Is Chapter 11 efficient?
The efficiency of the Chapter 11 bankruptcy process is examined by estimating the impact of Chapter 11 filings on the operating performance of bankrupt firms. We control for firm-level heterogeneity in prefiling characteristics using matching methods to select benchmark firms comparable to filing firms. We compare bankrupt firms' operating performances with those of matched nonbankrupt firms. Our results challenge the contention that Chapter 11 is an inefficient, debtor-friendly mechanism that rehabilitates economically nonviable firms. We demonstrate that firms that file under Chapter 11 perform no worse and, if anything, better than comparable nonfiling firms.
The efficiency of Chapter 11 bankruptcy proceedings has been the subject of debate among legal scholars and financial economists for a long time. Ideally, Chapter 11 provides a viable recontracting mechanism among claimholders of a bankrupt company mitigating transaction costs and bargaining problems associated with potentially value-enhancing restructuring initiatives (Aivazian and Callen, 1980, 1983; Brown, 1989; Giammarino, 1989; Wruck, 1990; Gilson, 1997). The implication is that Chapter 11 attenuates impediments to rational organizational and strategic changes so that collectively rational outcomes emerge for bankrupt firms. Critics of Chapter 11 contend that it is an overly debtor-friendly process that grants incumbent management too much controlling power and fails to liquidate a significant number of economically inefficient firms (Baird, 1986; Bebchuk, 1988, 2000; White, 1989, 1994; Bradley and Rosenzweig, 1991; Jensen, 1991; Aghion, Hart, and Moore, 1992). The contention is that Chapter 11 is an inefficient mechanism that rehabilitates economically nonviable firms. The results of this paper challenge this claim by demonstrating that firms that file under Chapter 11 perform no worse and, if anything, better than comparable nonfiling firms.
A growing literature assesses firm operating performance during Chapter 11 bankruptcy. Hotchkiss (1995) reports that 40% of reorganized firms continue to experience operating losses in the three years following bankruptcy. Maksimovic and Phillips (1998) find that filing firms' plant-level productivity and operating cash flows during Chapter 11 are not significantly different from those of their industry counterparts in low demand industries, but are worse than the performances of industry counterparts in high demand industries. Andrade and Kaplan (1998) estimate the costs of financial distress for 31 financially distressed companies by examining changes in operating and net cash flow margins from the onset of distress to its resolution. Adjusting for industry and market performance indicators, they find that firms' operating and net cash flow margins decline by 7% to 17% during Chapter 11 bankruptcy. Alderson and Betker (1999) study the total cash flows of reorganized firms over the five years after they emerge from bankruptcy and determine that total cash flows provide competitive returns when compared to the returns on benchmark portfolios. Zhang (2010) finds that when firms emerge from Chapter 11 bankruptcies, their industry competitors experience negative long-term equity returns and deteriorating financial performance. Denis and Rodgers (2007) relate the post-reorganization operating performance of filing firms to firm and industry characteristics and find that firms that significantly restructure their assets and liabilities during Chapter 11 are more likely to achieve positive industry-adjusted operating performance in the three years following emergence from bankruptcy.
Kalay, Singhal, and Tashjian (2007) examine changes in firm operating performance from the fiscal year-end immediately preceding the bankruptcy filing to the fiscal year-end immediately following emergence from bankruptcy. They follow Barber and Lyon (1996) and form a control group of firms that are in the same standard industrial classification (SIC) industries as the bankrupt firm and have performances within 10% of the performance of the bankrupt firm at the fiscal year-end preceding the firm's Chapter 11 filing. Using their industry and performance-adjusted measure, they find that firms experience significant improvement in their operating performance during Chapter 11, and that these improvements in operating performance are mainly related to the number of security classes in the reorganization plan and to the prefiling debt ratio.
We investigate the operating performance of a large number of public firms after they emerge from Chapter 11 reorganization. To assess the effect of Chapter 11 filings on firm operating performance, we benchmark on comparable firms to bankrupt firms in a way that reduces selection bias due to firm heterogeneity. To find the proper benchmark firms, it is important to account for self-selection problems and to control for firm heterogeneity in prefiling characteristics. Existing empirical studies use industry-adjusted or industry and performance-adjusted measures, to estimate the effect of bankruptcy on filing firms' operating performance. However, matching on one or two characteristics may not produce a proper group of comparable nonfiling firms since there may be substantial heterogeneity among firms and many reasons for filing for bankruptcy. We identify firm characteristics, such as profitability and liquidity, as well as measures reflecting agency and bargaining problems that affect the firm's bankruptcy filing decision. We then demonstrate that industry median firms, as well as industry and performance matched firms, are statistically different from bankruptcy filing firms in terms of prefiling characteristics. As such, they do not necessarily constitute a proper benchmark for assessing the effect of Chapter 11 filing on firm performance.
We employ empirical matching methods to account for the problems associated with endogenous filing decisions. We select from a comparison group of nonfiling firms, a group of control firms comparable to Chapter 11 filing firms in key prefiling characteristics. (1) We demonstrate that the matched (control) firms do not have statistically different attributes from the filing firms. We then compare changes in the operating performance of the control firms with those of bankrupt firms and find that when compared to the matched group, the filing firms' operating cash flows, on average, improve significantly from the prefiling year to the first postemergence year. These improvements are even larger when we consider average cash flow over the three years after emergence from bankruptcy. When compared with the industry-adjusted and industry and performance-adjusted estimation results, our empirical findings show that reorganized firms enjoy much greater improvement in operating cash flows after emerging from bankruptcy. Specifically, we find that matching on the industry median, as well as matching on industry and performance, generates results that are much smaller than those obtained by firm matching via the propensity score methods.
To trace the sources of these operating improvements, we examine changes in firms' income taxes and interest expenses from the prefiling year to the first postemergence year and find that, when compared to the matched firms, bankrupt firms' income taxes, scaled by total assets, increase slightly during bankruptcy while the ratio of interest expenses to assets is significantly reduced. We also find that bankrupt firms shed assets significantly and cut debt levels after emerging from bankruptcy. Thus, our results suggest that bankrupt firms undertake substantial organizational changes, including scaling down leverage significantly during bankruptcy, which help boost their net operating cash flow after emerging from bankruptcy. These results are shown to be robust to alternative matching estimates, different time periods, and various subsamples. To examine whether the improvements in cash flows differ among different types of firms, we conduct multivariate cluster analysis to determine natural groupings of bankrupt firms. Given the set of prefiling characteristics, we partition bankrupt firms into two nonoverlapping groups with the most distinct clustering and find that the benefits of reorganization are mainly enjoyed by firms with more liquid assets and higher market values. Taken together, our empirical results suggest that Chapter 11 reorganizations tend to boost the operating performances of firms that face temporary profitability problems, and they challenge the contention that Chapter 11 is an inefficient debtor-friendly mechanism.
The rest of the paper proceeds as follows. Section I discusses the use of matching methods in evaluating the effect of bankruptcy filings. Section II identifies the determinants of the bankruptcy filing decision used in our matching method. Section III describes the data, while Section IV presents the estimated effects of Chapter 11 on firm operating performance. Section V provides our conclusions.
I. Causal Inference about the Effect of Bankruptcy Filing
To assess the effect of Chapter 11 filing on firm operating performance, we need to identify a group of comparable firms and control for firm heterogeneity in prefiling characteristics. Ideally, the control firms would mimic the (counterfactual) performance of bankrupt firms if they had not filed for bankruptcy. This would isolate the pure effect of bankruptcy filing, so that the influence of firm heterogeneity and of other economic factors is eliminated from comparisons of "actual" outcomes with "would be" outcomes. The matching method is designed to minimize or eliminate any estimation bias associated with the inadequate control of firm attributes (Heckman, Ichimura, and Todd, 1998; Imbens, 2004).
We label a firm's filing decision a treatment and denote it by [Y.sub.i1] and [Y.sub.i0], the two possible performance outcomes for firm i if it files and does not file for bankruptcy, respectively. [D.sub.i] denotes the firm's bankruptcy status and is equal to one for filing firms and zero for all other firms. The parameter of interest is the average effect of Chapter 11 on the operating performance of the filing firm, E[[Y.sub.1] - [Y.sub.0]|D = 1]. This is commonly referred to as the average treatment effect on the treated (ATT) and captures the difference in the performance of the average filing firm relative to its performance if it had not filed. E[[Y.sub.1]|D = 1]. can be identified from data on filing firms, while E[[Y.sub.0]|D = 1]. is the counterfactual mean performance and cannot be identified from the data. Replacing E[[Y.sub.0]|D = 1]. with E[[Y.sub.0]|D = 0], for which data are available, leads to a selection bias in observational studies.
Matching is a way of estimating the evaluation parameter using observational data. It reduces the selection bias by constructing a group of comparable nonfiling firms, and is justified by the following two assumptions:
1. The conditional independence assumption:
([Y.sub.0], [Y.sub.1])D|X. (1)
2. The overlap assumption:
0 < Pr(D = 1|X) < 1. (2)
Given these two assumptions, the ATT is identified:
E[[Y.sub.1] - [Y.sub.0]|D = 1] = [E.sub.x] [E[[Y.sub.1] - [Y.sub.0]|X, D = 1]] = [E.sub.x] [E[[Y.sub.1]|X, D = 1] - E[[Y.sub.0]|X, D = 0]]. (3)
Assumption (1) indicates the independence of ([Y.sub.0], [Y,syb,1]) and D conditional on the observable set of firm attributes or covariates X. It asserts that the conditioning variables are sufficiently rich, so that the filing decision does not systematically depend on unobservables. Assumption (2) asserts that every firm has a chance of receiving the treatment, guaranteeing that matches can be found for all values of X. With the two assumptions, matching generates unbiased estimators. The matching method does not entirely eliminate the selection bias if certain unobservable factors systematically affect filing decisions even when conditioning on a rich set of observable firm attributes. For instance, one could argue that firms' filing decisions might be systematically dependent upon managers' ability to foresee the firms' prospects after bankruptcy. In this case, however, Heckman, Ichimura, and Todd (1997) show that matching still reduces the selection bias by eliminating two sources of the bias: 1) nonoverlapping support bias due to a comparison between noncomparable firms, and 2) misweighting bias due to a failure to reweigh the comparison firms. Any residual bias is due to selection on unobservables.
