUs) that have multiple inputs and outputs allows us to explore the dynamic change of corporate failure on
(2009) a time horizon
Iterative dichotomizer 3 ID3 method creates a decision tree that properly Like RPA, ID3 employs a non-backtracking splitting procedure, It assumes that the entire space of possible events
(ID3), Quinlan (1979) classifies the training sample which in turn maximizes the entropy of the split subsets begins as a single category (Braun and Chandler,
Balance sheet decomposition Examine changes in the structure of the balance sheet Significant changes in the composition of assets and liabilities, Assumes that firms try to maintain equilibrium in
measure, Lev (1973) indicates a firm is incapable of maintaining the equilibrium their financial structure
Gambler’s ruin theory, Scott Views the firm as a gambler The gambler, playing repeatedly with some probability of loss, Assumes a net positive probability that firm’s cash
(1980) continues to play until its net worth goes to zero(failure) flows will be consistently negative over a run of
Cash management theory, An imbalance in cash flows depicts failure of firm’s cash Short-term management of firm cash flow is a major concern Too simple
Aziz et al. (1988) management function, persistent of which may cause for a firm’s going concern
financial distress and ultimate failure

Note: A similar approach has been used by Aziz et al. (2006)

Table I.
IJLMA 2.2 Research question of the SLR
57,5 This paper integrates the approaches of prior reviewers (Aziz and Dar, 2006; Ravi and
Ravi, 2007; Balcaen and Ooghe, 2006) with the objective of answering the question:
What are the methodological issues in the literature? As discussed earlier, we achieve
our aim through the SLR approach, which involves the analysis of corporate failure
prediction uncovering gaps, both theoretical and empirical, in the process to facilitate
466 future research in the field of corporate failure prediction.

2.3 Search process of the SLR

We select five main sources to perform the SLR. These are as follows:
(1) Google Scholar (http://scholar.google.co.uk);
(2) Wiley Interscience (http://onlinelibrary.wiley.com);
(3) Science Direct (www.sciencedirect.com);
(4) Web of Knowledge (via Thomson Reuters); and
(5) Business Source Complete (http://web.ebscohost.com).

Our initial search in these sources for corporate failure prediction articles revealed that
authors used bankruptcy, insolvency, liquidation, financial distress and dissolution as
synonyms for corporate failure. Hence, we design a search framework that encapsulates
all the terms identified. In this respect, all the string defined is the following:
(“Prediction” OR “Predicting” OR “Forecasting”) AND (“Liquidation” OR “Bankruptcy”
OR “Insolvency” OR “Dissolution” OR “Failure”) AND (“Corporate” OR “Firms” OR

2.4 Inclusion and exclusion criteria

Generally, the search string depicts that it is sufficient to read the title of the article to
consider its inclusion in the SLR. However, in some cases, reading the title alone is not
sufficient. Confronted with such a situation, the abstract is read and, if need be, the
introduction and/or conclusion. In extreme cases, the whole article is read to consider its
inclusion in the SLR. For this reason, the full texts of the article must be present in an
electronic format and be written in English.
In contrast, to ensure the SLR is within reach, we devise exclusion criteria as follows.
First, we exclude articles with synonyms titles such as organization mortality and death
as well as organizational exit, decline, retrenchment, downsizing, setbacks and financial
distress. Thus, the SLR invokes Greenhalgh et al.’s (1988) assertion that these should not
be conceptualized as failure. Second, books, PhD Thesis, working papers, technical
reports as well as articles published in conference proceedings are excluded due in part
to the SLR quality bias. Accordingly, each article is scrutinized further to ensure that it
was published in a peer-reviewed journal.

2.5 Quality assessment of the SLR

The quality of the selected studies in Section 2.4 is addressed as follows:
• Publication place: Articles must have been published in at least Grade-3 ranking
journal as per the Association Business School Academic (ABS) Journal Quality
Guide Version 4 2010 Edition (hereafter the Guide) due in part to researchers’
quality bias. Thus, as ABS (2010) notes, the guide is a hybrid based partly on peer Predicting
review, citation and editorial judgement. corporate
• Estimation and validation procedures: In addition to the point above, papers failure
selected must have used reasonable estimation sample or validated models with
either ex-ante or ex-post data or Lachenbruch test (Collopy et al., 1994).
• Finally, each paper and its resultant model(s) pass through an in-depth scrutiny
on four critical dimensions, namely, robustness, predictive accuracy, adaptability 467
and explanatory capability.

2.6 Data collection of the SLR

An integrated template designed after taking a cue from prior reviewers, particularly
Ravi and Ravi (2007), Aziz and Dar (2006) and Balcaen and Ooghe (2006), is filled for
each selected work. The form consists of the research attributes such as results,
techniques used and methodological issues such as purpose, sample, variables,
matching, origin and operational definition, among others.

2.7 Data analysis of the SLR

The data collected are formulated to depict:
• The identifier allocated to the study in the SLR, its authors, bibliographic
reference, origin and year of publication.
• The organization of the study following the taxonomy suggested (presented in
Section 3.2).
• The aim for the research from prior researchers and models respective results are
• The research technique and method used to develop the failure prediction model.

Consistent with prior researchers (Aziz and Dar, 2006; Dimitras et al., 1996), papers that
used more than one method are counted as more than one empirical investigation. From
this point, the scene is set for a discussion of the results in Section 3 and particularly the
search results in Section 3.1.

3. Results and discussion of the SLR

3.1 Search results and deviations from protocol
The search string is used in the search engines of all sources. Table II presents a
summary of the number of papers found per source, candidates and selected studies.
Thus, the inclusion and exclusion criteria are used to identify the candidate and selected
studies, respectively. For parsimonious reasons, we present only the selected studies in
Table II.
We also identify 11 scholarly reviewed articles and one book (see Section 1 above)
through the search process. We read articles and book with a critical eye to generate a
picture of the state of the art about the SLR subject. The bibliographies of the prior
reviewed articles reveal relevant papers in aid of the research questions of the SLR.
Accordingly, the SLR includes ten papers that satisfy the quality criteria, but two:
Altman (1973) and Wilcox (1973). We acknowledge that this is a deviation of protocol
that concerns repetition of the SLR. However, the interest of working with more
complete set of papers must prevail.
IJLMA Source Studies found Candidate studies Selected studies
Google scholar 142 57 11
Wiley interscience 32 13 10
Science direct 11,357 32 23
Web of knowledge 54 17 16
468 Business source complete 128 13 13
Other-deviation from protocol 10
Total 11,713 132 83
Table II.
Search results Note: Identical studies in different sources have not yet been eliminated

Consequently, the present SLR covers 83 articles reporting 137 prediction failure models
(Table II), published in scholarly reviewed journals in nine main disciplines. The field of
study is as follows:
(1) finance (38);
(2) information management (30);
(3) operations research and management (32);
(4) accountancy (22);
(5) economics (4);
(6) general management (4);
(7) marketing (3);
(8) entrepreneurship small business management (3); and
(9) international business studies (1).

This suggests that corporate failure prediction is a multi-disciplinary field. In

particular, 16, 14, 14, 13, 10 and 9 studies are published in the following Journals,
• Expert System Application;
• Omega: The International Journal of Management Science;
• Business Finance Accounting;
• Decision Support Systems;
• Banking and Finance; and
• Accounting Research.

Furthermore, selected studies originate from 11 countries, with 53 per cent, 15 per cent
and 11 per cent of the studies using data set from the USA, Korea and the UK,
respectively. In addition, Australia, Belgium and France account for 4 per cent each,
whereas the rests of the world account for the remaining 9 per cent (i.e. Finland 3 per
cent, Greece 2 per cent, Italy 2 per cent, Brazil 1 per cent and Spain 1 per cent). These
disclosures have paved the way for an in-depth discussion of the methodological issues
in the extant literature, in the next sub-section.
3.2 Synthesis of the proposals Predicting
The purpose of this section is to provide a synthesis vision of the field of knowledge corporate
addressed by the SLR questions, to facilitate future research in this area. In particular,
this section answers the main question of the SLR: What are the methodological issues
in the literature? In this respect, the attributes of selected papers in this SLR are
tabulated in Table III[1]. From this point, taxonomy is proposed for the papers selected
in this SLR based on the three steps used by prior researchers to achieve their respective 469
contribution. These steps are as follows:
(1) introductory;
(2) application focus; and
(3) data collection.

