Keywords

Corporate Change, Corporate Governance, Restructuring, Reorganization, Merger, Acquisition, Leveraged Buyout, Market Share, Nonfinancial, Qualitative, Operational, Auditor

Introduction

Accountants, academics, economists, investors, regulators and other stakeholders often take distinct approaches for studying potential success or failure of firms in terms of financial performance. While some groups advance knowledge through studying influences external to the firm, others look to internal processes of the firm, and then others look at both internal and external forces, with emphasis on the comparative impact of external versus internal forces. Few researchers attempt to integrate these two categories of independent variables into a causality model. Theories from the internal compared to the external perspective of the challenges to successful performance tend to view the firm as either financially and managerially autonomous entities, or remotely controlled automatons. Our study focuses on internal factors that may indicate the firm’s capability to successfully restructure. Based on theory, we propose models for forecasting the likelihood of a firm to indeed, accomplish its official restructuring goals. These theories originate from the view that a firm’s processes are significant in a firm’s resilience, or autonomy. Regulatory bodies might emphasize the need for additional regulation, while accountants and academics might advocate additional ethics training, technology training or other forms of education when a firm must reorganize (Rosner, 2003).

We derive our evidence from a natural experiment of publicly known efforts toward reorganization. Thus, through observing firms that utilized Chapter 11 formalities during 2008 through 2020, we test the efficacy of our models. Our exploration employs key qualitive indicators involved by our sample of firms for the likelihood of successful restructuring. While some firms strategically file bankruptcy (CH11) for restructuring purposes (Chapter 11 reorganization), other firms simply file to completely dissolve the entity (Chapter 7 liquidation).

Shareholders may perceive the effects of management decisions on their wealth differently, and therefore react differentially to various corporate events, particularly types of restructuring decisions, changes in corporate governance practices, specific regulatory changes, risk shifting, claims ownership characteristics, and different categories or chapters of bankruptcy filing (Rose-Green & Dawkins, 2002; Bris et al., 2006; Barker et al., 2001; Moulton & Thomas, 1993; Brown & Mooradian, 1993).

Rose-Green & Dawkins (2002) find that stakeholders differentiate between firm strategy and financial distress based upon investor reactions between the use of a restructuring or reorganization bankruptcy (categorized as a Chapter 11 bankruptcy), contrasted with investor reactions around financial distress or liquidation bankruptcies (categorized as a Chapter 7 bankruptcy). Can certain actions by managers and corporate leaders of restructuring firms give stakeholders insights into strategic choices for the firm? How likely are those actions to coincide with outcomes consistent with the objectives of the firm's stakeholders (i.e., the shareholders, the employees, the creditors, and others)?

Our findings show significant relationships between qualitative factors and successful restructuring post-Chapter 11 (“reorganization-bankruptcy”) filing. We find that the Altman model is not significantly related to reorganization-bankruptcy filing. Moreover, we find that a number of qualitative factors acting alone or in concert with other qualitative factors have a significantly more important association with reorganization-bankruptcy filings than financial ratios. Further, we find that qualitative characteristics are significantly different when comparing reorganization-bankruptcy firms with Chapter 7 (distressed) firms and non-bankrupt firms. Evidence is provided that there are other incentives for filing reorganization-bankruptcy. For example, our findings suggest that a reorganization-bankruptcy is utilized as an efficient proxy or intermediate step for leveraged buy-outs, mergers, equity swapping, and going-private transactions, often to the detriment of bondholders and other debtholders.

By extension, we find that specific governance characteristics are significantly different in those firms that emerge from bankruptcy reorganized, in comparison to those that liquidate or fail to emerge. There is strong evidence that firms that initiate reorganization through changes in corporate governance and other risk shifting techniques are more likely to emerge from bankruptcy reorganization successfully reorganized. Finally, we find that there is significant CEO turnover surrounding reorganization-bankruptcy and that the new replacement CEO’s compensation package is significantly different in form and substance than that of the pre- reorganization-bankruptcy CEO.

Overall, a better specified model, based on theory, which includes qualitative considerations and which considers bankruptcy categories, balances the performance of bankruptcy models that include only financial variables. In conclusion, our findings suggest that reorganization-bankruptcy provides both an incentive and a mechanism to improve the quality of corporate governance (Ittner & Larcker, 1998b).

An implication of this research is that managers in distressed firms take different actions; some are value-maximizing while others are value-destroying. The purpose of this study is to examine whether certain operational and non-financial measures, which may be related to strategic choices are associated with the likelihood of corporate restructuring.

Background and Development of Hypotheses

Bankruptcy Prediction

After the seminal works of Beaver (1966) and Altman (1968) demonstrating the predictive value of financial ratios for firm failure, a number of researchers spent their time evaluating the effectiveness of these financial statement based models in determining bankruptcy across industries e.g., Cochran et al. (2006) with internet firms; Knox et al. (2009) hospitals, across time e.g., Grice & Ingram (2001) using a sample from 1985-1991; Beaver et al. 2005 using a sample from 1962 to 2002), and with regards to the length of the prediction interval e.g., Philosophov & Philosophov (2002) using prediction windows from one year to five years before a bankruptcy event). Ohlson (1980) on factors that affect the robustness of prediction regressions and Zmijewski's (1984) subsequent research on the methodological issues in early bankruptcy studies led to cautionary discussions of the pitfalls and limitations of bankruptcy prediction modeling (e.g., Grice & Dugan, 2001), as well as a stream of research regarding the technical aspects of modeling the probability of bankruptcy - most notably, Shumway's (2001) introduction of hazard models. This research encouraged further evaluations of the traditional logistic regression techniques (i.e., logit models) leading researchers to consider other traditional methods of identifying bankruptcy potential (e.g., a comparison of hazard modeling to auditor's propensity to give going concern opinions by Sun, 2007) or to explore applications of alternative techniques to the problem (e.g., binary quantile regression modeling as in Li & Miu, 2010; spline modeling as in Giordani et al., 2014; data envelopment analysis modeling as in Mousavi et al., 2015; fuzzy set qualitative comparative analysis as in Boraty?ska, 2016; neural network analysis as in Jones et al., 2017). Thus, while the predominant methods in accounting are logit and hazard models, we find innovation and exploration in research on prediction methods.

In contrast, the literature on expanding the variables beyond those which are available from the publicly filed financial statements is relatively limited. The earliest research on the nature of variables selected for inclusion in bankruptcy prediction models primarily focused on financial statement items or three areas of financial ability: profitability, cash flow generation, and leverage. The bankruptcy prediction literature since Shumway (2001) has included more focus on market-based variables (e.g., Hillegeist et al., 2004; Agarwal & Taffler, 2008; Charitou et al., 2007; Bauer & Agarwal, 2014; Tian et al., 2015) or other financial information not found in financial statements (such as bond downgrades as in Kim & Nabar, 2007; equity prices and residual prices in conjunction with macroeconomic indicators as in Tinoco & Wilson, 2013). One viewpoint is that accounting-based variables and market-based variables need not be considered mutually exclusive in prediction modeling if the combined sets add to predictive power by capturing a mix of information about the firm (Beaver et al., 2005).

We find that very little published evidence that researchers considered nonfinancial measures in bankruptcy prediction before the most recent decade. Keasey & Watson (1987) examined several non-financial factors in the prediction of small company failure in the U.K. context. They found that a model including non-financial factors related to 'normal business hazards' (e.g., strikes, the loss of a large customer, and ‘managerial defects’) performed better than a model including only financial variables. Smith & Taffler (2000) analyzed the association between the chairman's statement and subsequent firm. Using form-oriented and meaning-oriented variables, they found that discretionary narrative disclosures have information that is useful for decision making as well as for determining the likelihood of firms’ survival. Senteney et al. (2006) examined whether auditor opinions or auditor changes were associated with bankruptcy and found that failing firms lean towards changing auditors more than healthy firms and that auditor qualified opinions can indicate impending bankruptcy.