Matching serves to select a control group with comparable covariates, X, in effect reweighing the nonfiling firms in the region of common support. It explicitly controls for prefiling firm characteristics so that a comparison is made only between the filing firms and matched nonfiling firms with similar prefiling characteristics. The existing empirical studies often use industry-adjusted or industry and performance-adjusted measures to estimate the effect of bankruptcy on filing firms' operating performance. This implicitly assumes that the industry median and the prefiling performance are the only characteristics that need to be controlled for. Since there are many possible reasons why firms file for bankruptcy, partial matching based on one or two characteristics may not produce a proper control group of nonfiling firms.
To implement the matching method, a typical matching estimator takes the form:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (4)
where [I.sub.1] denotes the set of Chapter 11 filing firms, [I.sub.0] the set of all nonfiling firms, and [N.sub.1] is the number of firms in the set [I.sub.1]. The matches for each filing firm i [member of] [I.sub.1] are constructed as a weighted average over the outcomes of comparison firms where W(i, j) is a weight with [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] and depends upon the distance between [X.sub.i] and [X.sub.j].
In practice, matching may be difficult to implement when the set of variables in X is large. Rosenbaum and Rubin (1983) derive an important result that reduces a high dimensional matching problem to a one dimensional problem. They show that under the foregoing set of assumptions (i.e., conditional independence and overlap), matching can also be based on the propensity score, P(X) = P(D = 1|X), which is the probability of filing for Chapter 11 conditional on the covariates of X. We use the propensity score to select comparable non-filing firms and present our main estimation results based on propensity score matching in Section IV.
II. Bankruptcy Filing Determinants and Empirical Measures
In this section, we identify a set of firm characteristics that affect a firm's propensity to file for bankruptcy. Empirical proxy measures are constructed using data in Compustat, the source for all accounting information. These measures are used in Section IV to estimate firms' propensity scores on the basis of which the group of matched nonfiling firms is selected.
A firm's bankruptcy decision can be decomposed into two steps, whether to default and, if so, how to resolve the problem of default. The firm faces the option of either continuing its current operations (e.g., it may attempt to raise external funds to pay off current debt) or defaulting on its debt obligations. If it defaults, the firm can either reorganize through a private workout or a formal bankruptcy filing, or it can liquidate its assets. The equity holders in control of the firm make this decision based on equity value in three alternative situations: 1) continuation ([E.sub.c]), 2) reorganization ([E.sub.r]), and 3) liquidation ([E.sub.l]).
The equity value [E.sub.c] is affected by the firm's liquidity. If the firm lacks sufficient cash, or slack, to cover its current debt obligations and the costs of operations, it needs to access the external capital market to raise funds. Given transaction costs and asymmetric information, the marginal cost of external financing will exceed that of internal financing and the equity value would be reduced (Myers and Majluf, 1984). The equity values [E.sub.r] and [E.sub.l] depend on the possibility of renegotiation between shareholders and debt holders. Renegotiations can take the form of a private workout or can occur through a formal bankruptcy filing depending on the transaction costs of bargaining. The more severe the bargaining problems, the greater the likelihood of the firm to resort to filing Chapter 11. These considerations suggest a set of firm characteristics that affect the propensity to file for bankruptcy. We turn now to a discussion of these characteristics and summarize their empirical proxy measures in Table I. In the subsequent estimations, all proxy variables, except for the market value of equity, are measured at the end of the first fiscal year prior to the filing date.
A. Market Value of Equity
The larger the equity value in continuation, [E.sub.c], the less likely is the firm to file for bankruptcy. We calculate the market value of equity as the product of common shares outstanding and the close price at the fiscal year end. We then divide the market value of equity by the book value of total liabilities. In the sample we describe in Section III, some firms do not have trading data during the first fiscal year prior to the filing. To minimize any loss of bankrupt firms from our sample, we averaged the ratio of equity value to total liabilities over the three years preceding the bankruptcy filing. In subsequent estimations, the measure MV of equity/total liabilities represents a three-year average prior to filing. (2)
With asymmetric information, outside financing tends to be costly and reduces equity value. The more financial slack a firm has, the less prone it is to seek outside financing attenuating the reduction in equity value and the risk of bankruptcy. Two empirical measures are used as proxies for the firm's liquidity level. The first is the quick ratio, defined as cash plus receivables divided by current liabilities. The second is the interest coverage ratio, which is measured as earnings before interest and taxes (EBIT) over current interest expenses and captures the firm's ability to generate sufficient revenues to meet interest expenses.
We follow the literature by using profitability measures to assess the effect of Chapter 11 on firm operating performance. We also control for profitability in the prefiling firm attributes since lower profitability could lead to deeper financial difficulty. In our empirical work, we use either one of the two empirical measures for profitability: 1) operating income/total assets and 2) cash flow/total assets. Operating income is earnings before interest, taxes, depreciation, and amortization (EBITDA), and the operating cash flow measure is obtained directly from the statement of cash flows in the Compustat database.
D. Liquidation Value
Gilson, John, and Lang (1990) argue that assets are more likely to be sold when debt is restructured under Chapter 11 than privately. Chapter 11 is more costly for firms with a higher proportion of intangible assets. Thus, such firms are less likely to file for Chapter 11. We use intangible assets/total assets as one proxy for liquidation value. Alderson and Betker (1996) find that the liquidation costs of their sample firms have a significantly negative correlation to the fixed-to-total assets ratio. Also, the firm's fixed assets could serve as collateral enabling it to borrow to overcome default. Therefore, the importance of fixed assets in the choice of bankruptcy filing remains an empirical issue. We measure the proportion of fixed assets by the ratio of fixed assets/total assets, where fixed assets include the firm's net value of property, plant, and equipment.
E. Measures Reflecting Bargaining and Recontracting Problems
A firm's choice between a private workout and formal bankruptcy filing depends upon the severity of bargaining problems. Bargaining and recontracting problems make it less likely for a firm to achieve a successful private workout. Holdout and asymmetric information problems can be particularly serious sources of bargaining difficulties. The severity of the bargaining problems is affected by the heterogeneity of the debt claims as greater heterogeneity implies greater difficulties in having the debt restructured. Two categories of debt claims are examined, trade credits/total assets and secured debt/total liabilities, where the former is defined as accounts payable normalized by total assets and the latter as all secured long-term debt divided by total liabilities. As discussed in Gilson, John, and Lang (1990), the holdout problem is particularly severe for trade credit debt as the number of trade creditors is often large and their claims relatively heterogeneous. Achieving consensus via bargaining among trade creditors tends to be difficult. Alternatively, secured debt tends to be more homogenous as it is usually held by a relatively small number of institutional investors (e.g., banks and insurance companies). Moreover, institutional investors often have greater incentive and expertise to monitor a firm's operations. Thus, a firm with more secured debt tends to have smaller problems associated with heterogeneous debt and asymmetric information, and is more likely to rely on a private workout rather than a formal bankruptcy when reorganizing. Gilson, John, and Lang (1990) find that the presence of bank debt has a positive effect on the success of a workout. However, secured creditors could have a stronger bargaining position in renegotiations and, in this regard, are less likely to tender their claims in a private workout since their claims are secured by the pledge. Chatterjee, Dhillon, and Ramirez (1996) find that firms with more bank debt tend to choose Chapter 11 filings over other alternatives. The foregoing analysis suggests that the impact of secured debt on the firm's reorganization choice remains an empirical issue.
Finally, we use the auditor's opinion as an empirical proxy for asymmetric information problems. There are six categories for the auditor's opinion item in the Compustat database, and they are assigned a score of one to four (the higher the score, the better is the quality of the disclosure of information) as follows:
1. Either an auditor has expressed a negative opinion on the financial statements of the company or the financial statements reflect the effects of some limitation on the scope of the examination or unsatisfactory presentation of financial information.
2. Either financial statements are unaudited or the auditor refuses to express an opinion regarding the company's ability to sustain its operation as a going concern.
3. The auditor has expressed an unqualified opinion regarding the financial statements, but has added explanatory language to the auditor's standard report.
4. The financial statements reflect no unresolvable restrictions and the auditor makes no significant exceptions as to the accounting principles, the consistency of their application, and the adequacy of the information disclosed.
All of the aforementioned firm characteristics could affect the firm's filing decision and, as such, need to be properly controlled. The extant literature does not fully account for firm-level heterogeneity and may not have a proper benchmark of firms for comparison. As we demonstrate below in Table III, firms only matched by industry median or by industry and performance are significantly different from the bankrupt firms. We use the matching method to select control firms comparable to bankrupt firms in prefiling characteristics.
III. Sample and Summary Statistics
The Chapter 11 filing firms are drawn from the Bankruptcy Research Database (BRD), which contains Chapter 11 bankruptcy cases for large public firms with a real value of assets at the time of filing more than $100 million in 1980 dollars. (3) We start with all firms in the BRD that filed for Chapter 11 from 1987 to 2008, but delete those that are not in the Compustat database. We further exclude financial firms with historical SIC codes between 6000 and 6999. Figure 1 plots the distribution of the bankrupt nonfinancial firms in our initial sample.