We will discuss each in turn.

3.2.1 Introductory stage. We identify four common objectives from prior studies.
First, most researchers test the predictive ability of variables and/or statistical
technique (Beaver, 1966, 1968a, 1968b; Altman, 1968, 1973; Casey and Bartczak, 1984).
Second, others (Tam and Kiang, 1992) in consonance with criticisms of Joy and
Tollefson (1975) test the predictive accuracies of various statistical methods. Third,
others also test the predictive abilities of prior models, particularly, Altman’s (1968)
model (Deakin, 1972; Moyer, 1977; Booth, 1983). Four, other researchers focus on testing
the validity of methodological issues (Altman et al., 1977; Dambolena and Khoury, 1980;
Mensah, 1983; Back et al., 1996).
Nonetheless, to avoid overlaps, we follow Dietrich (1984) and, in turn, categorize the
aims of prior papers into two, namely, testing the association between financial distress
measures (R1), and developing models to predict corporate failure (R2). In this respect,
the former and the latter account for 85 per cent and 15 per cent, respectively. This
suggests that future research direction must focus on the development of models to
predict failure.
3.2.2 Issues associated to the application focus of models. We observe that prior
models are application-driven, implying that researchers apply an outsider’s viewpoint
and a financial approach to explain the company failure phenomena (Cybinski, 2000). In
other words, models are the outcome of a statistical search through a number of credible
traditional accounting ratios (Scott, 1980; Balcaen and Ooghe, 2006). Cybinski (2000)
contends that corporate failure prediction models are the result of “putting the cart
before the horse”. This suggests two main issues, namely, neglect of non-financial
information, and ad hoc selection of variables. We will discuss each in turn. Neglect of non-financial information. The neglect of non-financial information
remains one of the major gaps in the extant corporate failure literature. From this point,
there is a general agreement that annual accounts information in the form of financial
ratios (82 per cent) provides the best predictions of corporate failure (Aziz and Dar,
2006), due to the notion that financial ratios are objective measures based on publicly
available information (Laitinen, 1992). However, there is little agreement about the most
suitable ratios to use from, accrual-based financial ratios (Casey and Bartczak, 1984),
cash-based ratios (Gentry et al., 1987; Aziz et al., 1988) or both (Gentry et al., 1985). The
list of ratios[2] extracted from the papers reviewed suggests that the most frequently
used ratios are working capital/total assets, EBIT/total assets, total debt/total assets, net



Table III.
Summary of
attributes of papers

S001 Beaver (1966) R1 FR D5D6 23 5 79:79 Mix USA UA 90 22 5 34 8 H-out

S002 Altman (1968) R1 FR D5 23 5 58:99 M USA LDA 96 6 3 4 21 H-out
S003 Deakin (1972) R2 FR D5 123 5 43:55 M USA LDA 97 3 3 18 23 Ex-ante
S004 Casey et al. (1984) R1 FR D5 2 5 60:230 Mix USA UA 75 17 27 3 35 H-out
S005 Casey et al. (1984) R1 FR D5 2 5 60:230 Mix USA LDA 86 17 13 37 11 H-out
S006 Casey et al. (1985) R1 MIX D5 12 5 60:230 Mix USA LDA 87 13 13 13 33 H-out
S007 Casey et al. (1985) R1 MIX D5 12 5 60:230 Mix USA Logit 88 37 5 60 5 H-out
S019 Sharma and Mahajan (1980) R2 FR D5 23 5 36:36 R USA LDA 92 nr nr nr nr H-out
S020 Wilcox (1973) R1 FR D5 n/a n/a n/a n/a USA Ruin n/a 0 0 0 0 H-out
S021 Dambolena and Khoury (1980) R1 FR D5 12 5 23:23 R&M USA LDA 91 17 0 17 13 H-out
S022 Altman et al. (1977) R2 FR D5 12 3 23:35 Mix BRA LDA 91 12 7 21 15 H-out
S023 Altman (1977) R2 FR D5 6 5 56:107 S&L USA QDA 96 4 4 4 8 H-out
S026 Altman and Loris (1976) R2 MIX D5 6 1 40:113 Brokers USA QDA 90 10 10 18 12 H-out
S035 Lo (1986) R1 FR D5 123 1 38:38 Mix USA LDA nr 0 0 0 0 None
S036 Lo (1986) R1 FR D5 123 1 38:38 Mix USA Logit nr 0 0 0 0 None
S039 Santomero and Vinso (1997) R1 FR D5 6 nr 20:20 Utilities USA Ruin nr 0 0 0 0 None
S040 Blum (1974) R2 FR D5 123 8 115:115 Mix USA LDA 94 7 10 20 15 H-out
S041 Gentry et al. (1985) R1 FR D5 123 3 56:56 Mix USA Logit 83 21 12 26 30 H-out
S042 Aziz et al. (1988) R1 FR D5 23 5 49:49 Mix USA LDA 89 0 0 0 0 H-out
S043 Aziz et al. (1988) R1 FR D5 23 5 49:49 Mix USA Logit 92 14 2 14 16 H-out
S044 Begley et al. (1996) R1 FR D5 23 1 130:2600 Mix USA LDA 78 22 22 19 25 H-out
S045 Begley et al. (1996) R1 FR D5 23 1 265:3300 Mix USA Logit 72 12 17 11 27 H-out
S048 Mensah (1983) R1 FR D5 23 1 41:65 M USA LDA 100 0 0 54 31 Ex-ante
S049 Mensah (1983) R1 FR D5 23 1 41:65 M USA Logit 97 0 0 0 0 Ex-ante
S053 Jones and Hensher (2004) R1 FR D5D6 6 5 423:7818 Mix AUS Mxlogit 99 0 0 0 0 Ex-ante
S054 Wilson and Sharda (1994) R1 FR D5 23 1 65:64 Mix USA LDA 89 17 6 7 8 H-out
S055 Wilson and Sharda (1994) R1 FR D5 23 1 65:64 Mix USA BPNN 100 0 0 4 4 H-out
S056 Jones and Hensher (2004) R1 FR D5D6 6 5 423:7818 Mix AUS MNLogit nr 0 0 0 0 H-out
S073 Luoma and Laitinen (1991) R1 FR D1 23 1 36:36 SME: R&Ind FIN Logit 72 27 29 0 0 None
S074 Lussier (1995) R1 NFI D5 235 1 108:108 SMEs USA Logit 69 27 27 0 0 None
S075 Frydman et al. (1985) R1 FR D5 6 1 58:142 R&M USA LDA 70 18 12 0 0 None
S076 Pompe and Bilderbeek (2005) R1 FR D5 6 5 1369:3000 SMEs BEL LDA 76 29 25 29 54 H-out
S077 Pompe and Bilderbeek (2005) R1 FR D5 6 5 1369:3000 SMEs BEL NN 77 nr nr 0 0 H-out
S108 Ohlson (1980) R1 FR D5 6 3 105:2058 Mix USA Logit 96 11 27 12 17 H-out
S117 Tam and Kiang (1992) R1 FR D5 3 2 81:81 Banks USA BPNN 96 0 8 9 12 Jack
S118 Tam and Kiang (1992) R1 FR D5 3 2 81:81 Banks USA ID3 92 10 5 21 17 Jack

S119 Tam and Kiang (1992) R1 FR D5 3 2 81:81 Banks USA DA 89 0 22 17 11 Jack

S120 Tam and Kiang (1992) R1 FR D5 3 2 81:81 Banks USA Logit 92 9 7 12 17 Jack
S121 Tam and Kiang (1992) R1 FR D5 3 2 81:81 Banks USA kNN 69 36 25 19 31 Jack
S126 Frydman et al. (1985) R1 FR D5 6 1 58:142 R&M USA RPA 84 14 2 0 0 None