Xu & Zhang (2009) found that the added variables capturing the main bank system and Keiretsu structures unique to firms operating in Japan added value to bankruptcy prediction models beyond accounting and market based variables for their sample of 3,510 Japanese listed firms from 1992 to 2005. Wu et al. (2010) evaluated several bankruptcy prediction models and included firm characteristics with the customary accounting and market based variables. They found that diversification, measured by the number of business segments, and firm size are both related to the probability of bankruptcy in their sample that covers the years from 1980 to 2006. Eisdorfer & Hsu (2011) found that measures of innovation were important for bankruptcy prediction among technology firms in a sample from 1976 to 2005. Beams et al. (2013) found that CEO and CFO resignations are associated with future bankruptcy. Jones (2017) used a gradient boosting model that overcomes the limitations on the number of variables included in bankruptcy prediction. In addition to a wide set of variables used in earlier studies (e.g., macroeconomic, bond downgrades, size, and industry), his set of 91 variables included ownership concentration/structure data, compensation data, and analysts' estimates to analyze corporate bankruptcies from 1987 to 2013. Although Jones (2017) indicates that his model overcomes the 'number of variables' problem, he readily acknowledges that the gradient boosting model presents challenges for interpretation to parameter estimates. His results, however, are consistent with that above indicating that predictive power is enhanced using a variable set that includes nonfinancial measures along with the typical accounting-based and market-based measures. Furthermore, he found that the ownership concentration/structure and compensation variables outperformed the market-based variables in their ability to predict bankruptcy in his sample (Ittner & Larcker, 1998a).

Bankruptcy Prediction and Operational Measures

Bankruptcy may represent a strategic choice by management or it may be an unplanned event. No matter how bankruptcy results, it is not an event to be taken lightly given the potential for high costs to shareholders, employees, creditors, and other stakeholders¹. Therefore, managers facing financial distress or who believe a bankruptcy is an optimal strategic choice will make certain choices that are likely to affect operational outcomes.

Given the findings of prior literature on nonfinancial measures, we propose the following hypotheses related to operational actions and the likelihood of bankruptcy:

H_a: A bankruptcy prediction model that includes operational measures in addition to financial measures will be better able to predict bankruptcy in the short-term and long-term than a model purely based on financial measures.

H_b: After controlling for financial measures, operational measures are significantly related to future bankruptcy.

Sample Selection And Methodology

Sample Selection

The main source of bankruptcy data is LoPucki’s Bankruptcy Research Database (BRD) of big-case bankruptcies that lists bankruptcies in the years 1978 to 2008. Bankruptcy is defined as an original filing of Chapter 7 or Chapter 11. The primary information extracted from BRD includes firm names, industry code, case number, and date of filing. For inclusion in the BRD database, the firm has to have assets worth $100 million or more (measured in 1980 dollars) at the time of filing and is required to file 10-Ks with the SEC. The search resulted in 501 firms. This source of bankruptcy firms was supplemented by identifying additional bankruptcies from the Compustat database. However, inclusion in the sample was restricted to the same conditions for inclusion into the BRD database. This search resulted in 235 additional bankruptcy firms. The two bankruptcy datasets were combined for a total of 736 bankruptcy firms to form the initial bankruptcy sample. Bankruptcy firm-year observations were then developed from the available pre-bankruptcy data for this sample of bankruptcy firms. For this study, firms that file at least one bankruptcy are retroactively and prospectively classified as bankrupt. Bankruptcy firm-year observations that are not one of the three pre-bankruptcy years are then deleted from the sample (Said et al., 2003).

Financial data from Compustat was collected for the identified bankrupt as well as all non-bankrupt firms. In accordance with the BRD database criteria, non-bankruptcy firms were eliminated from the sample if assets were not worth $100 million measured in 1980 dollars at the time of observation. Since this study is concerned with longer-term bankruptcy detection, we delete bankruptcy firms that did not have all three years of financial data prior to bankruptcy. This resulted in a final sample of 314 bankruptcy firms (942 firm-year observations) and 48,506 non-bankruptcy firm-year observations. Table 1 provides the industrial and yearly distributions of this final sample.

Table 1 Panel A Industry Distribution For Bankruptcy & Non-Bankruptcy Firms
	Bankruptcy firms		Non-bankruptcy firms
	N	%	N	%
Natural Resources	25	7.96	2968	6.12
Construction and Metal	37	11.78	3845	7.93
Food	6	1.91	1958	4.04
Consumer Goods	25	7.96	1818	3.75
Paper and printing	14	4.46	2564	5.29
Chemical and petroleum	11	3.5	4427	9.13
Machinery and Equipment	27	8.6	8445	17.41
Transportation-related	28	8.92	3452	7.12
Telecommunications	38	12.1	6773	13.96
Wholesale and retail	57	18.15	6180	12.74
Entertainment	10	3.18	702	1.45
Business Services	15	4.78	3212	6.62
Health and other services	20	6.37	1972	4.07
Non-classifiable	1	0.32	190	0.39
Total	314	100	48506	100

Panel A presents the industrial distribution of both bankrupt and non-bankrupt firms. Most industries constitute less than 10% of the bankruptcy sample other than construction and metal (12%), telecommunications (12%), and wholesale and retail (18%). As for the non-bankruptcy sample, most industries constitute less than 10% to the sample, other than machinery and equipment (17%), telecommunications (14%), and wholesale and retail (13%). There does not appear to be any significant difference between the samples with respect to industry distribution. We present the yearly distribution of the bankruptcy and non-bankruptcy sample in panel B. There are no bankruptcy cases in years 1978 to 1980 since data for three years prior to bankruptcy was necessary. Bankruptcy in each of the remaining years are below 5% other than in years 1982, 1999, 2000, 2001, 2002, and 2003 (these constitute 5%, 6%, 12%, 11%, 10% and 7% of the full sample, respectively). Observations in the non-bankruptcy sample are approximately evenly distributed throughout the sample years.

Methodology

In this study, we replicate and extend the Altman (1984) model since this is the most commonly used model in practice (Altman, 2000). However, we update the methodology by employing a logit model rather than the discriminant analysis methodology used in (Altman, 1968). This method is preferable as it does not impose restrictions on prior bankruptcy percentages (Ohlson, 1980, 112). Furthermore, Gilbert et al. (1990) argued that the high predictive ability of models based on financial ratios could be due to the way the control group is selected. They showed that a financial ratio-based model was unable to distinguish bankruptcies from other distressed firms. Furthermore, they showed that the variables that are able to discriminate between bankrupt and distressed firms are different from those that discriminate between bankrupt and randomly selected non-bankrupt firms. Hence, we do not restrict our comparison to a randomly selected group of non-bankrupt firms but rather compare the bankruptcies to all non-bankrupt firms with available data. Alternatively, in robustness tests, we use a hazard model. The hazard model method uses all available information to determine each firm’s bankruptcy risk at each point in time and generally provides consistent and unbiased coefficients. The Altman (1968) model specification, termed model 1, is as follows:²

Prob(Bankruptcy) = β₁WC(alt) + β₂RE(alt) + β₃MVE(alt) + β₄EBIT(alt) + β₅S(alt) + ε     (1)

Where
WC(alt) = working capital scaled by total assets (Compustat data 179 / data 6);
RE(alt) = retained earnings scaled by total assets (data 36 / data 6);
MVE(alt) = market value of equity scaled by book value of debt ([data 25*data 199] / data 181);
EBIT(alt) = operating income scaled by total assets (data 178 / data 6); and
S(alt) = sales scaled by total assets (data 12 / data6).