We retrieve accounting data from Compustat and define annual observations on the basis of fiscal years as firms use a variety of fiscal year ends. Thus, we convert the dates in which the sample firms filed for Chapter 11 into corresponding fiscal years and label them as filing years. Whenever applicable, we also translate the dates in which the sample firms had their reorganization plan confirmed to corresponding fiscal years and refer to them as confirmation years. We eliminate observations preceding the filing years for which there is incomplete information on filing determinants and remove observations following the confirmation years where there is no information regarding operating income or cash flow. We also delete a bankrupt firm if its most recent year with available data is three or more years earlier than the filing year. The imposition of these requirements results in 464 bankrupt firms. In the refined sample, the prefiling year of a firm is the first fiscal year prior to its filing year and ranges from 1985 to 2006. Where applicable, postemergence years of a bankrupt firm refers to one to three fiscal years, depending upon the analysis, following the firm's confirmation year. Table II presents information on postbankruptcy outcomes for the 464 filing firms. Among them, 164 firms did not emerge from Chapter 11, while 285 did. By the definition employed in the BRD, a firm emerges from Chapter 11 if it is a stand-alone company and continues to operate after confirmation of the plan of reorganization. For firms that did not emerge, most were acquired or liquidated. Of the emerged firms, 127 were no longer tracked by Compustat. We examined the reasons for their deletions by reading footnote information in Compustat and found that 21 firms were deleted due to mergers and acquisitions, 27 were deleted due to bankruptcies, 20 went private after reorganizations, 42 no longer filed information with the Securities and Exchange Commission (SEC), and 17 were deleted for undocumented reasons. Our assessment of the effect of Chapter 11 focuses on the 158 firms that eventually emerged as independent companies and had accounting data in Compustat. It is this group, and particularly their postemergence operating performance, that raises concerns about whether Chapter 11 induces the survival of too many economically nonviable firms.
[FIGURE 1 OMITTED]
To find proper benchmarks for the Chapter 11 filing firms, we retrieve from Compustat all nonfinancial firms that are neither recorded in the BRD nor deleted from Compustat due to bankruptcies. We further focus on large firms and require that a nonbankrupt firm have assets worth more than $100 million, measured in 1980 dollars, for more than 80% of the time it is in the sample period. After removing observations lacking the necessary data information, we obtain a comparison group of 28,454 firm-years over the same range of prefiling years for the bankruptcy filing group. The first two panels of Table III summarize prefiling characteristics for the bankruptcy filing firms and comparison firms. On average, the bankruptcy filing firms have lower profitability, less liquidity, and smaller equity market value. The bankrupt firms have similar amounts of fixed assets, but higher intangible assets than nonbankrupt firms. Bankruptcy filing firms also have a higher proportion of secured debt. Similar to Chatterjee, Dhillon, and Ramirez (1996), we find that Chapter 11 filing firms have a higher trade credit to asset ratio suggesting that firms with more severe bargaining problems resort to formal bankruptcy filings. Finally, bankrupt firms receive less favorable opinions from auditors on their reporting standards.
The first two panels of Table III report that firm characteristics differ substantially between comparison firms and bankrupt firms. In the rest of Table III, we investigate whether the usual industry or industry and performance adjustments employed in the extant literature produce a group of firms comparable to bankrupt firms in terms of their prefiling characteristics.
In Panel C, we select, for each prefiling characteristic, median firms from the group of non-bankrupt firm-years that are in the same SIC industries as the bankrupt firms in their prefiling years. We then compute the industry median characteristics at the most disaggregated SIC level for which there are at least five nonbankrupt firms in the industry-year. We carry out a t-test (sign test) for matched pairs to examine the equality of means (medians) between bankrupt firms and their industry medians. The results indicate that for nine out of the ten prefiling characteristics, both means and medians are statistically different between the two groups.
In Panels D1 and D2, we compare the prefiling characteristics of bankrupt firms with those of industry and performance matched firms. A matched firm is one that is in the same major SIC industry as a bankrupt firm and has an operating performance closest to the bankrupt firm's in the prefiling year. To measure operating performance, Panel D1 uses operating income, while Panel D2 uses cash flow. We also conduct tests for the equality of means and medians between bankrupt firms and the industry and performance matched firms. The results show that bankrupt firms and their industry and performance matched firms have statistically different prefiling characteristics in most dimensions. In summary, Table III suggests that the industry median firms and the industry and performance matched firms differ significantly from bankrupt firms in terms of prefiling characteristics. In the next section, we employ matching methods to find a control group of nonbankrupt firms comparable to the bankrupt firms in a wide range of firm characteristics.
IV. Matching Estimates
A. Summary for Propensity Score Matched Firms
The main estimation results are obtained by matching firms on the basis of propensity scores. To estimate propensity scores, a logit model is used to predict firms' propensity to file for bankruptcy. The dependent variable is equal to one in the prefiling years for bankrupt firms and zero for the nonbankrupt firm-years that are in the range of bankrupt firms' prefiling years. The explanatory variables include the filing determinants, measured at the bankrupt firms' prefiling years. Summary statistics in the previous section indicate that bankruptcy filing firms have more intangible assets. To account for the possibility that the impact of intangible assets on filing decisions might be nonlinear, we also include its squared terms in the regression. Finally, year dummies are also included to control for macroeconomic effects.
The estimation results of the logit model are presented in Table IV. Panel A uses EBITDA/total assets as the profitability measure, while Panel B uses operating cash flow/total assets in the regression. The regression estimates are qualitatively the same for both operating performance measures, and most variables are significant and have the expected signs. Firms with lower profitability, less liquidity, and smaller market value of equity are more likely to file for bankruptcy. The proportion of secured debt in a firm's total liability has a positive effect on its filing decision, suggesting that secured creditors have a tougher bargaining position in renegotiations. The coefficients on the intangible assets measure imply that firms with more intangible assets are less likely to file only when the amount of intangible assets reaches a very high level. Finally, firms with a better quality of information disclosure are less likely to choose bankruptcy filings. (4)
We estimate propensity scores, predicted probabilities from the logit model, to find matched firms from the comparison group of nonbankrupt firms. The selection criterion is based on the difference between the propensity scores of the comparison firms and the bankrupt firms. First, all nonbankrupt firm-years are discarded if their estimated propensity scores are smaller (larger) than the minimum (maximum) of the estimated propensity scores for the bankrupt firms. This serves to impose a common support region and eliminates those not comparable to bankrupt firms. Next, a bankrupt firm is matched with k control firms whose estimated propensity scores in the bankrupt firm's prefiling year are closest to the bankrupt firm's propensity score. The k-nearest neighbors should be comparable to the bankrupt firms if the propensity score matching method selects proper benchmark firms.
Table V summarizes the characteristics of the matched firms with the propensity scores nearest those of the bankrupt firms. The results in the upper panel are obtained using propensity scores that are estimated by the logit regression in Panel A of Table IV. The results in the lower panel involve estimated propensity scores from the logit regression in Panel B of Table IV. When compared to the results in Table III, propensity score matched firms exhibit closer similarity to the bankrupt firms than do the industry-median or industry and performance matched firms. Further, the average characteristics of propensity score matched firms are now not significantly different from those of bankrupt firms in most dimensions. (5) These findings suggest that the propensity score matching method selected proper controls that are comparable to bankrupt firms in a wide range of prefiling characteristics.
B. The Effect of Chapter 11 Filing on Operating Performance
In this subsection, we employ matching methods to estimate the causal effect of Chapter 11 filings on firm operating performance. The parameter of interest, the ATT, captures the mean difference between the observed operating performance of reorganized firms and their counterfactual outcomes (i.e., how would they have performed had they not filed for Chapter 11?). To gauge a firm's operating performance, we use either the operating income/total assets measure or the cash flow/total assets measure. When operating income (or cash flow) is used as a performance measure, it is also included in the logit model to estimate a propensity score and to find matched firms.
Table VI presents estimation results of the k-nearest neighbor matching for k = 1, 4, 6, 10. (6) Panel A examines the short-term effect measured by the change in operating performance from the prefiling year to the first year after the firm emerges from Chapter 11. To find the counterfactual performance of a bankrupt firm, we obtain k nearest neighbor firms matched by propensity scores and calculate their average change in performance over the same observation period for the bankrupt firm. A k-nearest neighbor matching estimate is then computed as the mean difference between the bankrupt firms' observed and counterfactual performance changes. To assess the statistical significance of the estimate, we calculate the t-statistic for the differenced outcomes of the matched pairs and compare the t-statistic with the rejection region from the bootstrap sampling distribution based on 1,000 replications.
The results in Part 1, Panel A show that the observed average change in operating income for bankrupt firms is 0.041 and the average change for matched firms varies from 0.033 to 0.044. Therefore, the estimated ATT ranges from -0.003 to 0.008. These estimates indicate that the performance changes of bankrupt firms and of their matched peers are not significantly different. When we examine the change in cash flow from the prefiling year to the first postemergence year, we see a significant increase in the reorganized firms' cash flows, even in comparison to the matched benchmark firms. The observed average change in cash flow for bankrupt firms is 0.135 and the average change for matched firms ranges from 0.066 to 0.104, giving an estimate of the ATT from 0.031 to 0.069. These estimates are statistically significant in three out of four cases and account for 23.0% to 51.1% of the observed average change in cash flow for bankrupt firms. They are also economically significant. In summary, the results show that in the short term, reorganized firms do not fare worse in terms of operating income and their cash flows improve significantly during Chapter 11.