Note: A similar approach has been used by Aziz et al. (2006)

Table III.
IJLMA income/total assets, quick assets/current liabilities and current assets/current liabilities
57,5 (Dimitras et al., 1996).
There are disadvantages, however, of using financial ratios. First, small and medium
enterprises (SMEs) in most jurisdictions are not obliged to publish accounts, suggesting
that prior studies are limited to listed firms (95.63 per cent). In this respect, expect for
Luoma and Laitinen’s (1991) analysis in unincorporated entities, where the incidence of
472 failure is greater (Altman, 1968), literature on unincorporated entities is distinctively
lacking. Second, information from financial statements may not necessarily be true and
fair. For example, accounting measures are subject to manipulation due to the
accounting policies elected in respect of depreciation, inventory, revenue, expenditure
items and consolidation of accounts (Argenti, 1976; Chakravarthy, 1986; Keasey and
Watson, 1987; Rosner, 2003). From this point, third, models may be contaminated due in
part to occurrence of extreme ratio values, errors and missing values (Balcaen and
Ooghe, 2006). Finally, failure prediction models based on financial ratios assume that all
relevant failure indicators are reflected in the annual accounts. For that reason, Zavgren
(1985) contends that any econometric model containing only financial statement
information will not predict with certainty the failure of a firm (Argenti, 1985). Argenti
(1985) notes financial ratios give signpost on the verge of bankruptcy, but the crucial
question of why the signpost is not within its scope.
Accordingly, few researchers (17 per cent) such as Argenti (1976), Altman et al.
(1977), Ohlson (1980), Peel et al. (1986) and Peel and Peel (1988) favour financial and
non-financial indicators (Mix) in the corporate failure context. In fact, Lussier (1995),
using exclusively non-financial indicators from 108 matched paired firms, demonstrates
that professional advice, education, staffing and parent are significantly different
between the successful and failed US SMEs. Again, Peel et al. (1986) and Peel and Peel
(1988) show the ability of the timeliness of financial reporting to predict financial
distress, using UK data and ad hoc selection of variables.
We observe that 94.89 per cent of models in the corporate failure literature are
based on ad hoc selection of variables through statistical (60.58 per cent) and AIES
(34.31 per cent) techniques. Thus, model estimation is purely based on empiricism
due in part to the lack of real economic theory in the identification of variables. This
confirms Taffler’s (1983) notion that there is no coherent theory underpinning the
use of financial ratios and only very tenuous guidance on the appropriate measures
in different situations (Taffler, 1983). Thus, there is no uniformity in variable
selection and definition in the above studies, to answer the practical questions of
why business fails.
In this respect, researchers mostly develop criteria such as:
• all ratios used in a previous study, mostly Altman (1968) (Deakin, 1972; Collins,
1980; Sharma and Mahajan, 1980; Salchenberger et al., 1992);
• ratios popularity and predictive success in previous studies;
• analyst suggestions based on their experience;
• researchers’ speculation; and
• data availability for computation of ratios.

From this point, Beaver (1968b) notes that the selection of variables based on popularity
may be problematic because popular ratios are more likely to be subject to
window-dressing and, hence, are more likely to be unreliable. Accordingly, empirical Predicting
considerations are used to reduce the large set of potential financial ratios into seven key corporate
sets of decision dimensions: profitability, capital intensiveness, financial leverage,
short-term liquidity, cash position, inventory intensiveness and receivables
intensiveness (Pinches et al., 1973). Neophytou and Molinero (2004) argue that the
weakness associated with this brute empiricism approach is mitigated to some extent by
validating the model with a holdout sample from a future time. 473
Critics, e.g. Blum (1974), however, suggest that this brute empiricism approach
accounts in part for the numerous data-specific variables (Zavgren, 1985), which may
result in counter-intuitive signs for some coefficient (Keasey et al., 1990). Blum (1974), for
example, argues that in the absence of a theory of symptoms, researchers cannot use
statistical analysis of accounting ratios and anticipate a sustained correlation between
explanatory variables and bankruptcy prediction. Scott (1980, p. 1) has gone as far as
saying, “these models do not command full professional acceptance, in part because they
lack the underpinnings of an explicit and well-developed theory”.
This SLR notes the classical papers of Blum (1974), Gentry et al. (1985) and Aziz et al.
(1988) as the few exceptions to the present predicament. This, in turn, suggests that
future researchers must consider theoretical arguments to select suitable financial and
non-financial variables for their models.
3.2.3 Data collection stage. The data collection stage highlights six main issues, namely:
(1) arbitrarily defining the term corporate failure;
(2) arbitrarily matching technique;
(3) sample selectivity issues;
(4) choice of optimization cut-off point;
(5) assessment of classification results; and
(6) data sources.

