The Effect of Operating Cash Flow on the Likelihood and Duration of Survival for Marginally Distressed Firms in Taiwan

Abstract

The purpose of this study was to investigate the effect of operating cash flow (OCF) on the likelihood and the duration of distressed firms returning to a profitable position for survival. By selecting 309 marginally distressed firms that are Taiwan listed firms, we identified 218 firms that survived from financial distress and 91 firms that did not survive from financial distress for the logistic regression model. We found that the greater adequacy, stability, and growth of changes in OCF and the higher liquidity, growth, and size of firms significantly increased the likelihood of firm survival, suggesting that a distressed firm is more likely to return to profitability for survival if it can improve OCF after suddenly encountering financial distress. Moreover, applying duration analysis, this study took a further step to investigate the time dependence of firm survival among 218 surviving firms. The results suggest that firms generating more OCF in the post-distress period and possessing higher profitability, liquidity, and growth in the pre-distress period significantly took less time on resolving financial distress for survival. However, an economic recession can significantly impede the time and speed of firm survival. Overall, the study found consistent and robust evidence that OCF is a reliable instrument to predict the likelihood and duration of survival for financially distressed firms. The study also provides practical implications for managers, investors, policymakers, and lenders who intend to promote firm financial performance and sustainability.

1. Introduction

From a theoretical perspective, a firm conducts its business activities to achieve its basic goal of earning profits. However, while some firms succeed in achieving this goal so that the firm can continuously operate, others fail to achieve sustainability. Although financial distress does not necessarily lead to bankruptcy, a sustained decline in a firm’s financial performance affects investments, repayments, and both managers’ and shareholders’ wealth adversely [1] and therefore needs to be accurately predicted. Additionally, if a firm does experience a sudden financial shock, can managers, creditors, and investors predict which firms are likely to recover, and if so, the length of time in which firm recovery will occur [2]? This forms an empirical question.

However, survival and duration are two key performance issues while firms suffer financial distress. Managers who identify a firm that is in financial distress will take strategic actions in an effort to alleviate the distress; for example, a flow-based mitigation strategy that focused on improving operational profitability is a key determinant of survival from a situation of financial distress [3]. The survival of a firm depends on its capability to operate profitably and manage the timing of cash receipts and expenditures [4] because external financing may not be readily available to financially distressed firms [5]. Consequently, firms must rely on internal cash generation, such as operating cash flow (OCF) to offer adequate cash and liquidity to fund normal operations [5]. From this perspective, market participants may view positive OCF for a distressed firm as a sign that it is less risky and better able to survive financial difficulties. Nevertheless, there is a lack of empirical evidence on the relative value relevance of OCF in the context of distressed firms’ survival, especially at the early distressed stage.

To address the importance of OCF to the survival of distressed firms, there are three main reasons behind our choice of examining the effect of OCF on the likelihood and the duration of survival for financially distressed firms. First, there is a strong link between cash flow insufficiency and business distress [4,6], highlighting the importance of our analysis. While OCF can be viewed as the net inflow of liquid assets, the larger the inflow, the lower the probability of business failure [7]. Second, compared to cash-flow-based ratios (e.g., OCF ratios), accrual-based ratios (e.g., earnings ratios) are often easily manipulated by distressed firms in order to conceal their unfavorable situation [6,8,9,10]. However, cash-flow indicators seem to be neglected in the literature [11]. Third, the cash flow approach is a widely recognized way for estimating the business value, if businesses can meet the going concern assumption that they are financially viable [4]. Moreover, the survival and growth of the firm can be determined by the capacity to generate cash into business [12].

At this point, this study attempts to fill the gap and provide some insights into the related literature by exploring whether the changes in OCF after post-distress has a significant effect on the likelihood of survival in a binary logistic regression model and the duration of recovery in a dynamic discrete-time survival analysis model. More specifically, this study considers the adequacy, stability, and growth of changes in OCF as well as OCF indicators and the pre-distress firm characteristics to construct the theory-based variables for the empirical models, in addition to control variables such as firm size, economic recession dummy, and industry dummy. We analyzed a sample of 309 marginally distressed firms that are Taiwan listed firms, consisting of 218 surviving firms and 91 non-surviving firms over the period of 2003–2017.

Our main findings are that the adequacy, stability, and growth of incremental OCF significantly influence the likelihood of firms surviving from financial difficulties and shorten the time needed for a distressed firm to return to profitability. Therefore, this empirical evidence suggests that the greater adequacy, stability, and growth of changes in OCF in the distressed episode for a distressed firm, the higher probability and the shorter the time required for survival. For the firm characteristics, the distressed firms with higher short-term liquidity, greater sales growth, and larger size before the onset of financial distress are more likely to survive. Additionally, the distressed firms that had higher performance of profitability, liquidity, and growth in the pre-distress situation took less time in resolving financial distress for survival; however, an economic recession significantly slows up eventual survival. Our results are robust, regardless of if OCF indicators are alternatively proxied by either of the other measures of incremental OCF.