We further investigate (Panel B of Table VI) how reorganized firms fare in the long term. We follow a reorganized firm for three years after it emerges from Chapter 11. We calculate the average operating performance over the three postemergence years. In the process, we lose some observations as some firms emerged after 2006. As such, we do not have three years of observations prior to 2009, the last fiscal year of our sample. Additionally, some of them went private or were acquired after the first postemergence year. Focusing on firms with complete information, we obtain the change in operating performance by subtracting the prefiling performance from the average performance over the three years after emergence from bankruptcy. The results in Part 1 Panel B show that the change in operating income is stronger for bankrupt firms than for the benchmark firms, but it is not significant for half of the estimates. As for the cash flow measure, bankrupt firms enjoy a significantly greater increase over the three years after emergence than do the matched benchmark firms. The estimated ATT for the long-term change in cash flow ranges from 0.074 to 0.122, accounting for 45.4% to 74.8% of observed changes in cash flow for the bankrupt firms. These estimates are all statistically significant and much larger than the short-term effects. Overall, the results in Table VI suggest that reorganized firms' operating cash flows improved during Chapter 11 bankruptcy and the improvement was more pronounced in the long run.
To assess the effect of Chapter 11 filing on firm operating performance, we employed empirical matching methods to control for firm heterogeneity in the prefiling characteristics. The methods explicitly control for prefiling firm characteristics, so that a comparison is made strictly between filing firms and matched nonfiling firms with similar prefiling characteristics. This procedure helps separate the pure effect of bankruptcy filing from the influence of firm heterogeneity and other confounding economic factors. The firm matching estimates should produce, in principle, more robust and less biased estimates regarding the impact of Chapter 11 filing. We have demonstrated that firms matched using the propensity score method have very similar prefiling characteristics to the bankrupt firms, while the industry median firms and those matched according to industry and performance matching differ significantly from bankrupt firms in most dimensions. In Table VI, we compare the propensity score matching results pertaining to the effect of Chapter 11 on firm operating performance with those obtained by adjusting only for the industry effect. In both Panels A and B, Part 2 reports estimates using industry median matching as a benchmark, while Part 3 reports results using industry and performance matching. We use a t-test for paired data to examine the equality of means between the two groups. In comparing the changes in cash flows for bankrupt firms with those for matched industry median firms, we obtain an estimate of 0.10 in the short run and 0.027 in the long run. When we compare the bankrupt firms to their industry peers with the closest cash-flow-to-assets ratios prior to the filing, the estimated differences in the effects of Chapter 11 between reorganized and control firms are 0.017 in the short run and 0.029 in the long run. The industry benchmarking results are much smaller than those obtained using propensity score matching. For instance, the estimated long-term effect for the industry-and-cash flow matched result is 0.029, while the same effect for propensity score matching ranges from 0.074 to 0.122. Thus, when compared to the industry and performance matched results, propensity score matching generates estimates of increases in the ratios of cash flows to total assets that are 155% to 320% larger.
C. Improved Operating Performance and Organizational Changes
Table VI demonstrates that reorganized firms performed no worse and, if anything, better than comparable nonfiling firms. These improvements are derived largely from firms' operating cash flows. Alderson and Betker (1999) also find that although firms' postbankruptcy operating margins are poor, the total cash flow generates comparable returns to those in benchmark portfolios. To trace the sources of the operating improvements, we further investigate firms' operational and structural changes during bankruptcy.
Table VII summarizes the mean and median changes for reorganized firms and their nearest neighbor matches. The results are similar for reorganized firms and their k-nearest neighbors, where k = 4, 6, 10. The mean and median changes in income taxes for bankrupt and matched firms are not statistically different. The ratio of income taxes to total assets, on average, increases during bankruptcy for the reorganized firms, while there is virtually no change for the matched firms. Reorganized firms incur much lower interest expenses after the emergence of bankruptcy. The mean (median) reduction in interest payments is $34.095 ($18.451) million for reorganized firms, which is much larger than the reduction for the matched firms. After adjusting for total assets, the mean (median) reduction in the ratio of interest expenses is 0.029 (0.027) for the reorganized firms. This is significantly larger than the mean (median) reduction for the matched firms, which is only 0.009 (0.004). We also compared changes in working capital between the two groups and found that working capital investment, normalized by total assets, falls by 0.006 during bankruptcy for the average reorganized firm and it drops by 0.01 for the average matched firm. The difference in the decline in working capital investment between the two groups is not statistically significant. The preceding results suggest that reorganized firms reduce interest expenses during bankruptcy, thereby improving their net cash flow.
Previous studies find that bankrupt firms significantly reduce their assets and liabilities during Chapter 11 (Denis and Rodgers, 2007; Heron, Lie, and Rodgers, 2009). We examine changes in the firms' total assets and debt levels and find that firms shed assets and cut liabilities after they emerge from bankruptcy. The mean (median) change in total assets is -$441.985 (-$142.496) million for the reorganized firms, while the mean (median) change for comparable nonbankrupt firms is $4.022 (-$34.677) million. On the liability side, the sum of long-term and short-term debt decreases by $287.421 ($126.181) million for the mean (median) reorganized firm, while the mean (median) reduction for the matched nonbankrupt firm is $24.565 ($12.087) million. Normalizing firms' debt by total assets, we see that the mean (median) change in total debt/total assets is -0.202 (-0.178) for reorganized firms, while the change for matched firms is 0.021 (0.007). The differences between the two groups are both economically and statistically significant. Taken together, the results in Tables VI and VII suggest that when compared to otherwise similar nonbankrupt firms, reorganized firms significantly reduce their leverage and that bankruptcy forces firms to undertake organizational changes that help boost their net operating cash flows after emerging from bankruptcy.
D. The Effects of Bankruptcy and Prefiling Firm Characteristics
We have demonstrated that Chapter 11 filings help to improve reorganized firms' operating cash flows. A natural next step is to examine whether the benefits differ among different types of firms. To do this, we need to determine natural groupings of bankrupt firms. Given the set of prefiling characteristics, we use multivariate cluster analysis to partition bankrupt firms into k distinct nonoverlapping groups and investigate whether the effects of bankruptcy vary among these groups. Specifically, we start with k nearly equal segments formed from the bankrupt firms. The medians from these k groups are to be used as the starting group centers. Each bankrupt firm is then assigned to the group whose median is closest in Euclidean distance. Based on that classification, new group medians are computed, and the process is repeated. This procedure continues until all bankrupt firms remain in the same group from the previous iteration. Using the Calinski and Harabasz (1974) stopping rule, we find that k = 2 produces the most distinct clustering. Therefore, we compare the estimated effects and prefiling firm attributes between the two groups and present the results in Table VIII.
Panel A reports the short-term changes in operating cash flow for 69 firms in Group 1 and 89 firms in Group 2. For all matching estimates, the average change in cash flow for the first group is positive and much larger than the average change for the second group. Therefore, Chapter 11 reorganization appears to benefit Group 1 firms rather than Group 2 firms. A further analysis of prefiling characteristics suggests that a typical firm in Group 1 is less profitable and has lower earnings to cover their interest expenses. However, the representative firm in Group 1 has more liquid assets (as captured by its higher quick ratio) and a higher market value of equity.
Panel B of Table VIII presents the long-term changes in operating cash flows for the two distinct clusters. A comparison between the two groups leads to similar conclusions as previously presented. Group 1 firms, on average, enjoy greater cash flow improvements than their counterparts in Group 2. The typical firm in Group 1 is less profitable before filing and has more liquid assets. In summary, the results in Table VIII suggest that firms benefit more from bankruptcy reorganizations if they face temporary profitability problems and have larger liquidity.
E. Discussion of the Results
We used a sample of Chapter 11 cases filed from 1987 to 2008 to demonstrate that reorganized firms had significantly larger increases in cash flow when compared to otherwise identical nonfiling firms. In recent years (since 2003), there have been regulatory and practical changes in the US bankruptcy law, leading some financial economists and legal scholars to argue that Chapter 11 has shifted from a debtor-friendly to a creditor-friendly regime. (7) This alleged regime change raises the following question: "Are our foregoing results driven mainly by the fact that only the best performing firms have been allowed in recent years to emerge from Chapter 11?"
To answer this question, we examined the emergence rate and duration in bankruptcy for our sample of bankrupt firms that filed under Chapter 11 prior to 2003. In our pre-2003 sample of firms, 61% emerged from Chapter 11 and the emergence rate for the whole sample was 63%. The average (median) time spent in bankruptcy was 16.4 (13) months for the pre-2003 sample and 15.6 (12) months for the entire sample. We also investigated the change in the operating performance of 144 reorganized firms in the pre-2003 period. Our unreported propensity score matching results indicate that changes in cash flows for the pre-2003 sample are similar to those obtained for the entire sample. Thus, the recent regulatory changes do not seem to have had a dramatic impact on the bankruptcy outcomes of our sample firms. Our results do not indicate that only the strong firms survived Chapter 11 bankruptcy in the post-2003 period.