The subsections below will explain each of these issues in detail. Arbitrarily defining of the term “Corporate failure”. The definition of the term
corporate failure of each study was coded on a scale of D1 to D13, using Morris’ (1997)
spectrum of potential indicators of corporate financial distressed. These are as follows:
creditors’ or voluntary liquidation, appointment of receiver (D1), suspension of stock
exchange listing (D2), going concern qualification by the auditor (D3), composition with
the creditors (D4) and protection sought from creditors (D5). Others are breach of debt
covenants (D6), company reconstruction (D7), resignation of director (D8), take-over
(D9), closure or sale of part of business (D10) and a cut in dividend or reporting of losses
(D11). The rest are the reporting of profits below a forecast and capital becoming zero or
negative for D12 and D13, respectively.
Table III records 71 per cent and 13 per cent for D5 and D1, respectively. This, in turn,
suggests that there is a general agreement on the legal definition of corporate failure,
bankruptcy (D5) or insolvency (D1), in the USA or the UK, respectively. This definition
allows an objective criterion for dating the failing firms and splitting the sample into
failed and non-failed sub-samples (Charitou et al., 2004).
In contrast, researchers develop economic bankruptcy models but not legal
bankruptcy models (Shumway, 2001; Hillegeist et al., 2004). Hence, researchers (Beaver,
1966, Lau, 1987; Theodossiou, 1991; Altman et al., 1994; Lee et al., 1996; Jones and
IJLMA Henser, 2004; DuJardin and Séverin, 2011) broaden the legal definition to include more
57,5 than one of the Morris’ (1997) spectrum. Balcaen et al. (2006) concur, asserting that the
definition of corporate failure is artificial and applied inappropriately to the topic of
corporate failure prediction.
The definition of failed and non-failed companies is essential to minimize possible
outliers, and in turn Type I and II errors. The quest to minimize the latter errors due in
474 part to the heterogeneous nature of non-failed firms results in a careful selection of
non-failed firms (Altman and Loris, 1976; Taffler, 1983; El Hennawy and Morris, 1983).
Further, researchers prefer the legalistic definition to avoid issues as business closure,
delisting, mergers and acquisition termed as failures in part to minimize the Type I
errors. Watson and Everett (1996) suggest that closure firms may have sound financial
position, but closed for other reasons, is germane here. Arbitrarily matching technique. Having arbitrarily defined the term failure,
researchers’ next key task is to devise a procedure for matching the failed to the control
sample. For clarity, we code the matching techniques from prior papers on a scale of 1 to
8, for year, industry, size, inventory and depreciation method, location, random, bias,
industry averages, respectively. Table III shows that 63, 58, 49, 23, 15, 4, 2 and 1 studies
used size (e.g. assets, capital size, employees, etc.), industry, random, year, bias, location,
inventory and depreciation method and industry averages, respectively, for matching
failed and non-failed firms. This result indicates overlap due in part to combination of
one or two items in the matching scale. With the exception of Aly et al. (1992), the SLR
has enough evidence to support the assertion that size and industry dominate the
matching technique in the extant literature.
There are disadvantages, however, to match by size and industry. First, this leads to
sample selection bias (Martin, 1977; Zmijewski, 1984) and, in turn, asymptotic bias
parameter and probability estimates (Menski and Lerman, 1977). On one hand,
matching on the basis of size leads to too many small companies in the non-failed
sample. On the other hand, matching on the basis of industry will lead to too many
companies from recession-hit industries in the failed sample (Lennox, 1999). In this
respect, the estimation samples of failed and non-failed firms are not representative of
the whole population of firms (Ooghe et al., 1995). Second, the use of a relatively small
sample could lead to over-fitting, implying that the model is not stable over time (Platt
and Platt, 1990). Third, predictive accuracy of the estimated model is misleading. Casey
and Bartczak (1985) agree, emphasizing that matching by size would limit the
generalizability of the study’s findings in the predictive context. Four, it is not possible
to investigate the effects of industry and/or firm size on the probability of corporate
failure (Jones, 1987). In fact, Morris (1997) posits that the problem of sampling bias offers
an explanation why the market does not appear to behave in the way expected.
In contrast, Zmijewski (1984) argues that these estimation techniques do not appear
to provide different qualitative results from those provided by random sampling
technique. From this point, Ohlson (1980) notes that appropriate criteria for matching
are not obvious. Nevertheless, a sizeable proportion of the literature follows the classical
random sampling design (q, 1977; Taffler, 1983; Frydman et al., 1985; Etheridge et al.,
2000;Jones and Henser, 2004; Kim and Kang, 2010). Sample selectivity issues. The sample selectivity issues are discussed
logically as follows: neglect of the time dimension of corporate failure, neglect of large
size and arbitrary sample period.
• Neglect of the time dimension to corporate failure: The lack of data for the failed Predicting
companies, especially in developing countries, accounts for the numerous static corporate
models, one-year data. The SLR results register 63 and 32 out of 136 models
consider one- and five-year data, respectively. Furthermore, the US and the UK
studies account for 71 per cent and 16 per cent of the five-year prior models
reported in Table III, respectively.
There are demerits, however, of the dominance one-year failure prediction models. 475
First, these models are based on the following assumption: consecutive annual
accounts are independent, repeated measurements (i.e. uniform failure-process),
and failure is a steady state (Luoma and Laitinen, 1991). Second, application of the
static model to consecutive annual accounts of one particular firm may result in
conflicting predictions (Keasey et al., 1990; Luoma, and Laitinen, 1991), in an
ex-ante predictive context due in part to the neglect of time-series behaviour.
• Neglect of large sample size: The bulk of the literature is filled with small sample
size models. The few notable exceptions are Ohlson (1980), Min and Lee (2005),
Pompe and Bilderbeek (2005), Park and Han (2002), Hensher and Jones (2007),
Jardin and Séverin (2011) and Sánchez-Lasheras et al. (2012).
To overcome the limitations of small sample size, data are mostly gathered from
mix industries (51 per cent). Beaver (1966), Boritz et al. (1995) and Charitou et al.’s
(2004) models with data set from different industries record accuracies as high as
90 per cent (87 per cent), 99 per cent (99 per cent) and 96 per cent (86 per cent) for
estimation (validation), respectively, despite the presence of industry-sensitive
• Arbitrarily sample period: The definition of corporate failure is theoretical to a
certain arbitrary chosen year or period (Balcaen and Ooghe, 2006). Table III
records a minimum of 8 months (Hyghebaert et al., 2000) to maximum of 22 years
(Wilcox, 1973), for sample period for prior studies. On one hand, the time frame
must exceed one year to facilitate more samples for failed firms. Morris (1997)
suggests that 2 per cent of listed companies failed in each year, is germane here.
On the other hand, two main assumptions are vital for models reliability, namely,
data stationarity and stability (Altman and Eisenbeis, 1978; Zmijewski, 1984).
Stationarity implies stable relationship, first, within the independent variables
(Edmister, 1972) and, second, between the dependent and independent variables
(Mensah, 1984; Jones, 1987). While, data stability requires no change in
macroeconomic environment (Moyer, 1977; Mensah, 1983).

The synthetic definition of failure which is applied to a subjective period leads to

selection bias and/or contaminated population (Shumway, 2001; Taffler, 1983, 1983).
Accordingly, models are valid for sample-specific data within the chosen period (Ooghe
et al., 1995). Thus, the models are sensitive to the sample data examined and, in turn,
their predictive power shrinks with ex-ante sample (Moyer, 1977).
Attempts to overcome data instability include but are not limited to stability
measures (Altman et al. 1977; Dambolena and Khoury, 1980; Betts and Belhoul, 1987;
Pompe and Bilderbeek, 2005), industry-relative ratios (Platt and Platt, 1990; El Hennawy
and Morris, 1983) and deflated financial ratios (Mensah, 1983; Yang et al., 1999). The
results from these studies are mixed. For example, Yang et al. (1999) have evidence to
IJLMA support Platt et al.’s (1994) assertion that deflation improves the discriminant ability of
57,5 bankruptcy prediction. Dambolena and Khoury (1980) and Betts and Belhoul (1987) also
suggest that inclusion of standard deviation of ratios improves considerably the ability
of the discriminant function of large firms but not in the ex-ante results. On the other
hand, Platt and Platt (1990) suggest that industry-related variables do not significantly
increase the stability of the estimated coefficients. Also, Pompe and Bilderbeek’s (2005)
476 result is at variance with Dambolena and Khoury (1980). Choice of the optimization criteria. Another issue in the extant literature is the
choice of the optimization criteria. A failure prediction model depends largely on the
choice of the optimization measures and the modelling method. In this respect, there is
no superior modelling method and optimization cut off-point, both being exclusively left
to the researcher. In addition, researchers attempt to optimize various goodness-of-fit
measures such as the percentage of correct classifications, Type I or Type II
classification error, the Gini-coefficient, pseudo R-square and maximum likelihood
(Zavgren, 1985; Ooghe and Balcaen, 2002). Balcaen and Ooghe (2006) posit that there is
little point in finding a model that minimizes a performance measure that is irrelevant to
one’s aims. Koh (1992) contends that the non-consideration of misclassification costs in
earlier studies does not appear to be a serious limitation, although their consideration
may lead to different optimal cut-off points.
To mitigate the effects of this drawback, researchers validate their models with
ex-ante holdout sample and/or Lachenbruch Jack-knife validation test (Joy and
Tellefson, 1975). The rationale is to test for both over fitting and violation of the
stationarity assumption (Zavgren, 1983; Zmijewski, 1984; Mensah, 1984). Jones (1987),
however, is not in favour of the Lachenbruch method, as it does not test external validity.
The inter-temporal test, which is equivalent to the Lachenbruch Jack-knife validation
test, is a statistical technique widely accepted for validation of relatively small sample
size. We record 59.85 per cent, 13.14 per cent and10.95 per cent for holdout test, ex-ante
test and Lachenbruch test for the extant literature, respectively. As well, a sizeable
number (16.06 per cent) of models are not validated. This confirms the notion that
existing failure prediction models suffer from non-stationarity and data instability
problems (Moyer, 1977; Mensah, 1984; Charitou et al., 2004; and Balcaen and Ooghe,
2006). Mensah (1983), El Hennawy and Morris (1983), Taffler (1983), Peel and Peel (1988)
and Shin and Lee (2002) are among the few exceptions of prior work in favour of ex-ante
sample for validation. We, however, observe that prior models’ predictive ability is
questionable with ex-ante data. Assessment of classification results of prior studies. In general, assessment of
contribution of the independent variables can be made using one of the following:
• forward stepwise discriminant analysis;
• backward stepwise discriminant analysis;
• scaled vector test, separation of means test (Mosteller and Wallace, 1963; Joy and
Tollefson, 1975);
• conditional deletion test (Altman and Loris, 1976);
• univariate f-statistic; and
• Mosteller–Wallace analysis.
Obviously, this is also another tonic issue, as the various methods’ result differs. For Predicting
instance, the scaled vector method used by Altman (1968) has been over-criticized in the corporate
literature, particularly by Joy and Tollefson (1975) and Moyer (1977).
The SLR depicts that statistical techniques, AIES and theoretical models record
grand mean of overall predictive accuracy (OPA) estimation of 68.19, 67.76 and 63.14 per
cent, respectively. Furthermore, statistical techniques slightly maintains its supremacy
in the overall validation accuracy (VAL) of 55.13 per cent compared to 53.02 and 44.29 477
per cent for AIES and theoretical models, respectively. This, in turn, confirms the
findings of Aziz and Dar (2006), but sharply contrasts with Platt and Platt’s (1990)
assertion that the within-sample classification results one-year prior to failure are fairly
invariant with respect to methodology. We, however, have evidence to support Platt and
Platt in advancing the notion that validation results are disappointing, suggesting that
the values of certain independent variables differ markedly between the validation and
estimation periods (Wood and Piesse, 1987).
On the contrary, the assessment of Type I and II errors indicates that the results are
mixed. First, the AIES models record lower Type II error in the estimation (validation)
test of 7.03 per cent (6.03 per cent), followed by the statistical and theoretical models. In
contrast, the statistical models record a lower Type 1 error of 8.70 per cent (9.56 per cent)
for estimation (validation), compared to 9.31 per cent (9.53 per cent) and 8.71 per cent
(11.43 per cent) for the AIES and theoretical models, respectively. This inconsistency is
due in part to different techniques of validation. This, in part, suggests that choosing
between techniques is problematic (Aziz and Dar, 2006).
Nonetheless, the predictive accuracy estimation (validation) rates reported by
individual studies using AIES, statistical techniques and theoretical arguments are
between 100 per cent (99 per cent)-61 per cent (58 per cent), 100 per cent (100 per cent)-56
per cent (50 per cent) and 94 per cent (90 per cent)-83 per cent (68 per cent), respectively.
This confirms Scott’s (1980) findings that bankruptcy prediction is both empirically
feasible and theoretically explainable. Data sources for prior studies. The USA and UK offer multiplicity of data
sources for corporate failure prediction studies. The US data sources include, but are not
limited to, Moody Industrial Manual, Compustat Industrial Tape, OTC Manual, Wall
Street Journal Index and The Standard & Poor Stock Reports. The UK studies data
sources are mainly from the Jordan Dataquest Database, DataStream, Compustat and
Worldscope, Financial Times Actuaries All Share Index and Department of Trade
Databank. The Australian data sources include Sydney Stock Exchange and Australian
Graduate School of Management Annual Report File. The rests are Centrate dei Bilanci,
Korean Stock Exchange, Compustat, SABI Bureau Van Djik, French Database Diane
and National Bank of Belgian for Italy, Korea, Canada, Spain, France and Belgium,