To test the hypotheses of the study, we employ a combined model to determine whether a group of predictors will improve the model. The operational effects model tested is developed by combining non-financial and operational measures with the Altman (1968) model, in model 2, as follows:

Prob(Bankruptcy) = β₁ WC(alt) + β₂ RE(alt) + β₃ MVE(alt) + β₄ EBIT(alt) + β₅ Sales(alt) + β₆ ΔMktSh + β₇ ΔEmply + β₈ LitStlm + β₉ LitRisk + β₁₀ MergAcq + β₁₁ LevBuy + β₁₂ ΔAuditor + β₁₃ ΔAudOp + ε     (2)

Where the financial variables are as previously defined and the remaining variables are defined as follows:
ΔMktSh (Growth potential proxy) = change in firm sales from prior year to current year as a ratio of the total sales of the industry (data item 12 / industry total sales);
ΔEmply (Change in number of employees) = the difference in the log of number of employees from prior year to current year;
LitRisk (Litigation risk proxy) = an indicator variable set to 1 for litigious industry codes: bio-technology (2833-2836), computer hardware (3570-3577), electronics (3600-3674), retailing (5200-5961), computer software (7371-7379), and 0 otherwise;
LitStlm (participation in litigation settlement or insurance payout) = an indicator variable set to 1 if the firm participates in a litigation settlement or insurance payout (data 372), and 0 otherwise;
MergAcq = an indicator variable set to 1 if the firm participates in a merger/acquisition, and 0 otherwise;
LevBuy = an indicator variable set to 1 if the firm participates in a leveraged buyout, and 0 otherwise;
ΔAuditor = an indicator variable set to 1 if the auditor was changed from prior year to current year, and 0 otherwise; and
ΔAudOp = an indicator variable set to 1 if the audit opinion in current year was changed to a more adverse one than in the prior year, and 0 otherwise.

The non-financial and operational measures included in the bankruptcy model are based on prior research. Firstly, ΔMktShr is used as a proxy for growth potential (e.g. Amir & Lev 1996; Riley et al., 2003). We expect a negative coefficient since reduction in growth potential is an indicator of financial distress.

The change in employees, ΔEmply, is included as it has been shown that workforce reduction or downsizing may be associated with bankruptcy (Perotti & Spier, 1993; Stiglitz et al., 2003; Powell & Yawson, 2007; Campbell, 2008). As before, we expect a negative coefficient in the regression.

We employ two measures to proxy for litigation risk since prior research shows that filing for bankruptcy is sometimes used to circumvent litigation (Rice & Davis, 1999; Plevin, 2003). The first is litigation settlement (LitStlm) indicating whether the firm had settled any litigation claims during the year. We also include a litigation risk proxy (LitRisk) adopted from Francis et al. (1994) and Frankel et al. (2002) based on industry membership. We expect a negative coefficient for both variables, although the litigation risk proxy may be too general to provide significant results.

Variables that proxy for corporate restructuring are also included in the model, namely merger/acquisition (MergAcq) and leveraged buyout (LevBuy). Both merger/acquisition and leveraged buyouts are expected to eliminate the decision to file for bankruptcy (Gilson et al., 1990) so we expect a negative coefficient sign.

We also include the change in auditor (ΔAuditor) and expect a positive coefficient, indicating that a move to a different auditor will be associated with a higher likelihood of bankruptcy. This is expected given that change in auditor is associated with the perception of lower quality audits (Robertson et al., 2014) or an indication of issues with current auditors around going-concern (Senteney et al., 2006).³

Finally, we include the change in auditor opinion (ΔAudOp), and expect a positive coefficient, indicating that a move to a less favorable opinion will be associated with a higher likelihood of bankruptcy.

The appendix to the paper presents definitions and the measurement of the financial and non-financial or operational measures used in the above model.

Descriptives and Empirical Results

Descriptives

Table 2, panel A, presents the descriptive statistics and univariate differences for the variables in the sample of 942 bankruptcy firm-year observations and 48,506 non-bankruptcy firm-year observations.

Table 1 Panel B Yearly Distribution Of Bankruptcy And Non-Bankruptcy Firms
	Bankruptcy Firms		Non-Bankruptcy Firms
	N	%	N	%
1978	0	0	1,249	2.57
1979	0	0	1,242	2.56
1980	0	0	1,215	2.5
1981	4	1.27	1,206	2.49
1982	16	5.1	1,166	2.4
1983	4	1.27	1,381	2.85
1984	8	2.55	1,375	2.83
1985	9	2.87	1,362	2.81
1986	11	3.5	1,368	2.82
1987	4	1.27	1,388	2.86
1988	5	1.59	1,356	2.8
1989	4	1.27	1,347	2.78
1990	11	3.5	1,330	2.74
1991	9	2.87	1,370	2.82
1992	8	2.55	1,430	2.95
1993	12	3.82	1,555	3.21
1994	3	0.96	1,669	3.44
1995	5	1.59	1,806	3.72
1996	5	1.59	1,927	3.97
1997	5	1.59	2,049	4.22
1998	13	4.14	2,109	4.35
1999	18	5.73	2,167	4.47
2000	39	12.42	2,268	4.68
2001	37	11.78	2,180	4.49
2002	31	9.87	2,199	4.53
2003	23	7.32	2,232	4.6
2004	11	3.5	2,308	4.76
2005	12	3.82	2,304	4.75
2006	5	1.59	1,948	4.02
2007	2	0.64	0	0
Total	314	100	48,506	100

The Table 2, panel A, shows that mean (median) financial statement ratios of non-bankruptcy firms, other than sales (S(alt)), are significantly greater than that of the ratios of bankruptcy firms. For example, the mean (median) working capital ratio (WC(alt)) is -0.004(0.087) and 0.193(0.170) for bankruptcy and non-bankruptcy firms, respectively. The mean (median) market value of equity ratio (MVE(alt)) is 1.059(0.277) and 3.410(1.382) for bankruptcy and non-bankruptcy firms, respectively. Bankruptcy firms also have a lower increase in employees and less prevalence of litigation settlement than non-bankruptcy firms (mean and median differences in ΔEmply are -0.034 and -0.041, respectively, significant at the 1% level, and mean difference in LitStlm is -1.59%, significant at the 5% level). Furthermore, bankruptcy firms have a higher incidence of change in auditors as well as opinion (mean difference in ΔAuditor is 1.81%, significant at the 5% level, and mean difference in ΔAudOp is 9.96%, significant at the 1% level).