Our empirical results shed some light on the efficiency of the Chapter 11 bankruptcy process. The intent of Chapter 11 is to help rehabilitate economically viable firms facing temporary difficulties. The magnitude of a type II error wherein nonviable firms are reorganized rather than liquidated drives much of the controversy over the efficiency of Chapter 11. Our findings suggest that the type II error is not substantial and, that contrary to some claims, Chapter 11 is not a debtor-friendly process. A related efficiency issue pertaining to Chapter 11 concerns the type I error (i.e., that economically viable firms are wrongly liquidated during bankruptcy). (8) We did not address the question of type l errors as data limitations do not allow us to keep track of a firm's asset liquidation, nor do the data enable us to find the liquidation value of the firm. Yet, there is some evidence in the literature to suggest that the type I error is not substantial. Using plant-level data, Maksimovic and Phillips (1998) find that plants have higher Total Factor Productivity when sold to new owners post Chapter 11. Kalay, Singhal, and Tashjian (2007) utilize data on prebankruptcy market value and postbankruptcy liquidation value for 37 firms liquidated or acquired during Chapter 11. They find that value based and risk-adjusted returns for the liquidated or acquired firms do not statistically differ from those of reorganized firms.
F. Robustness Checks
In this section, we first investigate whether the main results differ in subsamples of bankrupt firms that have different durations in bankruptcy. For the 158 reorganized firms in Table VI, the mean (median) time spent in Chapter 11 is 15.2 (11.5) months. We divide the bankrupt firms into two groups: 1) firms that stay in bankruptcy for less than one year and 2) firms in bankruptcy for more than one year. Propensity score matching estimation is carried out for each group separately. The untabulated results are consistent with our general conclusion that operating income for the average reorganized firm does not increase significantly during bankruptcy, but the improvement is significant for cash flow. Comparing the increased cash flow measure between the two subsamples, we find that improvement is greater for firms staying in bankruptcy for less than one year.
The previous results are valid to the extent that the estimated propensity score correctly captures a firm's probability to file. As a final robustness check, we use an alternative method to select matched firms. The selection procedure is directly based on the set, X, of prefiling characteristics and chooses a control firm whose X is closest to that of the bankrupt firm. To determine the distance between a bankrupt firm's prefiling characteristics, [X.sub.i], and a control firm's, [X.sub.j], we use the following metric:
[parallel][X.sub.i] - [X.sub.j][parallel] = ([X.sub.i] - [X.sub.j])'. [[summation].sup.-1.sub.xi] x ([X.sub.i] - [X.sub.j]), (5)
where [[summation].sup.-1.sub.xi] is the covariance matrix of prefiling characteristics for the group of bankrupt firms, and it helps to reduce differences in X within matched pairs in all directions. The unreported results of the alternative matching method are similar to the main results in Table VI. In short, cash flows improve significantly after firms emerge from bankruptcy, and this improvement is more pronounced in the long term.
We evaluated the effect of Chapter 11 filings on the operating performance of bankrupt firms. Critics of Chapter 11 have suggested that it is an overly debtor-friendly process that fails to liquidate a significant number of inefficient firms, impeding collectively rational outcomes for bankrupt firms. The results of this paper challenge this claim. To assess the effect of Chapter 11 filings, one needs to identify a group of comparable firms that mimic the (counterfactual) performance of bankrupt firms. To find proper benchmark firms, it is important to control for firm-level heterogeneity in prefiling characteristics. We demonstrated that the usual industry-median or industry and performance adjustments do not necessarily yield a comparable set of firms to the bankrupt firms in terms of key prefiling characteristics. We used matching methods to select control firms that constitute a proper benchmark group to the filing firms. We then compared changes in the operating performance of bankruptcy filing firms with those of matched nonbankrupt firms. We found that when compared to the matched group, the operating income of the filing firms did not change significantly during bankruptcy, while their net cash flows improved significantly. These results were robust to alternative matching estimations, different time periods, and various subsamples. We also investigated changes in firms' tax obligations and interest expenses over the period stretching from the prefiling year to the first postemergence year. We determined that when compared to matched firms, bankrupt firms' income taxes, normalized by assets, increased slightly during bankruptcy, while the ratio of interest expenses to assets was significantly reduced. Also, bankrupt firms significantly lowered their debt-to-assets ratios after emerging from bankruptcy.
Our findings suggest that Chapter 11 forces bankrupt firms to scale down their leverage, and that the ensuing organizational changes help to boost firms' operating cash flow after they emerge from bankruptcy. We find further evidence that Chapter 11 reorganizations are more beneficial to firms with more liquid assets and higher market value prior to filing for bankruptcy. Our results challenge the contention that Chapter 11 is an inefficient debtor-friendly mechanism that rehabilitates economically nonviable firms. We demonstrate that firms that file under Chapter 11 perform no worse and, if anything, better than comparable nonfiling firms.
Aghion, P., O. Hart, and J. Moore, 1992, "The Economics of Bankruptcy Reform," Journal of Law, Economics, and Organization 8, 523-546.
Aivazian, V.A. and J.L. Callen, 1980, "Corporate Leverage and Growth: The Game-Theoretical Issues," Journal of Financial Economics 8,379-399.
Aivazian, V.A. and J.L. Callen, 1983, "Reorganization in Bankruptcy and the issue of Strategic Risk," Journal of Banking and Finance 7, 119-133.
Alderson, M.J. and B.L. Betker, 1996, "Liquidation Costs and Accounting Data," Financial Management 25, 25-36.
Alderson, M.J. and B.L. Betker, 1999, "Assessing Post-Bankruptcy Performance: An Analysis of Reorganized Firms' Cash Flows," Financial Management 28, 68-82.
Andrade, G. and S.N. Kaplan, 1998, "How Costly Is Financial (Not Economic) Distress? Evidence from Highly Leveraged Transactions that Became Distressed," Journal of Finance 53, 1443-1493.
Baird, D.G., 1986, "The Uneasy Case for Corporate Reorganizations," Journal of Legal Studies 15,127-147.
Baird, D.G. and R.K. Rasmussen, 2002, "The End of Bankruptcy," Stanford Law Review 55, 751-790.
Baird, D.G. and R.K. Rasmussen, 2003, "Chapter 11 at Twilight," Stanford Law Review 56, 673-700.
Barber, B.M. and J.D. Lyon, 1996, "Detecting Abnormal Operating Performance: The Empirical Power and Specification of Test Statistics," Journal of Financial Economics 41,359-399.
Bebchuk, L.A., 1988, "A New Approach to Corporate Reorganizations," Harvard Law Review 101,775-804.
Bebchuk, L.A., 2000, "Using Options to Divide Value in Corporate Bankruptcy," European Economic Review 44, 829-843.
Bradley, M. and M. Rosenzweig, 1991, "The Untenable Case for Chapter 11," Yale Law Journal 101, 1043-1095.
Bris, A., I. Welch, and N. Zhu, 2006, "The Costs of Bankruptcy: Chapter 7 Liquidation versus Chapter 11 Reorganization," Journal of Finance 61, 1253-1303.
Brown, D.T., 1989, "Claimholders Incentive Conflicts in Reorganization: The Role of Bankruptcy Law," Review of Financial Studies 2, 109-123.
Calinski, T. and J. Harabasz, 1974, "A Dendrite Method for Cluster Analysis," Communications in Statistics-Theory and Methods 3, 1-27.
Chatterjee, S., U.S. Dhillon, and G.G. Ramirez, 1996, "Resolution of Financial Distress: Debt Restructurings via Chapter 11, Prepackaged Bankruptcies, and Workout," Financial Management 25, 5-18.
Denis, D.K. and K.J. Rodgers, 2007, "Chapter 11: Duration, Outcome, and Post-Reorganization Performance," Journal of Financial and Quantitative Analysis 42, 101-118.
Giammarino, R.M., 1989, "The Resolution of Financial Distress," Review of Financial Studies 2, 25-47.
Gilson, S.C., 1997, "Transactions Costs and Capital Structure Choice: Evidence from Financially Distressed Firms," Journal of Finance 52, 161-196.
Gilson, S.C., K. John, and L.H. Lang, 1990, "Troubled Debt Restructurings: An Empirical Study of Private Reorganization of Firms in Default," Journal of Financial Economics 27, 315-353.
Heckman, J.J., H. Ichimura, and P.E. Todd, 1997, "Matching As an Econometric Evaluation Estimator: Evidence From Evaluating a Job Training Programme," Review of Economic Studies 64, 605-654.
Heckman, J.J., H. Ichimura, and P.E. Todd, 1998, "Matching As an Econometric Evaluation Estimator," Review of Economic Studies 65, 261-294.
Heron, R., E. Lie, and K. Rodgers, 2009, "Financial Restructuring in Fresh-Start Chapter 11 Reorganizations," Financial Management 38, 727-745.
Hotchkiss, E.S., 1995, "Post-bankruptcy Performance and Management Turnover," Journal of Finance 50, 3-21.
Imbens, G.W., 2004, "Nonparametric Estimation of Average Treatment Effects Under Exogeneity: A Review," Review of Economics and Statistics 86, 4-29.
Jensen, M.C., 1991, "Corporate Control and the Politics of Finance," Journal of Applied Corporate Finance 4, 13-33.
Kalay, A., R. Singhal, and E. Tashjian, 2007, "Is Chapter 11 Costly?" Journal of Financial Economics 84, 772-796.
Maksimovic, V. and G. Phillips, 1998, "Asset Efficiency and Reallocation Decisions of Bankrupt Firms," Journal of Finance 53, 1495-1532.
Myers, S.C. and N. Majluf, 1984, "Corporate Financing and Investment Decisions When Firms Have Information that Investors Do Not Have," Journal of Finance 13, 187-221.
Rosenbaum, P.R. and D.B. Rubin, 1983, "The Central Role of the Propensity Score in Observational Studies for Causal Effects," Biometrika 70, 41-55.
Warren, E. and J.L. Westbrook, 2003, "Secured Party in Possession," American Bankruptcy Institute Journal 22 12, 52-53.
White, M.J., 1989, "The Corporate Bankruptcy Decision," Journal of Economic Perspectives 3, 129-151.