4. Limitations of the SLR

The present SLR limitations are as follows. First, the searches in this SLR are defined by
using certain synonymous words with corporate failure to six search engines (see
Section 3.1.3). Thus, we cannot promise that all relevant papers to the scope of the SLR
have been found. For instance, ten important papers in this SLR are directly included in
the review as a result of their citation in prior reviews. However, after checking the
references to the papers included in the SLR, 40 and 95 models have been published in
IJLMA Grade-4 and Grade-3 journals of ABS Guide, respectively. This implies that we have
57,5 analysed the contents of an illustrative sample of the field.
This notwithstanding, the repetition of the SLR is not feasible due to the deviation
from protocol stated in 3.1. However, this deviation is inevitable in other to have a
complete sample for more comprehensive discussion. We consider reasonable sample
estimation and validation sample as well as models’ robustness, predictive accuracy,
478 adaptability and explanatory capability to be part of quality assessment criteria (see
Section 2.5), but these items are difficult to quantify with precision.

5. Conclusion and further work

This paper presents an SLR, during the period 1966-2012, in the application of
statistical, intelligent techniques and theoretical approaches to solve the bankruptcy
prediction problem faced by firms in 12 different countries. This review answers the
question: What are the methodological issues in the literature? The rationale is to
uncover the main research gap in the literature.
The search engines of Google Scholar, Wiley Interscience, Science Direct, Web of
Knowledge and Business Source Complete of the Social Sciences were systematically
searched for the reviewed papers using a defined search string. The review modifies the
approaches used by previous review researchers, namely, Aziz and Dar (2006), Ravi and
Ravi (2007) and Balcaen and Ooghe (2006). Thus, for each paper, we extract study
attributes such as author, year, period, accuracy, errors, sample, matching technique,
variables, origin, firm type and journal of publication (Aziz and Dar, 2006; Ravi and Ravi
(2007). Secondly, following Balcaen and Ooghe (2006), we propose a synthesis of the
proposal to discuss the methodological issues in the literature. Finally, the rationale
behind research and operational definition of failure is extracted and discussed using a
template from Dietrich (1984) and Morris (1997), respectively.
After a detailed review of the literature, the researchers are overwhelmed by the
valuable contributions made by studies reviewed and those cited to the existing body of
accounting, finance and economics literature, particularly in the area of corporate failure
prediction. We observe that most of the studies also achieved remarkable results using
one-year financial data prior to failure. Results of models, however, deteriorate two to
five years prior to failure. The approach followed in all the papers reviewed may be
termed ex-post empirical: a group of actual failures is identified from an arbitrary
period, and the characteristics of these firms one or more years prior to failure are
compared with a group of firms which did not fail.
In addition, our results indicate that statistical techniques dominate the literature at
60.58 per cent, compared to 34.31 per cent and 5.11 per cent for AIES and theoretical
approaches, respectively, suggesting that a theoretically sound, highly accurate, simple
and widely used corporate failure prediction model for stakeholders is yet to be
developed. Thus, despite the countless models, the question of when and why
businesses failed is still a grey area. In summary, the SLR reiterates Neophytou and
Molinero’s (2004) assertion that:
[ ] all the aforementioned studies regardless of the approach used, have one common
impediment: they are not based on an economic theory in choosing the variables for
distinguishing between failing and non-failing firms (Neophytou and Molinero, 2004, p. 8).
Again, Taffler (1983) posits that “The lack of any real theory relating to the use of Predicting
financial ratio analysis constitutes a serious gap in the accounting literature” (Taffler, corporate
1983. p. 3).
For this reason, future researchers must consider the link between the corporate
failure phenomena and theoretical arguments from proponents of existing economics
theories. In particular, sound theoretical arguments based on legalistic, managerial
hegemony, agency and resources dependency perspectives using the corporate 479
governance lens may offer insight and depend our understanding in the corporate
failure process. In short, we argue that these theories can answer the grey question of
when and why businesses fail.

1. To conserve space, we present only models published in Grade-4 Journals; the full list,
however, is available upon request. Abbreviations: First row: year of publication, AIM –
objective of study, VAR –independent variables, TERM – definition of corporate failure,
MATC – matching technique, YPFT – number of years prior to failure considered, F-SAMP –
full sample, ORIG – origin, OPA – overall predictive accuracy in per cent, ETI – estimation
Type I error in per cent, ETII – estimation Type II error in per cent, TEST – validation Type
I errors in per cent, VTII – validation Type II errors in per cent, V MET – validation method.
Fifth column: FR – financial ratios, MIX – financial and non-financial ratios, NFI – non
financial information. Tenth column: M – manufacturing, R – retail, n/a – not applicable, C –
construction, D – distribution, Ind – industrial, S&L – savings and loans.
2. To conserve space, the full list of ratios per each study is available upon request.