Table 2 Panel A  Descriptive Statistics: Comparative Mean & Median Values Of Variables For Bankruptcy & Non-Bankruptcy Firms
	Bankruptcy  Firms		Non-Bankruptcy Firms
Variable	Mean	Median	Mean	Median	Mean difference	Median difference
WC(alt)	-0.004	0.087	0.193	0.17	-0.197a	-0.084a
RE(alt)	-0.506	-0.046	0.18	0.219	-0.686a	-0.264a
EBIT(alt)	-0.052	0.027	0.092	0.092	-0.144a	-0.065a
MVE(alt)	1.059	0.277	3.41	1.382	-2.351a	-1.105a
S(alt)	1.149	1.029	1.151	0.993	-0.002	0.036
?MktSh	0.407	0.046	0.377	0.097	0.03	-0.051a
?Emply	0.147	-0.015	0.182	0.027	-0.034	-0.041a
LitStlm	3.50%		5.09%		-1.59%b
LitRisk	21.34%		23.88%		-2.54%c
MergAcq	21.34%		38.12%		-16.78%a
LevBuy	0.00%		0.46%		-0.46%b
DAuditor	6.79%		4.98%		1.81%b
DAudOp	20.49%		10.53%		9.96%a
Firm-Year Observations	942			48,506

Table 2, panel B, provides Pearson correlation coefficients between the financial and non-financial measures. The correlation between RE(alt) and EBIT(alt) is relatively high (r = 0.575). However, concerns with collinearity problems are customarily overlooked in the Altman (1968) model in favor of retaining this combination of variables in the model. Moreover, since the group performs well in discriminant analysis, the literature historically utilizes the model as it is. As for the operational effects variables, while many of the variables are significantly related to each other, the coefficient values of the related variables are rather low, with the highest coefficient value at r = -0.137 for the negative correlation between MergAcq and LitStlm. Hence, none of these correlations appears large enough to present collinearity problems.

Table 2 Panel B Pearson Correlation Coefficients: N=49,448
	RE(alt)	EBIT(alt)	MVE(alt)	S(alt)	?MktSh	?Emply	LitStlm	LitRisk	MergAcq	LevBuy	DAuditor	?AudOp
WC(alt)	0.246a	0.208a	0.241a	0.154a	0.000	-0.004	-0.013a	0.234a	0.026a	0.009b	-0.043a	-0.056a
RE(alt)		0.575a	0.018a	0.080a	-0.007	-0.005	-0.037a	-0.073a	0.006	0.012a	-0.032a	-0.045a
EBIT(alt)			0.072a	0.163a	-0.009b	-0.003	-0.011b	-0.035a	-0.002	0.010b	-0.036a	-0.059a
MVE(alt)				-0.064a	0.004	0.004	0.014a	0.189a	-0.033a	-0.015a	0.006	-0.008c
S(alt)					-0.010b	-0.006	-0.032a	0.135a	0.029a	0.043a	-0.043a	-0.040a
?MktSh						0.017a	-0.003	-0.003	-0.005	-0.001	-0.001	0.008c
?Emply							-0.002	0.000	0.006	-0.001	0.021a	-0.003
LitStlm								0.028a	-0.137a	-0.016a	-0.008c	0.071a
LitRisk									-0.027a	-0.005	0.005	0.019a
MergAcq										-0.053a	-0.023a	-0.059a
LevBuy											-0.011b	-0.021a
?Auditor												0.018a

Results of Logit Regressions

We begin by presenting the results of the logistic regressions. Table 3 provides the empirical results for the sample of bankrupt and non-bankrupt firms one-year, two-years, and three-years prior to bankruptcy (Liedtka, 2002).

Table 3 Logistic Regression Coefficients (P-Values) For Sample Of Non-Bankrupt Firms And Bankrupt Firms One-Year, Two-Years And Three-Years Prior To Bankruptcy: N=48,820
		One-year prior to bankruptcy		Two-years prior to bankruptcy		Three-years prior to bankruptcy
Independent Variable	Sign             Estimate	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
Intercept	?	-2.198	-2.173	-2.257	-2.159	-2.268	-2.133
		(0.000a)	(0.000a)	(0.000a)	(0.000a)	(0.000a)	(0.000a)
WC(alt)	_	-1.687	-1.617	-0.520	-0.487	0.002	0.012
		(0.000a)	(0.000a)	(0.000a)	(0.002a)	(0.988)	(0.940)
RE(alt)	_	-0.071	-0.066	-0.040	-0.039	-0.038	-0.039
		(0.000a)	(0.000a)	(0.004a)	(0.003a)	(0.001a)	(0.000a)
EBIT(alt)	_	-2.325	-2.253	-1.984	-2.013	-1.458	-1.485
		(0.000a)	(0.000a)	(0.000a)	(0.000a)	(0.000a)	(0.000a)
MVE(alt)	_	-0.136	-0.132	-0.042	-0.043	-0.073	-0.076
		(0.098c)	(0.021b)	(0.396)	(0.422)	(0.011b)	(0.013b)
S(alt)	_	0.118	0.122	0.045	0.052	0.016	0.020
		(0.000a)	(0.000a)	(0.041b)	(0.019b)	(0.503)	(0.395)
?MktShr	_		-0.092		0.000		0.000
			(0.362)		(0.257)		(0.392)
?Emply	_		-0.055		0.000		0.000
			(0.816)		(0.817)		(0.186)
LitStlm	_		-0.201		-0.175		-0.372
			(0.056c)		(0.145)		(0.005a)
LitRisk	_		0.040		-0.046		-0.011
			(0.616)		(0.640)		(0.888)
MergAcq	_		-0.347		-0.324		-0.326
			(0.000a)		(0.000a)		(0.000a)
LevBuy	_		-4.502		-4.112		-3.000
			(0.000a)		(0.000a)		(0.000a)
?Auditor	+		0.121		0.060		-0.179
			(0.190)		(0.554)		(0.098c)
?AudOp	+		0.381		0.058		-0.062
			(0.000a)		(0.428)		(0.399)
Industry		Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
Year		Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
Psuedo R2		26.14%	28.79%	8.22%	9.65%	5.07%	6.78%

The first columns of results provide the empirical results of the logistic regression analysis utilized to replicate the Altman (1968) model (model 1) as well as the combined model for bankruptcy firms (model 2) one-year prior to bankruptcy. Results of model 1 indicate that four of the five financial statement variables (working capital, retained earnings, earnings before interest and taxes, and market value of equity) are significantly related to the probability of bankruptcy in the predicted direction. Sales is significantly related to the probability of bankruptcy but in the opposite direction (coefficient on S(alt) is 0.118, significant at the 1% level). Results for model 2 which includes the operational variables show the same pattern as model 1 on the five financial ratio variables. Furthermore, four operational factors are significantly associated with the probability of bankruptcy: litigation settlement is significantly negative (coefficient = -0.201, significant at the 10% level), indicating that bankrupt firms are less likely to enter into settlement agreements; merger and acquisition activity is significantly negative (coefficient = -0.347, significant at the 1% level), indicating that bankruptcy firms are less likely to enter into merger and acquisition activities one year prior to bankruptcy; incidence of a leveraged buyout has a significant and negative coefficient (coefficient = -4.502, significant at the 1% level), indicating that bankrupt firms are less likely to engage in a leveraged buyout one year prior to bankruptcy; and the change in the auditor opinion is positive and significant (coefficient = 0.381, significant at the 1% level), indicating that bankrupt firms are more likely to move to a less favorable opinion in the year prior to bankruptcy. The explanatory power of the model is improved (R²=28.79%) as compared to the Altman (1968) model (R²=26.14%). As expected one year prior to bankruptcy, after controlling for financial measures, as hypothesized, several non-financial and operational measures are significantly related to future bankruptcy: litigation settlement, merger and acquisition activity, incidence of a leveraged buyout and the change in the auditor opinion with the likelihood of moving to a less favorable opinion in the year prior to bankruptcy.