White, M.J., 1994, "Corporate Bankruptcy as a Filtering Device: Chapter 11 Reorganizations and Out-of-Court Debt Restructurings," Journal of Law, Economics, and Organization 10, 268-295.
White, M.J., 2007, "Bankruptcy Reform and Credit Cards," Journal of Economic Perspectives 21, 175-199.
White, M.J., 2009, "Bankruptcy: Past Puzzles, Recent Reforms, and the Mortgage Crisis," American Law and Economics Review 11, 1-23.
Wruck, K.H., 1990, "Financial Distress, Reorganization, and Organizational Efficiency," Journal of Financial Economics 27, 419-444.
Zhang, G., 2010, "Emerging From Chapter 11 Bankruptcy: Is It Good News or Bad News for Industry Competitors?" Financial Management 39, 1719-1742.
(1) An alternative approach is to assume bivariate normality of the error terms in treatment-effects regressions and apply Heckman's two-step procedure. Bris, Welch, and Zhu (2006) take this approach to examine a variety of measures associated with bankruptcy costs including the change in the firm's value during bankruptcy, the time spent in bankruptcy, legal expenses, and recovery rates for creditors.
(2) In our sample, the equity-to-liabilities ratio measured at the first prefiling year has a high correlation (0.89) with the three-year average ratio. Our empirical results remain qualitatively the same when we use the ratio of the first prefiling year.
(3) The Bankruptcy Research Database is maintained by Professor Lynn LoPucki. More information about the database is available at http://lopucki.law.ucla.edu/.
(4) Our main estimation results of bankruptcy effects are valid to the extent that the specification of our logit regression is correct and the estimated propensity score correctly captures a firm's probability to file. As a robustness check, we carry out an alternative matching method in Section E which does not rely on estimating propensity scores. It is reassuring to note that the results are similar to those obtained by propensity score matching.
(5) When bankrupt firms are matched with their k-nearest neighbors, for k = 4, 6, 10, the unreported results are similar and indicate that an average firm in the matched group is not statistically different from an average bankrupt firm in terms of most pre-filing characteristics.
(6) The nearest neighbor matching pairs the bankrupt firm with k matches from the universe of nonbankrupt firms. In effect, it assigns equal weights to the k matches and zero weight to the other nonbankrupt firms. In the empirical estimation, we also conducted an estimation using kernel-based matching (Heckman, Ichimura, and Todd, 1997, 1998). It matches the bankrupt firm with all comparison firms and uses a kernel function to assign greater (smaller) weights to comparison firms that are closer to (more distant from) the bankrupt firm. The untabulated results are similar to the ten nearest neighbor matching.
(7) White (2007, 2009) analyzes the effects of major changes in the US bankruptcy code in 2005, when the Bankruptcy Abuse Prevention and Consumer Protection Act was signed into law. While many of the new changes are aimed at consumer bankruptcy cases, the amendments also significantly affect creditors' rights in Chapter 11, granting them increased protections and more certainty in bankruptcy proceedings. Baird and Rasmussen (2002, 2003) and Warren and Westbrook (2003) also argue that Chapter 11 has recently become more favorable to creditors.
(8) Table II shows that for the 164 bankrupt firms that did not emerge from Chapter 11, most of them were liquidated by various means.
We are grateful for the comments and suggestions of an anonymous referee and Bill Christie (Editor). Any errors are our own.
Varouj A. Aivazian and Simiao Zhou *
* Vorouj A. Aivazian is a Professor of Economics and Finance at the University of Toronto in Ontario, Canada. Simiao Zhou is an Assistant Professor of Finance at the Shanghai University of Finance and Economics
in Shanghai. China.
Table I. Empirical Measures for Bankruptcy Filing Determinants This table summarizes the empirical measures for the bankruptcy filing determinants. Compustat data items are provided in parentheses. Operating Income/Total Assets Operating income (Data 13) divided by total assets (Data 6), where operating income is measured as earnings before interest, taxes, depreciation, and amortization (EBITDA). Cash Flow/Total Assets The operating cash flow measure is obtained from the statement of cash flows in Compustat. In the initial sample, 2% of the observations have missing values in operating cash flows. They are approximated by the sum of income before extraordinary items (Data 18) and depreciation (Data 14) in the income statement. Quick Ratio The quick ratio is defined as cash and short-term investments (Data 1) plus receivables (Data 2) divided by current liabilities (Data 5). Interest Coverage Ratio The interest coverage ratio is measured as earnings before interest and taxes (Data 13--Data 14) divided by interest and related expenses (Data 15). MV of Equity/Total Liabilities The market value of equity is calculated as the product of common shares outstanding (Data 25) and the close price at the fiscal year end (Data 199). In our sample, some firms do not have trading data in the first fiscal year prior to the filing. To minimize any loss of bankrupt firms from our sample, we averaged the ratio of equity value to total liabilities (Data 181) over the three years preceding the bankruptcy filing. Fixed Assets/Total Assets Fixed assets are measured by a firm's net value of property, plant, and equipment (Data 8). Intangible Assets/Total Assets Intangible assets are measured by the net value of intangible assets (Data 33) that have no physical existence in themselves, but represent rights to some privilege. Secured Debt/Total Liabilities The proportion of secured debt in total liabilities, where secured debt represents all long-term debt secured or collateralized by a mortgage, property, receivable, stock, or other assets (Data 241). Trade Credits/Total Assets The ratio of trade credits to total assets, where trade credits are trade obligations due within one year or within the normal operating cycle of the company (Data 70). Auditor's Opinion The variable indicates the auditor's opinion (Data 149) on a company's financial statements and has a score of 1-4 with a higher value representing a better quality of information disclosure in the company's financial statement. Table II. Distribution of Firms by Postbankruptcy Profiles This table reports the breakdown of bankrupt firms by different profiles after they file for bankruptcy. The initial sample has 464 nonfinancial firms in the Bankruptcy Research Database that filed for Chapter 11 from 1987 to 2008 and have prefiling accounting data available in the Compustat database. Number of Firms Firms that emerged and had postemergence earnings 158 information Firms that emerged, but were deleted from 127 Compustat postemergence Firms that were deleted because of merger and 21 acquisition Firms that were deleted because of bankruptcies 27 Firms that were deleted because they became 20 private Firms that were deleted because they no longer 42 filed with the SEC Firms that were deleted for unknown reasons 17 Firms that did not emerge from Chapter 11 164 Firms that sold almost all assets under [section] 70 363 of Chapter 11 Firms that were liquidated or acquired when plans 83 were confirmed Firms whose filings were converted to Chapter 7 11 proceedings Firms whose filings were dismissed 1 Firms whose outcomes were unclassifiable 3 Firms whose filings were still pending 11 Total 464 Table III. Summary Statistics for Prefiling Characteristics Panel A summarizes prefiling characteristics for 158 nonfinancial firms in the Bankruptcy Research Database that filed for Chapter 11 from 1987 to 2008 and have prefiling and postemergence data available in the Compustat database. The bankrupt firms' prefiling years are the first fiscal years preceding the filing dates. Panel B includes the universe of comparison observations consisting of 28,454 nonbankrupt, nonfinancial firm-years in Compustat. Their statistics are summarized over the same range of prefiling years for bankrupt firms. Panel C reports summary statistics for the matched industry median firms in the same SIC industries as the bankrupt firms. Panel DI summarizes characteristics for industry and operating income matched firms that have the same major SIC codes and have the closest prefiling operating income to the bankrupt firms. Panel D2 presents descriptive statistics for industry and cash flow matched firms that have the same major SIC codes and have the closest prefiling cash flow to the bankrupt firms. To test the equality of means (medians) between bankrupt firms and the matched firms, the Mest (sign-test) for matched-pair data is carried out. Mean Median SD Panel A. Bankruptcy Filing Firms Operating income/total assets 0.039 0.051 0.108 Cash flow/total assets -0.121 -0.050 0.241 Quick ratio 0.569 0.419 0.561 Interest coverage ratio -0.773 -0.028 3.882 MV of equity/total liabilities 0.579 0.278 0.965 Fixed assets/total assets 0.410 0.393 0.240 Intangible assets/total assets 0.132 0.062 0.164 Secured debt/total liabilities 0.170 0.073 0.219 Trade