Aldrich, J.H and Nelson, F.D. (1984), Linear Probability, Logit and Probit Models, Sage, p. 45.
Altman, E.I. (1968), Financial ratios, discriminant analysis and the prediction of corporate
bankruptcy, Journal of Finance, Vol. 23 No. 4, pp. 589-609.
Altman, E.I. (1973), Predicting railroad bankruptcies in America, The Bell Journal of Economics
and Management Science, Vol. 4 No. 1, pp. 184-211.
Altman, E.I. (1984), The success of business failure prediction models: an International Survey.
Journal of Banking & Finance, Vol. 8 No. 2, pp. 171-198.
Altman, E.I. and Eisenbeis, R.A. (1978), “Financial applications of discriminant analysis: a
clarification”, Journal of Financial & Quantitative Analysis, Vol. 13 No. 1, pp. 185-195.
Altman, E.I., Haldeman, R.G. and Narayanan, P. (1977), “ZETAM analysis a new model to identify
bankruptcy risk of corporations”, Journal of Banking & Finance, Vol. 1 No. 1, pp. 29-54.
Altman, E.I. and Loris, B. (1976), “A financial early warning system for over-the-counter
broker-dealers”, Journal of Finance, Vol. 31 No. 4, pp. 1201-1217.
Altman, E.I. and Narayanan, P. (1996), “Business failure classification models: an international
survey”, Working Paper Series, New York University, New York, NY.
Altman, E.I., Marco, G. and Varetto, F. (1994), “Corporate distress diagnosis: comparisons using
linear discriminant analysis and neural networks (The Italian Experience)”, Journal of
Banking & Finance, Vol. 18 No. 3, pp. 505-529.
Aly, I.M., Barlow, H.A. and Jones, R.W. (1992), “The usefulness of SFAS No. 82 (Current Cost)
information in discriminating business failure: an empirical study”, Journal of Accounting,
Auditing & Finance, Vol. 7 No. 2, pp. 217-229.
IJLMA Argenti, J. (1976), Corporate Collapse: The Causes and the Symptoms, McGraw-Hill, Maidenhead,
Argenti, J. (1985), “Signs of corporate collapse”, IEEE Engineering Management Review, Vol. 13
No. 4, pp. 36-38.
Aziz, A., Emanuel, D.C., Lawson, G.H. (1988), “Bankruptcy prediction-an investigation of cash
flow based models”, Journal of Management Studies, Vol. 25 No. 5, pp. 419-437.
480 Aziz, M.A. and Dar, H.A. (2006), “Predicting corporate bankruptcy: where we stand?”, Corporate
Governance, Vol. 6 No. 1, pp. 18-33.
Back, B., Laitinen, T. and Sere, K. (1996), “Neural network and genetic algorithm for bankruptcy
prediction”, Expert Systems with Applications, Vol. 11 No. 4, pp. 407-413.
Balcaen, S. and Ooghe, H. (2006), “35 years of studies on business failure: an overview of the classic
statistical methodologies and their related problems”, The British Accounting Review,
Vol. 38 No. 1, pp. 63-93.
Beaver, W.H. (1966), “Financial ratios as predictors of failure”, Journal of Accounting Research,
Vol. 4 No. 3, pp. 71-111.
Beaver, W.H. (1968a), “Market prices, financial ratios and the prediction of failure”, Journal of
Accounting Research, Vol. 6 No. 2, pp. 179-192 (Autumn).
Beaver, W.H. (1968b), “The information content of annual earnings announcements”, Journal of
Accounting Research, Vol. 6 No. 3, pp. 67-92.
Begley, J., Ming, J. and Watts, S. (1996), “Bankruptcy classification errors in the 1980s: an
empirical analysis of Altman’s and Ohlson’s models”, Review of Accounting Studies, Vol. 1
No. 4, pp. 267-284.
Bell, T., Ribar, G. and Verchio, J. (1990), “Neural networks vs. Logistic regression in predicting
bank failures”, in Srivastava, P. (Ed.), Auditing Symposium X, University of Kansas,
Betts, J. and Belhoul, D. (1987), “The effectiveness of incorporating stability measures in company
failure models”, Journal of Business Finance & Accounting, Vol. 14 No. 3, pp. 323-334.
Biolchini, J., Main, G.P., Natali, A.C.C. and Travassos, G.H. (2005), “Systematic review in software
engineering”, TR-ES 670/05, System Engineering and Computer Science Rio De Janeiro,
Department COPPE/UFRIJ.
Blum, M.P. (1974), “Failing company discriminant analysis”, Journal of Accounting Research,
Vol. 12 No. 1.
Booth, P.J. (1983), “Decomposition measures and the prediction of financial failure”, Journal of
Business Finance & Accounting, Vol. 10 No. 1, pp. 67-82.
Boritz, J.E., Kennedy, D.B. and De Miranda, E.A. (1995), “Predicting corporate failure using a
neural network approach”, International Journal of Intelligent Systems in Accounting
Finance & Management, Vol. 4 No. 2, pp. 95-111.
Braun, H. and Chandler, J.S. (1987), “Predicting stock market behaviour through rule induction: an
application of the learning-from-examples approach”, Decision Sciences, Vol. 18 No. 3,
pp. 415-429.
Casey, C. and Bartczak, N. (1985), “Using operating cash flow data to predict financial distress:
some extensions”, Journal of Accounting Research, Vol. 23 No. 1, pp. 384-401.
Casey, C.J. and Bartczak, N.J. (1984), “Cash flow–it’s not the bottom line”, Harvard Business
Review, Vol. 62 No. 4, pp. 61-66.
Chakravarthy, B.S. (1986), “Measuring strategic performance”, Strategic Management Journal,
Vol. 7 No. 5, pp. 437-458.
Charitou, A., Neophytou, E. and Charalambous, C. (2004), “predicting corporate failure: empirical Predicting
evidence for the UK”, European Accounting Review, Vol. 13 No. 3, pp. 465-497.
Collins, R.A. (1980), “An empirical prediction on bankruptcy prediction models”, Financial
Management (1972), Vol. 9 No. 2, pp. 52-57.
Collopy, F., Adya, M. and Armstrong, J.S. (1994), “Principles for examining predictive validity: the
case of information systems spending forecasts”, Information Systems Research, Vol. 5
No. 2, pp. 170-179. 481
Cox, D.R. and Oakes, D. (1984), Analysis of Survival Analysis Data, Chapman & Hall, London.
Cybinski, P. (2000), “The path to failure: where are bankruptcy studies at now?”, Journal of
Business and Management, Vol. 7 No. 1, pp. 11-39.
Dambolena, I.G. and Khoury, S.J. (1980), “Ratio stability and corporate failure”, Journal of Finance,
Vol. 35 No. 4, pp. 1017-1026.
Deakin, E.B. (1972), A Discriminant Analysis of Predictors of Business Failure, Wiley-Blackwell.
Dietrich, J.R. (1984), “Discussion of methodological issues related to the estimation of financial
distress prediction models”, Journal of Accounting Research, pp. 83-86.
Dimitras, A.I., Zanakis, S.H. and Zopounidis, C. (1996), “A survey of business failures with an
emphasis on prediction methods and industrial applications”, European Journal of
Operational Research, Vol. 90 No. 3, pp. 487-513.
Edmister, R.O. (1972), Financial Ratios as Discriminant Predictors of Small Business Failure,
Eisenbeis, R.A. and Avery, R.B. (1972), Discriminant Analysis and Classification Procedures:
Theory and Applications, DC Heath & Co, Lexington, MA.
El Hennawy, R.H.A. and Morris, R.C. (1983), “The significance-of base year in developing failure
prediction models”, Journal of Business Finance & Accounting, Vol. 10 No. 2, pp. 209-223.
Etheridge, H.L., Sriram, R.S. and Hsu, H.Y.K. (2000), “A comparison of selected artificial neural
networks that help auditors evaluate client financial viability”, Decision Sciences, Vol. 31
No. 2, pp. 531-550.
Fisher, R.A. (1936), “The use of multiple measurements in taxonomic problems”, Annual
Gugenics, Vol. 7 No. 2, pp. 78-188.
Frydman, H., Altman, E.I. and Kao, D. (1985), “Introducing recursive partitioning for financial
classification: the case of financial distress”, The Journal of Finance, Vol. 40 No. 1,
pp. 269-291.
Gentry, J.A., Newbold, P. and Whitford, D.T. (1985), “Classifying bankrupt firms with funds flow
components”, Journal of Accounting Research, Vol. 23 No. 1, pp. 146-160.
Gentry, J.A., Newbold, P. and Whitford, D.T. (1987), “Funds flow components, financial ratios, and
bankruptcy”, Journal of Business Finance & Accounting, Vol. 14 No. 4, pp. 595-606.
Greenhalgh, L., Lawrence, A.T. and Sutton, R.I. (1988), “Determinants of workforce reduction
strategies in declining organizations”, Academy of Management Review, Vol. 13 No. 2,
pp. 241-254.
Hensher, D.A. and Jones, S. (2007), “Forecasting corporate bankruptcy: optimizing the
performance of the mixed logit model”, Abacus, Vol. 43 No. 3, pp. 241-264.
Hillegeist, S., Keating, E., Cram, D. and Lundstedt, K. (2004), “Assessing the probability of
bankruptcy”, Review of Accounting Studies, Vol. 9 No. 1, pp. 5-34.
Hyghebaert, N., Gaeremynck, A., Roodhooft, F. and Van De Gucht, L.M. (2000), “New firm
survival: the effects of start-up characteristics”, Journal of Business Finance & Accounting,
Vol. 27 Nos 5/6, pp. 627-651.
IJLMA Jardin, P.D. and Séverin, E. (2011), “Predicting corporate bankruptcy using a self-organizing map:
an empirical study to improve the forecasting horizon of a financial failure model”, Decision
57,5 Support Systems, Vol. 51 No. 3, pp. 701-711.
Jones, F.L. (1987), “Current techniques in bankruptcy prediction”, Journal of Accounting
Literature, Vol. 6, pp. 131-164.
Jones, S. and Hensher, D.A. (2004), “Predicting firm financial distress: a mixed logit model”, The
482 Accounting Review, Vol. 79 No. 4, pp. 1011-1038.
Joy, O.M. and Tollefson, J.O. (1978), “Some clarifying comments on discriminant analysis”, Journal
of Financial & Quantitative Analysis, Vol. 13 No. 1, pp. 197-200.
Kahya, E. and Theodossiou, P. (1999), “Predicting corporate financial distress: a time-series cusum
methodology”, Review of Quantitative Finance and Accounting, Vol. 13 No. 4, pp. 323-345.
Keasey, K., McGuinness, P. and Short, H. (1990), “Multilogit approach to predicting corporate
failure-further analysis and the issue of signal consistency”, Omega, International Journal
of Management Science, Vol. 18 No. 1, pp. 85-94.
Keasey, K. and Watson, R. (1987), “Current cost accounting and the prediction of small company
performance”, Journal of Business Finance and Accounting, pp. 51-70.
Keasey, K. and Watson, R. (1991a), “Financial distress prediction models: a review of their
usefulness1”, British Journal of Management, Vol. 2 No. 2, pp. 89-102.
Keasey, K. and Watson, R. (1991b), “Financial distress prediction models: a review of their
usefulness”, British Journal of Management, pp. 89-102.
Khanna, T. (1990), Foundations of Neural Networks Reading, Addison-Wesley, Massachusetts, MA.
Kim, M. and Kang, D. (2010), “Ensemble with neural networks for bankruptcy prediction”, Expert
Systems with Applications, Vol. 37 No. 4, pp. 3373-3379.
Kingdom, J. and Feldman, K. (1995), “Genetic algorithms for bankruptcy prediction”, Search Space