The next two columns present the results of the logistic regression two years prior to bankruptcy. The Pseudo R² for model 2 is 9.65% as compared to that of the Altman model of 8.22%. The coefficients are similar to those one year prior to bankruptcy, except the market value of equity is not significant in either model two years prior to bankruptcy. The two variables, litigation settlement and change in auditor opinion, are no longer significantly related to the probability of bankruptcy, although the direction of relationship remains as before. As expected, two years prior to bankruptcy, leveraged buyout and merger and acquisition activity continue as significantly strong indications of corporate failure. Change in market share slightly increased in significance, although the strength in significance two years prior to bankruptcy of the bankruptcy indicators, change in employees, litigation settlement, litigation risk, change in auditor and change in audit opinion unexpectedly declined. In that there remains a relationship between the hypothesized variables and bankruptcy, albeit weak, we interpret this as a possible timing issue i.e., indicative of untimely or delayed reporting or recognition of the hypothesized nonfinancial activities. Delays or untimely reporting typically impair the decision usefulness of proxies for firms’ performance. A primary underlying motivation for such delays is the impact of such proxies (for loss recognition) on equity incentives (see Dechow et al., 2011, 264).

The final two columns present the results three years prior to bankruptcy. As before, the explanatory power of model 2 exceeds that of model 1. The Pseudo R² for model 2 is 6.78% as compared to that of the Altman model of 5.07%. Two financial ratio variables, working capital and sales, are no longer significantly related to the probability of bankruptcy. Four non-financial and operational measures remain associated with bankruptcy: litigation settlement, undertaking a merger and acquisition, undertaking a leveraged buyout, and changing audit firms (coefficient of LitStlm is -0.372, significant at the 1% level; coefficient of MergAcq is -0.326, significant at the 1% level; coefficient of LevBuy is -3.000, significant at the 1% level; coefficient of ΔAuditor is -0.179, significant at the 10% level). As expected, years before their ultimate failure, firms tend to informally work toward resolving issues related to unbooked liabilities. For instance, our results indicate that three years prior to bankruptcy, firms attempt to control bad news by strategically settling litigation out-of-court, changing auditors, engaging in merger and acquisition negotiations, and partaking in leveraged buyouts.

Overall, the results show an improvement in models that include not only financial ratios, but non-financial and operational measures. The non-financial measures that are associated with the probability of bankruptcy are litigation settlement and restructuring in the form of a merger/acquisition or a leveraged buyout. Moreover, change in auditor opinion or change in auditor is also associated with a greater probability of bankruptcy, yet only when they occur in specific time intervals relative to the period of bankruptcy. Contrary to our expectations, one measure that was not found to be significantly associated with the probability of bankruptcy is the growth potential, captured by the change in market share and the change in number of employees. We also did not expect to find a differential effect across time periods for the change in auditor and change in auditor's opinion variables. Issues such as untimely reports of liabilities or unreported liabilities, unreported internal activities, and the use of data captured during business periods employing outdated accounting standards may be contributing factors that impair accurately measuring nonfinancial variables. Even though the above tests do not strongly indicate which model performs better in terms of detection rates equally across every pre-bankruptcy year, it is clear that there are non-financial or operational factors that are strongly associated with bankruptcy at different times in pre-bankruptcy decision process that are not captured in a purely financial detection model.

Classification of Bankruptcy Firms

In this section, we present discriminant analysis results for the financial and combined models to enable us to compare the misclassification of bankrupt and non-bankrupt firms using both models. However, rather than using the full sample to measure the discriminant analysis function as well as estimate misclassification, we use the U-method, which is based on the “leave one-out” principle. This method is similar to selecting a hold-out sample on which to test the discriminant analysis function. Specifically, the discriminant function is fitted to repeatedly drawn samples of the original population. Therefore, a dataset with 100 observations would involve 100 different discriminant analyses being performed, each on 99 of the 100 observations. Each time the discriminant function is calculated, it is used to classify the remaining observation that was not involved in the calculation of the function Table 4.

Table 4 Percentage Of Correctly Classified And Misclassified Bankruptcies And Non-Bankruptcies Using Discriminant Analysis For Models 1 And 2 Using U-Method N=48,820
	Model 1		Model 2
	Percent Correct	Percent Error	Percent Correct	Percent Error
One-year prior to bankruptcy
Bankrupt	57.32	42.68	60.83	39.17
Non-Bankrupt	95.86	4.14	95.35	4.65
Two-years prior to bankruptcy
Bankrupt	46.18	53.82	50.64	49.36
Non-Bankrupt	88.63	11.37	87.85	12.15
Three-years prior to bankruptcy
Bankrupt	40.76	59.24	50.96	49.04
Non-Bankrupt	83.89	16.11	80.84	19.16

We find that the combined model outperforms the financial model for each of the three years prior to bankruptcy. The results show that model 1 correctly identifies 57%, 46%, and 41% of bankrupt firms one-year, two-years and three-years prior to bankruptcy, respectively. By contrast, the combined model correctly identifies 61%, 51% and 51% of the bankruptcy firms for these same pre-bankruptcy periods. The range of difference for correct classification is from 4 to 10%. For all years prior to bankruptcy, correct classification of non-bankrupt firms is similar for both models with a range of difference from 1 to 3%. For example, for non-bankrupt firms, the financial model is successful in classifying 96%, 89% and 84% and the combined model classifies 95%, 88%, and 81% correctly one, two and three years prior to bankruptcy, respectively. The classification rates of bankrupt firms are much smaller than those found in previous literature (Altman, 1968; Ohlson, 1980) since their sampling methodology is over-represented with bankrupt firms and this can overestimate detection rates (Gilbert et al., 1990).

Hazard Model Results

Table 5 Proportional Hazard Model Coefficients (P-Values) For Sample Of Non-Bankrupt Firms And Bankrupt Firms One-Year, Two-Years And Three-Years Prior To Bankruptcy  N=48,820
		One-year prior to bankruptcy		Two-years prior to bankruptcy		Three-years prior to bankruptcy
Independent Variable	Sign             Estimate	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
WC(alt)	_	-0.146	-0.251	-0.673	-0.601	-0.114	-0.053
		(0.092c)	(0.005a)	(0.001a)	(0.003a)	-0.65	-0.83
RE(alt)	_	-0.021	-0.039	0.354	0.345	-0.094	-0.1
		-0.285	(0.061c)	(0.000a)	(0.000a)	(0.047b)	(0.044b)
MVE(alt)	_	-1.363	-0.851	-3.187	-3.141	-0.189	-0.14
		(0.000a)	(0.001a)	(0.000a)	(0.000a)	-0.255	-0.418
EBIT(alt)	_	-1.094	-1.048	-0.334	-0.365	-0.299	-0.328
		(0.000a)	(0.000a)	(0.000a)	(0.000a)	(0.000a)	(0.000a)
S(alt)	_	-0.19	-0.158	0.002	0.008	-0.038	-0.046
		(0.008a)	(0.030b)	-0.97	-0.908	-0.583	-0.519
?MktShr	_		-0.052		0		0
			-0.692		-0.96		-0.752
?Emply	_		-0.521		0		0.001
			(0.038b)		-0.966		-0.809
LitStlm	_		-0.337		-0.33		-1.021
			-0.238		-0.247		(0.008a)
LitRisk	+		0.461		0.206		0.195
			(0.002a)		-0.164		-0.187
MergAcq	_		-0.791		-0.881		-0.924
			(0.000a)		(0.000a)		(0.000a)
LevBuy	_		-11.549		-11.839		-11.792
			-0.961		-0.962		-0.959
?Auditor	+		0.344		0.108		-0.479
			(0.083c)		-0.634		-0.12
?AudOp	+		1.367		0.289		-0.05
			(0.000a)		(0.069c)		-0.785
χ2		442.57	626.31	213.78	270.93	122.52	187.43

Proportional hazard models relate the time that passes before some event occurs (in this case, bankruptcy) to variables that may be related to the event. The main benefit of using this methodology is that a hazard model is not static and controls for each firm’s period at risk (Shumway, 2001). Table 5 reports hazard model coefficients for individual model effects (analysis of maximum likelihood estimates) on the same variables reported in Table 3. The hazard analysis indicates that operational factors could lessen the importance of financial factors (working capital, retained earnings, and sales) in determining the likelihood of filing for bankruptcy the further away the analysis is from the bankruptcy filing.