credits/total assets 0.094 0.076 0.082 Auditor's opinion 3.114 3.000 0.677 Panel B. Comparison Firm-Years Operating income/total assets 0.139 0.132 0.076 Cash flow/total assets 0.091 0.089 0.074 Quick ratio 1.243 0.912 1.492 Interest coverage ratio 15.056 3.758 54.916 MV of equity/total liabilities 2.618 1.339 4.386 Fixed assets/total assets 0.408 0.359 0.244 Intangible assets/total assets 0.107 0.032 0.158 Secured debt/total liabilities 0.084 0.002 0.168 Trade credits/total assets 0.083 0.065 0.073 Auditor's opinion 3.647 4.000 0.534 Panel C. Matched Industry Median Firms Operating income/total assets 0.127 *** 0.128 *** 0.035 Cash flow/total assets 0.085 *** 0.082 *** 0.026 Quick ratio 0.881 *** 0.874 *** 0.351 Interest coverage ratio 3.749 *** 2.779 *** 3.784 MV of equity/total liabilities 1.589 *** 1.232 *** 1.686 Fixed assets/total assets 0.407 0.394 * 0.189 Intangible assets/total assets 0.064 *** 0.036 ** 0.083 Secured debt/total liabilities 0.059 *** 0.013 *** 0.103 Trade credits/total assets 0.080 ** 0.068 0.047 Auditor's opinion 3.829 *** 4.000 *** 0.369 Panel D1. Industry and Operating Income Matched Firms Operating income/total assets 0.050 *** 0.054 * 0.087 Quick ratio 1.534 *** 1.062 *** 1.566 Interest coverage ratio 1.397 *** 0.811 *** 10.567 MV of equity/total liabilities 3.132 *** 1.074 *** 6.787 Fixed assets/total assets 0.377 0.349 * 0.233 Intangible assets/total assets 0.101 * 0.024 ** 0.168 Secured debt/total liabilities 0.123 ** 0.014 * 0.184 Trade credits/total assets 0.091 0.070 0.080 Auditor's opinion 3.640 *** 4.000 *** 0.552 Panel D2. Industry and Cash Flow Matched Firms Cash flow/total assets -0.062 *** -0.017 *** 0.130 Quick ratio 1.476 *** 0.975 *** 1.996 Interest coverage ratio -0.088 0.391 * 7.492 MV of equity/total liabilities 2.774 *** 0.861 *** 5.250 Fixed assets/total assets 0.342 *** 0.313 *** 0.233 Intangible assets/total assets 0.121 0.061 0.171 Secured debt/total liabilities 0.107 *** 0.018 0.184 Trade credits/total assets 0.080 * 0.058 ** 0.075 Auditor's opinion 3.607 *** 4.000 *** 0.520 P25 P75 Panel A. Bankruptcy Filing Firms Operating income/total assets -0.008 0.096 Cash flow/total assets -0.187 0.028 Quick ratio 0.173 0.743 Interest coverage ratio -1.018 0.623 MV of equity/total liabilities 0.137 0.603 Fixed assets/total assets 0.220 0.598 Intangible assets/total assets 0.000 0.223 Secured debt/total liabilities 0.000 0.294 Trade credits/total assets 0.040 0.123 Auditor's opinion 3.000 3.000 Panel B. Comparison Firm-Years Operating income/total assets 0.094 0.178 Cash flow/total assets 0.059 0.128 Quick ratio 0.614 1.354 Interest coverage ratio 1.860 8.411 MV of equity/total liabilities 0.727 2.738 Fixed assets/total assets 0.205 0.609 Intangible assets/total assets 0.000 0.156 Secured debt/total liabilities 0.000 0.071 Trade credits/total assets 0.037 0.103 Auditor's opinion 3.000 4.000 Panel C. Matched Industry Median Firms Operating income/total assets 0.105 0.145 Cash flow/total assets 0.071 0.099 Quick ratio 0.642 1.113 Interest coverage ratio 1.807 4.348 MV of equity/total liabilities 0.855 1.771 Fixed assets/total assets 0.250 0.537 Intangible assets/total assets 0.003 0.094 Secured debt/total liabilities 0.000 0.065 Trade credits/total assets 0.046 0.101 Auditor's opinion 4.000 4.000 Panel D1. Industry and Operating Income Matched Firms Operating income/total assets 0.004 0.095 Quick ratio 0.678 1.637 Interest coverage ratio -1.359 3.235 MV of equity/total liabilities 0.571 2.384 Fixed assets/total assets 0.201 0.561 Intangible assets/total assets 0.000 0.150 Secured debt/total liabilities 0.000 0.202 Trade credits/total assets 0.037 0.116 Auditor's opinion 3.000 4.000 Panel D2. Industry and Cash Flow Matched Firms Cash flow/total assets -0.148 0.028 Quick ratio 0.656 1.494 Interest coverage ratio -1.678 1.628 MV of equity/total liabilities 0.541 1.933 Fixed assets/total assets 0.161 0.525 Intangible assets/total assets 0.000 0.173 Secured debt/total liabilities 0.000 0.134 Trade credits/total assets 0.036 0.104 Auditor's opinion 3.000 4.000 *** Significant at the 0.01 level. ** Significant at the 0.05 level. * Significant at the 0.10 level. Table IV. Estimates of the Logit Regressions This table reports logit estimates of firms' propensity to file for bankruptcy. The dependent variable is one for bankrupt firm-years and zero otherwise. The explanatory variables are the bankruptcy determinants presented in Table I and measured in the bankrupt firms' prefiling years. Panel A presents estimates when operating performance is measured by operating income, while Panel B provides the results when operating performance is measured by cash flow. The regressions also control for year dummy variables. ME is the marginal effects evaluated at the 85th (90th) percentile of intangible assets-total assets for bankrupt firms in Panel A (B) and at the sample means for all other variables. The marginal effect for the auditor's opinion dummy variable is for the discrete change from zero to one. z-statistics are presented in parentheses and are computed using robust standard errors. Coef. SE Panel A. Operating Performance Measured by Operating Income Operating income/total assets -11.271 *** (2.509) Quick ratio -1.494 *** (0.406) Interest coverage ratio -0.089 *** (0.029) MV of equity/total liabilities -0.930 *** (0.359) Fixed assets/total assets -0.517 (0.517) Intangible assets/total assets 4.952 *** (1.550) [(Intangible assets/total assets).sup.2] -7.678 *** (2.538) Secured debt/total liabilities 1.880 *** (0.393) Trade credits/total assets 0.498 (1.002) Auditor's opinion dummy2 1.085 (0.741) Auditor's opinion dummy3 -0.142 (0.511) Auditor's opinion dummy4 -1.846 *** (0.497) Constant -3.150 *** (1.093) Number of observations Log pseudolikelihood Pseudo [R.sup.2] Panel B. Operating Performance Measured by Cash Flow Cash flow/total assets -8.970 *** (0.774) Quick ratio -1.074 *** (0.342) Interest coverage ratio -0.107 *** (0.017) MV of equity/total liabilities -0.924 *** (0.305) Fixed assets/total assets -0.214 (0.539) Intangible assets/total assets 4.628 *** (1.616) [(Intangible assets/total assets).sup.2] -6.882 ** (2.706) Secured debt/total liabilities 1.582 *** (0.417) Trade credits/total assets 0.085 (1.223) Auditor's opinion dummy2 0.859 (0.924) Auditor's opinion dummy3 -0.216 (0.595) Auditor's opinion dummy4 -1.720 *** (0.572) Constant -4.457 *** (1.148) Number of observations Log pseudolikelihood Pseudo [R.sup.2] ME Panel A. Operating Performance Measured by Operating Income Operating income/total assets -0.00093 Quick ratio -0.00012 Interest coverage ratio -0.00001 MV of equity/total liabilities -0.00008 Fixed assets/total assets -0.00004 Intangible assets/total assets -0.00001 [(Intangible assets/total assets).sup.2] Secured debt/total liabilities 0.00016 Trade credits/total assets 0.00004 Auditor's opinion dummy2 0.00016 Auditor's opinion dummy3 -0.00001 Auditor's opinion dummy4 -0.00024 Constant Number of observations 28,612 Log pseudolikelihood -639.11 Pseudo [R.sup.2] 0.3472 Panel B. Operating Performance Measured by Cash Flow Cash flow/total assets -0.00068 Quick ratio -0.00008 Interest coverage ratio -0.00001 MV of equity/total liabilities -0.00007 Fixed assets/total assets -0.00002 Intangible assets/total assets -0.00003 [(Intangible assets/total assets).sup.2] Secured debt/total liabilities 0.00012 Trade credits/total assets 0.00001 Auditor's opinion dummy2 0.00010 Auditor's opinion dummy3 -0.00002 Auditor's opinion dummy4 -0.00020 Constant Number of observations 28,612 Log pseudolikelihood -580.07 Pseudo [R.sup.2] 0.4075 *** Significant at the 0.01 level. ** Significant at the 0.05 level. Table V. Summary Statistics for the Propensity Score Matched Firms This table presents the summary statistics for the matched nonbankrupt firms whose estimated propensity scores are closest to those of bankrupt firms. The propensity scores in Panel A are estimated by the logit regression in Table IV using operating income as the performance measure, while those in Panel B are estimated by the logit regression using cash flow as the performance measure. The prefiling characteristics of bankrupt firms are presented in Panel A of Table III and the summary statistics for the propensity score matched firms are measured in bankrupt firms' prefiling years. To test the equality of means (medians) between bankrupt firms and the matched firms, the t-test (sign-test) for matched-pair data is employed. Mean Median SD Panel A. Operating Income is Used to Estimate Propensity Scores Operating income/total assets 0.033 0.055 0.081 Quick ratio 0.589 0.532 0.391 Interest coverage ratio -0.058 * 0.790 *** 3.600 MV of equity/total liabilities 0.492 0.428 0.411 Fixed assets/total assets 0.451 0.417 0.258 Intangible assets/total assets 0.133 0.068 0.171 Secured debt/total liabilities 0.219 * 0.098 0.257 Trade credits/total assets 0.080 * 0.056 0.070 Auditor's opinion 3.095 3.000 0.772 Panel B. Cash Flow is Used to Estimate Propensity Scores Cash flow/total assets -0.087 -0.042 0.139 Quick ratio 0.578 0.517 0.410 Interest coverage ratio -0.290 0.216 *** 4.312 MV of equity/total liabilities 0.518 0.393 0.543 Fixed assets/total assets 0.422 0.396 0.218 Intangible assets/total assets 0.155 0.104 0.168 Secured