Research Report No 01-95, Search Space, London.
Koh, H.C. (1992), “The sensitivity of optimal cut-off points to misclassification costs of Type I and
Type II errors in the going-concern prediction context”, Journal of Business Finance &
Accounting, Vol. 19 No. 2, pp. 187-197.
Kolodner, J. (1993), Case-Based Reasoning, Morgan Kaufmann, San Mateo, CA.
Laitinen, E.K. (1992), “Prediction of failure of a newly founded firm”, Journal of Business
Venturing, Vol. 7 No. 4, pp. 323-340.
Laitinen, E.K. and Laitinen, T. (1998), “Cash management behavior and failure prediction”,
Journal of Business Finance & Accounting, Vol. 25 Nos 7/8, pp. 893-919.
Lane, W.R., Looney, S.W. and Wansley, W. (1986), “An application of the cox proportional hazards
model to bank failure”, Journal of Banking and Finance, Vol. 10 No. 4, pp. 511-531.
Lau, A.H. (1987), “A five-state financial distress prediction model”, Journal of Accounting
Research, Vol. 25 No. 1, pp. 127-138.
Lee, K.C., Han, I. and Kwon, Y. (1996), “Hybrid neural network models for bankruptcy
predictions”, Decision Support Systems, Vol. 18 No. 1, pp. 63-72.
Lennox, C. (1999), “Identifying failing companies: a re-evaluation of the logit, probit and da
approaches”, Journal of Economics and Business, Vol. 51 No. 4, pp. 347-364.
Lev, B. (1973), “Decomposition measures for financial analysis”, Financial Management, Spring,
Vol. 2 No. 1, pp. 56-63.
Lo, A.W. (1986), “Logit versus discriminant analysis: a specification test and application to
corporate bankruptcies”, Journal of Econometrics, Vol. 31 No. 2, pp. 151-178.
Luoma, M. and Laitinen, E.K. (1991), “Survival analysis as a tool for company failure prediction”,
Omega, Vol. 19 No. 6, pp. 673-678.
Lussier, R.N. (1995), “A nonfinancial business success versus failure prediction mo”, Journal of Predicting
Small Business Management, Vol. 33 No. 1, p. 8.
Martin, D. (1977), “Early warning of bank failure a logit-regression approach”, Journal of Banking
and Finance, Vol. 1 No. 3, pp. 249-276. failure
Mensah, Y.M. (1983), “The differential bankruptcy predictive ability of specific price level
adjustments: some empirical evidence”, Journal of Accounting Review, Vol. 58 No. 2,
pp. 228-246.
Mensah, Y.M. (1984), “An examination of the stationarity of multivariate bankruptcy prediction
models: a methodological study”, Journal of Accounting Research, Vol. 22 No. 1, pp. 380-395.
Menski, C.F. and Lerman, S.R. (1977), “The estimation of choice probabilities from choice based
samples”, Econometrica, Vol. 45 No. 8, pp. 1977-1988.
Meyer, P.A. and Pifer, H.W. (1970), “Prediction of bank failures”, Journal of Finance, Vol. 25 No. 4,
pp. 853-868.
Min, J.H. and Lee, Y.C. (2005), “Bankruptcy prediction using support vector machine with optimal
choice of kernel function parameters”, Expert Systems with Application, Vol. 28 No. 4,
pp. 603-614.
Morris, R. (1997), Early Warning Indicators of Corporate Failure: A Critical Review of Previous
Research and Further Empirical Evidence, Ashgate, Aldershot.
Mosteller, F. and Wallace, D.L. (1963), “Inference in an authorship problem”, Journal of American
Statistical Association, Vol. 58 No. 302, pp. 275-309.
Moyer, R.C. (1977), “Forecasting financial failure: a re-examination”, Financial Management,
Vol. 6 No. 1, pp. 11-17.
Neophytou, E. and Molinero, C.M. (2004), “Predicting corporate failure in the UK: a
multidimensional scaling approach”, Journal of Business Finance & Accounting, Vol. 31
Nos 5/6, pp. 677-710.
Nicolas, J. and Toval, A. (2009), “On the generation of requirements specifications from software
engineering models: a systematic literature review”, Information and Software Technology,
Vol. 51 No. 9, pp. 1291-1307.
Ohlson, J.A. (1980), “Financial ratios and the probabilistic prediction of bankruptcy”, Journal of
Accounting Research, Vol. 18 No. 1, pp. 109-131.
O’Leary, D.E. (1998), “Using neural network to predict corporate failure”, International Journal of
Intelligent Systems in Accounting Finance and Management, Vol. 7 No. 3, pp. 187-197.
Ooghe, H. and Balcaen, S. (2002), “Are failure prediction models transferable from one country to
another? An empirical study using Belgian financial statements”, Proceedings of the 9th
Annual Conference of the Multinational Finance Society, Cyprus, 30 June-3 July.
Ooghe, H., Joos, P. and Bourdeaudhuij, C. (1995), “Financial distress models in Belgium: the results
of a decade of empirical research”, International Journal of Accounting, Vol. 30 No. 3,
pp. 245-274.
Park, C. and Han, I. (2002), “A case-based reasoning with the feature weights derived by analytic
hierarchy process for bankruptcy prediction”, Expert Systems with Applications, Vol. 23
No. 3, pp. 255-264.
Peel, M., Peel, D. and Pope, P. (1986), “Predicting corporate failure – some results for the UK
corporate sector”, Omega, Vol. 14 No. 1, pp. 5-12.
Peel, M.J. and Peel, D.A. (1988), “A multilogit approach to predicting corporate failure-some
evidence for the UK corporate sector”, Omega, Vol. 16 No. 4, pp. 309-318.
Pinches, G., Mingo, K. and Caruthers, J. (1973), “The stability of financial patterns in industrial
organizations”, Journal of Finance, Vol. 28 No. 2, pp. 389-396.
IJLMA Platt, H.D. and Platt, M.B. (1990), “Development of a class of stable predictive variables: the case of
bankruptcy prediction”, Journal of Business Finance & Accounting, Vol. 17 No. 1, pp. 31-51.
Platt, H.D., Platt, M.B. and Pedersen, J. (1994), “Bankruptcy discrimination with real variables”,
Journal of Business Finance & Accounting, Vol. 21, pp. 491-509.
Pompe, P.P.M. and Bilderbeek, J. (2005), “The prediction of bankruptcy of small- and
medium-sized industrial firms”, Journal of Business Venturing, Vol. 20 No. 6, pp. 847-868.
484 Premachandra, I.M., Bhabra, G.S. and Sueyoshi, T. (2009), “DEA as a tool for bankruptcy
assessment: a comparative study with logistic regression technique”, European Journal of
Operational Research, Vol. 193 No. 2, pp. 412-424.
Quinlan, J.R. (1979), “Discovering rules by induction from large collections of examples”, in
Michie, D. (Ed.), Expert Systems in the Microelectronic Age, Edinburgh University Press,
Edinburgh, pp. 168-201.
Ravi, P. and Ravi, K.V. (2007), “Bankruptcy prediction in banks and firms via statistical and
intelligent techniques-a review”, European Journal of Operational Research, Vol. 180 No. 1,
pp. 1-28.
Rosner, R.L. (2003), “Earnings manipulation in failing firms”, Contemporary Accounting
Research, Vol. 20 No. 2, pp. 361-408.
Salchenberger, L.M., Mine Cinar, E. and Lash, N.A. (1992), “Neural networks: a new tool for
predicting thrift failures”, Decision Sciences, Vol. 23 No. 4, pp. 899-916.
Sánchez-Lasheras, F., De Andrés, J., Lorca, P., De Cos Juez, F.J. (2012), “A hybrid device for the
solution of sampling bias problems in the forecasting of firms’ bankruptcy”, Expert Systems
with Applications, Vol. 39 No. 8, pp. 7512-7523.
Santomero, A.M and Vinso, J.D. (1997), “Estimating the probability of failure for commercial
banks and the banking system”, Journal of Banking and Finance, Vol. 1, pp. 185-205.
Scott, J. (1980), “The probability of bankruptcy”, Journal of Banking and Finance, Vol. 5,
pp. 317-344.
Sharma, S. and Mahajan, V. (1980), “Early warning indicators of business failure”, Journal of
Marketing, Vol. 44 No. 4, pp. 80-89.
Shin, K.S. and Lee, Y.J. (2002), “A genetic algorithm application in bankruptcy prediction
modelling”, Expert Systems with Applications, Vol. 23 No. 3, pp. 321-328.
Shumway, T. (2001), “Forecasting bankruptcy more accurately: a simple hazard model”, The
Journal of Business, Vol. 74 No. 1, pp. 101-124.
Slowinski, R. and Zopounidis, C. (1995), “Application of the rough set approach to evaluation of
bankruptcy risk”, International Journal of Intelligent Systems in Accounting Finance &
Management, Vol. 4 No. 1, pp. 27-41.
Taffler, R.J. (1983), “The assessment of company solvency and performance using a statistical
model”, Accounting & Business Research, Vol. 13 No. 52, pp. 295-307.
Taffler, R.J. (1984), “The use of the Z-Score in practice”, City Business School Centre for Empirical
Research in Finance and Accounting Working Paper.
Tam, K. (1991), “Neural network models and the prediction of bank bankruptcy”, Omega, Vol. 19
No. 5, pp. 429-445.
Tam, K.Y. and Kiang, M.Y. (1992), “Managerial applications of neural networks: the case of bank
failure predictions”, Management Science, Vol. 38 No. 7, pp. 926-947.
Tay, F.E.H. and Shen, L. (2002), “Economic and financial prediction using rough set model”,
European Journal of Operational Research, Vol. 141 No. 3, pp. 641-659.
Theodossiou, P. (1991), “Alternative models for assessing the financial condition of business in
Greece”, Journal of Business Finance and Accounting, Vol. 18 No. 5, pp. 697-720.
Watson, J. and Everett, J.E. (1996), “Small business failure rates: choice of definition and the size Predicting
effect”, The Journal of Entrepreneurial and Small Business Finance, Vol. 5 No. 3, pp. 271-285.
Wilcox, J.W. (1973), “A prediction of business failure using accounting data”, Journal of
Accounting Research, Vol. 11 No. 3, pp. 163-179. failure
Wilson, R.L. and Sharda, R. (1994), “Bankruptcy prediction using neural networks”, Decision
Support Systems, Vol. 11 No. 5, pp. 545-557.
Wood, D. and Piesse, J. (1987), “The information value of MDA based financial indicators”, Journal 485
of Business Finance and Accounting, Vol. 14 No. 1, pp. 27-38.
Yang, Z.R., Platt, M.B. and Platt, H.D. (1999), “Probabilistic neural networks in bankruptcy
prediction”, Journal of Business Research, Vol. 44 No. 2, pp. 67-74.
Zavgren, C.V. (1983), “The prediction of corporate failure: the state of the art”, Journal of
Accounting Literature, Vol. 2, pp. 1-38.
Zavgren, C.V. (1985), “Assessing the vulnerability to failure of American industrial firms: a
logistic analysis”, Journal of Business Finance & Accounting, Vol. 12 No. 1, pp. 19-45.
Zhang, G., Hu, Y., Eddy, P.B. and Indro, C.D. (1999), “Artificial neural networks in bankruptcy
prediction: general framework and cross-validation analysis”, European Journal of
Operational Research, Vol. 116 No. 1, pp. 16-32.
Zmijewski, M.E. (1984), “Methodological issues related to the estimation of financial distress
prediction models”, Journal of Accounting Research, Vol. 22, pp. 59-82.