Table 5 Proportional Hazard Model Coefficients (P-Values) For Sample Of Non-Bankrupt Firms And Bankrupt Firms One-Year, Two-Years And Three-Years Prior To Bankruptcy  N=48,820
		One-year prior to bankruptcy		Two-years prior to bankruptcy		Three-years prior to bankruptcy
Independent Variable	Sign             Estimate	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
WC(alt)	_	-0.146	-0.251	-0.673	-0.601	-0.114	-0.053
		(0.092c)	(0.005a)	(0.001a)	(0.003a)	-0.65	-0.83
RE(alt)	_	-0.021	-0.039	0.354	0.345	-0.094	-0.1
		-0.285	(0.061c)	(0.000a)	(0.000a)	(0.047b)	(0.044b)
MVE(alt)	_	-1.363	-0.851	-3.187	-3.141	-0.189	-0.14
		(0.000a)	(0.001a)	(0.000a)	(0.000a)	-0.255	-0.418
EBIT(alt)	_	-1.094	-1.048	-0.334	-0.365	-0.299	-0.328
		(0.000a)	(0.000a)	(0.000a)	(0.000a)	(0.000a)	(0.000a)
S(alt)	_	-0.19	-0.158	0.002	0.008	-0.038	-0.046
		(0.008a)	(0.030b)	-0.97	-0.908	-0.583	-0.519
?MktShr	_		-0.052		0		0
			-0.692		-0.96		-0.752
?Emply	_		-0.521		0		0.001
			(0.038b)		-0.966		-0.809
LitStlm	_		-0.337		-0.33		-1.021
			-0.238		-0.247		(0.008a)
LitRisk	+		0.461		0.206		0.195
			(0.002a)		-0.164		-0.187
MergAcq	_		-0.791		-0.881		-0.924
			(0.000a)		(0.000a)		(0.000a)
LevBuy	_		-11.549		-11.839		-11.792
			-0.961		-0.962		-0.959
?Auditor	+		0.344		0.108		-0.479
			(0.083c)		-0.634		-0.12
?AudOp	+		1.367		0.289		-0.05
			(0.000a)		(0.069c)		-0.785
χ2		442.57	626.31	213.78	270.93	122.52	187.43

The results show that, in Model 1, one year before bankruptcy, all but one of the financial variables (retained earnings) are associated with bankruptcy. However, some financial variables lose significance in periods over one year prior to bankruptcy. For example, two years prior to bankruptcy, sales loses significance in both models and three years prior to bankruptcy, three variables, working capital, market value of equity and sales lose significance in both models. Similarly, the operational variables that are associated with bankruptcy vary over the years. For example, one year prior to bankruptcy, in the combined model, five of the operational variables (change in employees, litigation risk, merger/ acquisition, auditor change and audit opinion change) included in the combined model are associated with bankruptcy (coefficient of ΔEmply is -0.521, significant at the 5% level; coefficient of LitRisk is 0.461, significant at the 1% level; coefficient of MergAcq is -0.791, significant at the 1% level; coefficient of ΔAuditor is 0.344, significant at the 10% level; coefficient of ΔAudOp is 1.367, significant at the 1% level). The results show the failure of bankruptcy firms to effectively restructure through the self-help mechanisms of consummating relationships with new business alliances, layoffs of the workforce, and distancing the firms from past auditors and audit opinions one year before bankruptcy. Moreover, the effects of litigation risk indicate a strong association between bankruptcy and the firm’s industry placement (Maines et al., 2002).

Two years prior to bankruptcy, decreased working capital, increased retained earnings, decreased earnings, decreased market value of equity, audit opinion change and merger/ acquisition are strongly associated with a firm’s risk of filing bankruptcy (coefficient of MergAcq is -0.881, significant at the 1% level; coefficient of DAudOp is 0.289, significant at the 10% level). Three years prior to bankruptcy, only two financial variables (retained earnings and earnings) and two operational variables (litigation settlement and merger/acquisition) are associated with the likelihood of filing bankruptcy (coefficient of LitStlm is -1.021, significant at the 1% level; coefficient of MergAcq is -0.924, significant at the 1% level).

Two years prior to bankruptcy, decreased working capital, increased retained earnings, decreased earnings, decreased market value of equity, audit opinion change and merger/ acquisition are strongly associated with a firm’s risk of filing bankruptcy (coefficient of MergAcq is -0.881, significant at the 1% level; coefficient of DAudOp is 0.289, significant at the 10% level). Three years prior to bankruptcy, only two financial variables (retained earnings and earnings) and two operational variables (litigation settlement and merger/acquisition) are associated with the likelihood of filing bankruptcy (coefficient of LitStlm is -1.021, significant at the 1% level; coefficient of MergAcq is -0.924, significant at the 1% level).

In comparing the results from the logistic regression to the hazard model, we find that the results are similar, except that the change in employees’ variable is only significant using the hazard model one-year prior to bankruptcy. Furthermore, litigation settlement is only significant three years prior to bankruptcy using the hazard model in contrast to the logistic regression results in which it is also significant one-year prior to bankruptcy. While the results indicate a negative association between bankruptcy and leveraged buyouts, the coefficients lose their significance in the hazard model in contrast to the strong significance in the logistic regression model for all three years prior to bankruptcy. While the effects of auditor change and audit opinion change vary, the hazard model indicates a strong direct association between bankruptcy and auditor change in the short-term (one year prior to bankruptcy), as well as a strong link to audit opinion change two years and one year prior to bankruptcy.

Overall, the significant operational variables in the hazard model that are consistent with the logistic regression model are litigation settlement and mergers and acquisitions, as well as the change in the auditor’s opinion (Beaver et al., 2005, Powell & Yawson, 2007).

Conclusion

Bankruptcy research indicates that there are now secular changes in the explanatory power of financial ratios with respect to bankruptcy. In particular, forces over the last 40 years potentially affect the ability of financial ratios to predict bankruptcy. This study extends bankruptcy prediction research by investigating a subset of variables that provide explanatory power to bankruptcy models. The extended model that includes non-financial and operational measures in addition to the traditional financial ratios is tested for its overall performance over a series of three pre-bankruptcy years.  We find that these variables prove to be significantly useful indicators of forthcoming bankruptcy. The empirical results provide evidence that the informativeness of financial ratios weakens as the length between the prediction period and the bankruptcy filing date increases. Furthermore, we find that the performance of a model that includes both financial ratios and non-traditional measures is stronger than a model that includes only financial variables. These findings show that these non-traditional measures are informative and provide relevant and timely signals to stakeholders in advance of bankruptcy. Our findings contribute to the literature by developing a relevant theoretically-based model which includes qualitative considerations. The results suggest that due to information asymmetry in financial reporting, other factors should be considered in determining the likelihood of bankruptcy.

Going forward, Maines advance the notion “that research suggests that noncomparability among types and formats likely hampers investors’ ability to use nonfinancial measures. There is a trade-off, however, between ease of comparability and having firms provide disclosures that reflect the economics of the firm and management's strategy for the firm.” Hence, we contend that to the extent that policymakers, stockholders, bondholders and other stakeholders would benefit from early detection of signs of bankruptcy, this study provides insight into important factors associated with bankruptcy.