debt/total liabilities 0.172 0.040 0.220 Trade credits/total assets 0.083 0.056 ** 0.083 Auditor's opinion 3.133 3.000 0.659 P25 P75 Panel A. Operating Income is Used to Estimate Propensity Scores Operating income/total assets -0.014 0.090 Quick ratio 0.360 0.764 Interest coverage ratio -0.750 1.507 MV of equity/total liabilities 0.207 0.628 Fixed assets/total assets 0.248 0.684 Intangible assets/total assets 0.000 0.221 Secured debt/total liabilities 0.000 0.417 Trade credits/total assets 0.033 0.110 Auditor's opinion 3.000 4.000 Panel B. Cash Flow is Used to Estimate Propensity Scores Cash flow/total assets -0.162 0.027 Quick ratio 0.321 0.753 Interest coverage ratio -0.917 1.453 MV of equity/total liabilities 0.145 0.633 Fixed assets/total assets 0.259 0.599 Intangible assets/total assets 0.001 0.266 Secured debt/total liabilities 0.000 0.368 Trade credits/total assets 0.033 0.110 Auditor's opinion 3.000 3.000 *** Significant at the 0.01 level. ** Significant at the 0.05 level. * Significant at the 0.10 level. Table VI. Estimation by Propensity Score Matching and Industry Benchmarking This table presents alternative estimation results for a group of bankrupt nonfinancial firms in the Bankruptcy Research Database that filed for Chapter 11 from 1987 to 2008 and emerged as public companies. Panel A illustrates short-term changes in operating performance from the prefiling year to the first postemergence year; Panel B summarizes long-term performance changes that are calculated by subtracting the prefiling performance from the average performance over the three years after emergence. Two measures are used to gauge operating performance. The results for the EBITDA- to-total-assets ratio are presented in Columns 2-6, while the results for the ratio of operating cash flow to total assets can be found in Columns 7-11. Part 1 reports propensity score matching estimates where bankrupt firms are compared with k matched neighbors that have the nearest propensity scores. Part 2 provides estimation results comparing bankrupt firms with the industry median firms in the same SIC industries. Part 3 presents the results obtained when comparing bankrupt firms with those firms that have the same major SIC codes and have the closest operating performance in prefiling years. "No." refers to the number of bankrupt firms. "Observed" is the average observed operating performance for bankrupt firms. "Counter." is their average counterfactual performance, and "Control" refers to the average performances of industry median firms in Part 2 and industry and performance matched firms in Part 3. "ATT" is the average difference between bankrupt firms' observed and counterfactual performances, and "Est." refers to the difference between "Observed" and "Control." Panel A. Short-Term Effects on Operating Performance Part 1: Propensity Score Matched Results Short-Term Change in Operating Income k No. Observed Counter. ATT t-stat 1 158 0.041 0.044 -0.003 (-0.33) 4 158 0.041 0.038 0.003 (0.35) 6 158 0.041 0.037 0.004 (0.46) 10 158 0.041 0.033 0.008 (0.96) Part 2: Industry Median Matched Results No. Observed Control Est. t-stat 157 0.042 0.125 -0.083 *** (-9.16) Part 3: Industry and Performance Matched Results No. Observed Control Est. t-stat 158 0.041 0.052 -0.011 (-1.26) Panel B. Long-Term Effects on Operating Performance Part 1: Propensity Score Matched Results Long-Term Change in Operating Income k No. Observed Counter. ATT t-stat 1 114 0.046 0.037 0.009 (0.97) 4 114 0.046 0.036 0.011 (1.22) 6 114 0.046 0.033 0.014 * (1.64) 10 114 0.046 0.026 0.021 *** (2.52) Part 2: Industry Median Matched Results No. Observed Control Est. t-stat 113 0.049 0.138 -0.089 *** (-8.93) Part 3: Industry and Performance Matched Results No. Observed Control Est. t-stat 114 0.046 0.050 -0.004 (-0.49) Short-Term Change in Cash Flow k No. Observed Counter. ATT t-stat 1 158 0.135 0.104 0.031 (1.24) 4 158 0.135 0.090 0.045 * (1.88) 6 158 0.135 0.075 0.060 ** (2.50) 10 158 0.135 0.066 0.069 *** (2.89) Part 2: Industry Median Matched Results No. Observed Control Est. t-stat 157 0.136 0.126 0.010 (0.39) Part 3: Industry and Performance Matched Results No. Observed Control Est. t-stat 158 0.135 0.118 0.017 (0.75) Panel B. Long-Term Effects on Operating Performance Part 1: Propensity Score Matched Results Long-Term Change in Cash Flow k No. Observed Counter. ATT t-stat 1 114 0.163 0.089 0.074 *** (4.20) 4 114 0.163 0.064 0.098 *** (5.76) 6 114 0.163 0.051 0.112 *** (6.37) 10 114 0.163 0.041 0.122 *** (6.76) Part 2: Industry Median Matched Results No. Observed Control Est. t-stat 113 0.165 0.138 0.027 (1.23) Part 3: Industry and Performance Matched Results No. Observed Control Est. t-stat 114 0.163 0.134 0.029 * (1.72) *** Significant at the 0.01 level. ** Significant at the 0.05 level. * Significant at the 0.10 level. Table VII. Changes in Leverage, Taxes, and Interest Expenses This table presents changes in leverage, taxes, and interest expenses for the reorganized firms and their nearest neighbor firms matched by propensity scores. All level variables are in real values of 1980 dollars, and the changes are from the prefiling year to the first postemergence year. Total debt is long-term debt plus short- term debt. The t-test (sign-test) for matched pair data is used to test the equality of means (medians) between reorganized firms and the matched firms. Mean Changes Reorganized Matched Change in income taxes 15.207 -9.825 Change in interest expenses -34.095 -23.006 Change in (income taxes/total assets) 0.011 * 0.001 Change in (interest expenses/total assets) -0.029 *** -0.009 Change in total assets -441.985 ** 4.022 Change in total debt -287.421 *** -24.565 Change in (total debt/total assets) -0.202 0.021 Median Changes Reorganized Matched Change in income taxes 0.505 0.297 Change in interest expenses -18.451 *** -4.503 Change in (income taxes/total assets) 0.004 0.001 Change in (interest expenses/total assets) -0.027 *** -0.004 Change in total assets -142.496 *** -34.677 Change in total debt -126.181 *** -12.087 Change in (total debt/total assets) -0.178 *** 0.007 *** Significant at the 0.01 level. ** Significant at the 0.05 level. * Significant at the 0.10 level. Table VIII. Summary Statistics for Firms in Different Clusters Multivariate cluster analysis is employed to partition bankrupt firms into k distinct nonoverlapping groups. The Calinski and Harabasz stopping rule shows that k = 2 produces the most distinct grouping. For firms in each group, this table summarizes the estimated changes in cash flow and prefiling firm characteristics. Panel A (B) is comprised of firms that are used in Table VI to estimate the short-term (long-term) effects on cash flow. Definitions of prefiling firm characteristics are summarized in Table 1. Group 1 N Mean Median Panel A. Short-Term Changes in Cash Flow and Prefiling Characteristics 1-neighbor matching estimates 69 0.086 0.074 4-neighbor matching estimates 69 0.100 0.097 6-neighbor matching estimates 69 0.121 0.107 10-neighbor matching estimates 69 0.135 0.115 Cash flow/total assets 69 -0.228 -0.168 Quick ratio 69 0.603 0.451 Interest coverage ratio 69 -2.737 -1.165 MV of equity/total liabilities 69 0.808 0.389 Fixed assets/total assets 69 0.412 0.405 Intangible assets/total assets 69 0.115 0.058 Secured debt/total liabilities 69 0.129 0.024 Trade credits/total assets 69 0.108 0.083 Auditor's opinion 69 2.957 3.000 Panel B. Long-Term Changes in Cash Flow and Prefiling Characteristics 1-neighbor matching estimates 49 0.107 0.123 4-neighbor matching estimates 49 0.143 0.139 6-neighbor matching estimates 49 0.160 0.142 10-neighbor matching estimates 49 0.170 0.152 Cash flow/total assets 49 -0.204 -0.168 Quick ratio 49 0.688 0.523 Interest coverage ratio 49 -2.915 -1.134 MV of equity/total liabilities 49 0.864 0.382 Fixed assets/total assets 49 0.397 0.405 Intangible assets/total assets 49 0.120 0.063 Secured debt/total liabilities 49 0.123 0.024 Trade credits/total assets 49 0.102 0.064 Auditor's opinion 49 2.939 3.000 Group 2 N Mean Median Panel A. Short-Term Changes in Cash Flow and Prefiling Characteristics 1-neighbor matching estimates 89 -0.012 0.007 4-neighbor matching estimates 89 0.002 -0.010 6-neighbor matching estimates 89 0.012 0.005 10-neighbor matching estimates 89 0.019 0.010 Cash flow/total assets 89 -0.039 0.010 Quick ratio 89 0.543 0.390 Interest coverage ratio 89 0.750 0.584 MV of equity/total liabilities 89 0.402 0.231 Fixed assets/total assets 89 0.409 0.384 Intangible assets/total assets 89 0.146 0.068 Secured debt/total liabilities 89 0.202 0.116 Trade credits/total assets 89 0.083 0.070 Auditor's opinion 89 3.236 3.000 Panel B. Long-Term Changes in Cash Flow and Prefiling Characteristics 1-neighbor matching estimates 65 0.049 0.034 4-neighbor matching estimates 65 0.065 0.043 6-neighbor matching estimates 65 0.076 0.054 10-neighbor matching estimates 65 0.085 0.058 Cash flow/total assets 65 -0.055 -0.004 Quick ratio 65 0.485 0.354 Interest coverage ratio 65 0.768 0.596 MV of equity/total liabilities 65 0.411 0.241 Fixed assets/total assets 65 0.412 0.384 Intangible assets/total assets 65 0.153 0.093 Secured debt/total liabilities 65 0.177 0.089 Trade credits/total assets 65 0.075 0.069 Auditor's opinion 65 3.200 3.000