Further reading
Ball, R. and Foster, G. (1982), “Corporate financial reporting: a methodological review of empirical
research”, Journal of Accounting Research, Vol. 20, pp. 161-234.
Dimitras, A.I., Slowinski, R., Susmaga, R. and Zopounidis, C. (1999), “Business failure prediction
using rough sets”, European Journal of Operational Research, Vol. 114 No. 2, pp. 263-280.
Innes, J., Mitchell, F. and Sinclair, D. (2000), “Activity-based costing in the UK’s largest companies:
a comparison of 1994 and 1999 survey results”, Management Accounting Research, Vol. 11
No. 3, pp. 349-362.
Jones, S. and Hensher, D.A. (2007), “Modelling corporate failure: a multinomial nested logit
analysis for unordered outcomes”, The British Accounting Review, Vol. 39 No. 1, pp. 89-107.
Kasey, C.J. and Bartczak, N.J. (1984), “Cash flow–it’s not the bottom line (Cover Story)”, Harvard
Business Review, Vol. 62 No. 4, pp. 61-66.
Lee, Y. (2005), “Bankruptcy prediction using support vector machine with optimal choice of kernel
function parameters”, Expert Systems with Applications, Vol. 28 No. 4, pp. 603-614.
Taffler, R. and Tisshaw, H. (1977), “Going, going, gone-four factors which predict”, Accountancy,
Vol. 88 No. 1003, pp. 50-54.

Corresponding author
Joseph Arthur can be contacted at: jarthur1989@gmail.com

For instructions on how to order reprints of this article, please visit our website:
Or contact us for further details: permissions@emeraldinsight.com
Reproduced with permission of copyright owner. Further
reproduction prohibited without permission.