Appendix

Appendix Variable Definitions
Variable Name	Measurement	Definition	COMPUSTAT Data Item Number
Financial Variables:
WC(alt)	WC/Ta	Working capital	179/6
RE(alt)	RE/Ta	Retained earnings	36/6
EBIT(alt)	EBIT/Ta	Earnings before interest and tax	178/6
MVE(alt)	MVE/TL	Market value of equity	199*25/181
S(alt)	S/Ta	Sales	6-Dec
Non-financial and Operational Variables:
?MktShr	Change in firm revenue / Total revenue of industry from prior year to current year	Change in market share	12
? Emply	Change in log(Emply) from prior year to current year	Change in employees (logarithmic)	29
LitStlm	1=Litigation settlement, 0=otherwise	Litigation settlement dummy	372
LitRisk	1=Litigious Industries,	Litigation risk
	0=otherwise
MergAcq	1=had a merger or acquisition,	Mergers and acquisitions
	0=otherwise
LevBuy	1=underwent a leveraged buyout,	Leveraged buyout
	0=otherwise
?Auditor	1=auditor in current year is different than that in prior year, adjusted for Arthur Anderson change in year 2002,
	0=otherwise	Auditor change	149
?AudOp	1=auditor opinion in current year is more adverse than that in prior year, 0=otherwise	Auditor opinion change	149

Acknowledgement

1The finance literature includes several papers measuring the costs associated with bankruptcy (e.g., Warner, 1977; Ang, Chua, and McConnell 1982; Bris, Welch, and Zhu 2006).

2Firm and year subscripts have been deleted for simplicity.

3We treat a change from Arthur Anderson to any auditor during the year 2002 as a no-change observation since firms were forced to change auditors following the bankruptcy of Arthur Anderson.

References

Agarwal, V., & Taffler, R. (2008). Comparing the performance of market-based and accounting-based bankruptcy prediction models.Journal of Banking & Finance,32(8), 1541-1551.

Indexed at, Google Scholar, Cross Ref

Altman, E.I. (1968). Financial ratios, discriminant analysis, and the prediction of corporate bankruptcy. Journal of Finance, 23(4), 589-609.

Indexed at, Google Scholar, Cross Ref

Altman, E.I. (1984). A further empirical investigation of the bankruptcy cost question.The Journal of Finance,39(4), 1067-1089.

Indexed at, Google Scholar, Cross Ref

Altman, E.I. (2000). Predicting Financial Distress of Companies. Working paper, New York University.

Indexed at, Google Scholar, Cross Ref

Amir, E., & Lev, B. (1996). Value-relevance of nonfinancial information: The wireless communications industry. Journal of Accounting and Economics, 22(1), 3-30.

Indexed at, Google Scholar, Cross Ref

Bauer, J., & Agarwal, V. (2014). Are hazard models superior to traditional bankruptcy prediction approaches? A comprehensive test.Journal of Banking & Finance,40, 432-442.

Indexed at, Google Scholar, Cross Ref

Beams, J.D., Huang, H., & Yan, Y. (2013). Top management resignation and firms' subsequent bankruptcy. Accounting and the Public Interest, 13(1), 39-54.

Indexed at, Google Scholar, Cross Ref

Beaver, W. (1966). Financial ratios as predictors of failure. Journal of Accounting Research, 4, 71-111.

Indexed at, Google Scholar, Cross Ref

Beaver, W., McNichols, M.F., & Rhie, J.W. (2005). Have financial statements become less informative? Evidence from the ability of financial ratios to predict bankruptcy. Review of Accounting Studies, 10(1), 93-122.

Indexed at, Google Scholar, Cross Ref

Boraty?ska, K. (2016). FsQCA in corporate bankruptcy research. An innovative approach in food industry.Journal of Business Research,69(11), 5529-5533.

Indexed at, Google Scholar, Cross Ref

Bris, A., Welch, I., & Zhu, N. (2006). The costs of bankruptcy: Chapter 7 liquidation versus chapter 11 reorganization.The Journal of Finance,61(3), 1253-1303.

Indexed at, Google Scholar, Cross Ref

Campbell, D. (2008). Nonfinancial performance measures and promotion-based incentives.Journal of Accounting Research,46(2), 297-332.

Indexed at, Google Scholar, Cross Ref

Charitou, A., Lambertides, N., & Trigeoris, L. (2007). Managerial discretion in distressed firms. British Accounting Review, 9(4), 323-346.

Indexed at, Google Scholar, Cross Ref

Cochran, J.J., Darrat, A.F., & Elkhal, K. (2006). On the bankruptcy of internet companies: An empirical inquiry.Journal of Business Research,59(10), 1193-1200.

Indexed at, Google Scholar, Cross Ref

Dechow, P.M., Ge, W., Larson, C.R., & Sloan, R.G. (2011). Predicting material accounting misstatements. Contemporary Accounting Research, 28(1), 17-82.

Indexed at, Google Scholar, Cross Ref

Eisdorfer, A., & Hsu, P. (2011). Innovate to survive: The effect of technology competition on corporate bankruptcy.Financial Management,40(4), 1087-1117.

Indexed at, Google Scholar, Cross Ref

Francis, J.D., Philbrick, D., & Schipper K. (1994). Shareholder litigation and corporate disclosures. Journal of Accounting Research, 32, 137-164.

Indexed at, Google Scholar, Cross Ref

Frankel, R.M., Johnson, M., & Nelson, K. (2002). The relation between auditors' fees for nonaudit services and earnings management. The Accounting Review, 77(s-1), 71-105.

Indexed at, Google Scholar, Cross Ref

Gilbert, L.R., Menon, K., & Schwartz, K.B. (1990). Predicting bankruptcy for firms in financial distress. Journal of Business Finance & Accounting, 17(1), 161-171.

Indexed at, Google Scholar, Cross Ref

Gilson, S.C., John, K., & Lang, L.H.P. (1990). Troubled debt restructurings: An empirical study of private reorganization of firms in default. Journal of Financial Economics, 27(2), 315-353.

Indexed at, Google Scholar, Cross Ref

Giordani, P., Jacobson, T., von Schedvin, E., & Villani, M. (2014). Taking the twists into account: Predicting firm bankruptcy risk with splines of financial ratios. Journal of Financial and Quantitative Analysis,49(4), 1071-1099.

Indexed at, Google Scholar, Cross Ref

Grice, J.S., & Dugan, M.T. (2001). The limitations of bankruptcy prediction models: Some cautions for the researcher.Review of Quantitative Finance and Accounting,17(2), 151-166.

Indexed at, Google Scholar, Cross Ref

Grice, J.S., & Ingram, R.W. (2001). Tests of the generalizability of Altman's bankruptcy prediction model.Journal of Business Research,54(1), 53-61.

Indexed at, Google Scholar, Cross Ref

Hillegeist, S.A., Keating, E.K., Cram, D.P., & Lundstedt, K.G. (2004). Assessing the probability of bankruptcy.Review of Accounting Studies,9(1), 5-34.

Google Scholar

Ittner, C. D., & Larcker, D.F. (1998a). Are nonfinancial measures leading indicators of financial performance? An analysis of customer satisfaction. Journal of Accounting Research, 36(Supplement), 1-35.

Indexed at, Google Scholar, Cross Ref

Ittner, C.D., & Larcker, D.F. (1998b). Innovations in performance measurement: Trends and research implications.Journal of Management Accounting Research,10, 205-238.

Indexed at, Google Scholar

Jones, S. (2017). Corporate bankruptcy prediction: A high dimensional analysis. Review of Accounting Studies, 22(3), 1366-1422.

Indexed at, Google Scholar, Cross Ref

Jones, S., Johnstone, D., & Wilson, R. (2017). Predicting corporate bankruptcy: An evaluation of alternative statistical frameworks.Journal of Business Finance & Accounting,44(1-2), 3-34.

Indexed at, Google Scholar, Cross Ref

Keasey, K., & Watson, R. (1987). Non-financial symptoms and the prediction of small company failure: A test of  Argenti's hypothesis. Journal of Business Finance & Accounting, 14(3), 335-354.

Indexed at, Google Scholar, Cross Ref

Kim, Y., & Nabar, S. (2007). Bankruptcy probability changes and the differential informativeness of bond upgrades and downgrades.Journal of Banking & Finance,31(12), 3843-3861.

Indexed at, Google Scholar, Cross Ref

Knox, K.J., Blankmeyer, E.C., Trinidad, J.A., & Stutzman, J.R. (2009). Predicting bankruptcy in the Texas nursing facility industry.Quarterly Review of Economics and Finance,49(3), 1047-1064.

Indexed at, Google Scholar, Cross Ref

Li, M.Y.L., & Miu, P. (2010). A hybrid bankruptcy prediction model with dynamic loadings on accounting-ratio-based and market-based information: A binary quantile regression approach.Journal of Empirical Finance,17(4), 818-833.

Indexed at, Google Scholar, Cross Ref

Liedtka, S. (2002). The information content of nonfinancial measures in the airline industry. Journal of Business Finance & Accounting, 29(7-8), 1105-1121.

Indexed at, Google Scholar, Cross Ref

Maines, L.A., Bartov, E., Fairfield, P.M., Hirst, D.E., Iannaconi, T.E., Mallett, R., & Vincent, L. (2002). Recommendations on disclosure of nonfinancial performance measures. Accounting Horizons, 16(4), 353-362.

Indexed at, Google Scholar, Cross Ref

Mousavi, M.M., Ouenniche, J., & Xu, B. (2015). Performance evaluation of bankruptcy prediction models: An orientation-free super-efficiency DEA-based framework.International Review of Financial Analysis,42, 64-75.

Indexed at, Google Scholar, Cross Ref

Ohlson, J.A. (1980). Financial ratios and the probabilistic prediction of bankruptcy. Journal of Accounting Research, 18(1), 109-131.

Indexed at, Google Scholar, Cross Ref

Perotti, E.C., & Spier, K.E. (1993). Capital structure as a bargaining tool: The role of leverage in contract renegotiation. The American Economic Review, 83(5), 1131-1141.

Indexed at, Google Scholar

Philosophov, L.V., & Philosophov, V.L. (2002). Corporate bankruptcy prognosis: An attempt at a combined prediction of the bankruptcy event and time interval of its occurrence.International Review of Financial Analysis,11(3), 375-406.

Indexed at, Google Scholar, Cross Ref

Plevin, M.D. (2003). Asbestos bankruptcy basics and selected related insurance issues, presented at Mealey's Asbestos Bankruptcy Conference, Dallas, June 3.

Google Scholar

Powell, R., & Yawson. A. (2007). Are corporate restructuring events driven by common factors? Implications for takeover prediction. Journal of Business Finance & Accounting, 34(7-8), 1169-1192.

Indexed at, Google Scholar, Cross Ref

Rice, J.F., & Davis, N.W. (1999). The future of mass tort claims: Comparison of settlement class action to bankruptcy treatment of mass tort claims. South Carolina Law Review, 50(Winter), 405.

Google Scholar

Riley, R., Pearson, T.A., & Trompeter, G. (2003). The value relevance of non-financial performance variables and accounting information: The case of the airline industry. Journal of Accounting and Public Policy, 22(3), 231-254.

Indexed at, Google Scholar, Cross Ref

Robertson, J.C., Stefaniak, C.M., & Houston, R.W. (2014). Do PCAOB inspection reports influence corporate executives' perceptions of audit quality and the likelihood of switching auditors? Accounting and the Public Interest, 14(1), 48-71.

Indexed at, Google Scholar, Cross Ref

Rose-Green, E., & Dawkins, M. (2002). Strategic bankruptcies and price reactions to bankruptcy filings. Journal of Business Finance & Accounting, 29(9-10), 1319-1335.

Indexed at, Google Scholar, Cross Ref

Rosner, R. (2003). Earnings manipulation in failing firms. Contemporary Accounting Research, 20(2), 361-408.

Indexed at, Google Scholar, Cross Ref

Said, A.A., HassabElnaby, H.R., & Wier, B. (2003). An empirical investigation of the performance consequences of nonfinancial measures.Journal of Management Accounting Research,15, 193-223.

Indexed at, Google Scholar, Cross Ref

Senteney, D.L., Bazaz, M.S., & Ahmadpour, A. (2006). Tests of the incremental explanatory power of auditor qualified opinion and audit firm changes in predicting impending bankruptcy. International Journal of Accounting, Auditing and Performance Evaluation, 3(4), 434-451.

Indexed at, Google Scholar, Cross Ref

Shumway, T. (2001). Forecasting bankruptcy more accurately: A simple hazard model. The Journal of Business, 74(1), 101-124.

Indexed at, Google Scholar, Cross Ref

Smith, M., & Taffler, R.J. (2000). The chairman's statement - A content analysis of discretionary narrative disclosures. Accounting, Auditing & Accountability Journal, 13(5), 624-646.

Indexed at, Google Scholar, Cross Ref

Stiglitz, J.E., Orszag, J.M., & Orszag, P.R. (2003). The impact of asbestos liabilities on workers in bankrupt firms. Norton Journal of Bankruptcy Law and Practice, 12(1), 51-96 or 98.

Google Scholar

Sun, L. (2007). A re-evaluation of auditors' opinions versus statistical models in bankruptcy prediction.Review of Quantitative Finance and Accounting,28(1), 55-78.

Indexed at, Google Scholar, Cross Ref

Tian, S., Yu, Y., & Guo. H. (2015). Variable selection and corporate bankruptcy forecasts.Journal of Banking & Finance,52, 89-100.

Indexed at, Google Scholar, Cross Ref

Tinoco, M.H., & Wilson, N. (2013). Financial distress and bankruptcy prediction among listed companies using accounting, market and macroeconomic variables.International Review of Financial Analysis, 30, 394-419.

Indexed at, Google Scholar, Cross Ref

Warner, J.B. (1977). Bankruptcy costs - Some evidence.The Journal of Finance,32(2), 337-347.

Indexed at, Google Scholar, Cross Ref

Wu, Y., Gaunt, C., & Gray, S. (2010). A comparison of alternative bankruptcy prediction models.Journal of Contemporary Accounting & Economics, 6(1), 34-45.

Indexed at, Google Scholar, Cross Ref

Xu, M., & Zhang, C. (2009). Bankruptcy prediction: The case of Japanese listed companies.Review of Accounting Studies,14(4), 534-558.

Indexed at, Google Scholar, Cross Ref

Zmijewski, M.E. (1984). Methodological issues related to the estimation of financial distress prediction models.Journal of Accounting Research, 22(Supplement), 59-82.

Indexed at, Google Scholar, Cross Ref

Received: 22-Sep-2022, Manuscript No. AAFSJ-22-12599; Editor assigned: 23-Sep -2022, PreQC No. AAFSJ-22-12599(PQ); Reviewed: 07-Oct-2022, QC No. AAFSJ-22-12599; Revised: 11-Oct-2022, Manuscript No. AAFSJ-22-12599(R); Published: 18-Oct-2022