Our paper attempts to contribute to the existing literature on the information content of OCF and distressed firms’ survival measures in several ways. Firstly, this paper provides empirical findings from a much earlier stage in the firm financial distress cycle to provide an early warning regarding firms that are heading to recovery or failure. Specifically, the emphasis in our study is on collecting and tracing the sample of marginally distressed firms (e.g., [2]) from a much earlier period through the firm performance in the distress cycle, thereby complementing recent studies that focus on the post-distress stage, such as firms with a default rating [4,6,13,14], distressed Z-score [3,12], and bankruptcy [15]. Secondly, the literature on accounting and financial management has focused increasing attention on the costs and benefits of OCF. However, less is known on how the variety dimensions of OCF are shared on resolving financial distress for firm survival. In this study, the characteristics of OCF were measured in three cash-flow-based dimensions: adequacy, stability, and growth. This innovative analysis would be more comprehensive and informative if we are able to capture more information content of OCF that can be helpful in predicting the likelihood of firm performance and their survival after financial distress. Thirdly, the empirical results also can assist managers, investors, policymakers, and lenders who intend to promote firm financial sustainability and performance. The identification of factors that determine the probability of success or failure and the speed of recovery would be a useful tool for managers making decisions to turnaround distressed firms in the future. Knowledge of the associated explanatory determinants of firm survival will be helpful for investors in judging whether and when to buy or sell their position on distressed firms. This analysis also has implications for strategy making to policymakers and potential benefit to lenders.

The remainder of the paper is organized as follows. Section 2 provides an overview of the literature and develops the hypothesis. Section 3 describes sample data, methodology, and variables used in our analysis. Section 4 presents the main empirical results. Section 5 presents the conclusions of the paper.

2. Literature Review and Hypotheses Development

2.1. Related Literature on Determinants and Consequences of Finicial Distress

It is not surprising that the study of firm bankruptcy or financial distress has received a lot of attention in accounting and finance literature. For example, both the accounting-based distress prediction model [3,6,7,16,17,18,19,20,21,22] and the market-based prediction model [23,24,25,26,27,28] are fairly commonly used to examine the determinants and consequences of financial distress. The mainstream accounting and finance research still seems to believe that the ability of financial statements to assess financial distress is important [1], although this approach suffers from some inherent limitations, such as its use of historical costs, failure to capture intangible assets and expected future volatility, and limited scope [29]. Finicial distress offers a context in which to evaluate the usefulness of financial statements [1], and most of research is geared toward testing the informative content of financial statements as a predictor, looking for a relationship between accounting data and future solvency. In contrast, market-based indicators as predictors of distress require data from capital markets, which limits their application in environments with an underdeveloped capital market, such as emerging markets [6]. This study therefore adopts accounting-based measures to proxy for financial distress and to examine the likelihood of distressed firms that return to profitability and the duration of their turnaround since initial distress.

The event-based variables, such as debt restructuring, shareholder changes, asset replacement, and the changes in operation information, greatly improve the ability to predict the probability of subsequent distress [30]. However, the cash-flow-based proxies (especially cash from operation) are more useful in assessing the likelihood of emergence from distress as they capture progress in ongoing strategic efforts [3]. For example, following the cash-flow-based proxies, substantial recent studies have specifically focused on the effect of OCF in predicting financial distress [4,6,12,14], and their results show that OCF has a negative effect on financial distress [6]. Specifically, firms having high OCF means that they have sources of funds to carry out their operating activities. If the OCF generated by the firm has increased, it is less likely that the firm will experience financial distress; however, if the firm’s OCF has decreased continuously without being overcome, the firm can experience financial distress [4]. Therefore, the information content of OCF is known to be predictive of firm failure, financial distress, and bankruptcy. Whether it is useful to predict the survival of firms from a situation of marginal distress is an empirical question.

2.2. Hypotheses Development

In the long run, the cash flows generated by operating activities are inadequate to cover financial obligations, resulting in firms having both difficulties paying interest on loans and also having a tendency to fall into financial distress [31,32]. Thus, these firms have an inferior ratio of liquidity that is measured by OCF divided by total liabilities, and the probability of business failure is higher in such cases [31]. Moreover, an efficiency ratio of OCF to total assets illustrates how well the firms utilize their assets to generate cash flows in their operations [33]. This ratio is good for both internal and external users as it captures the effectiveness of a firm in using its assets to keep away from financial difficulties [34]. Following the arguments of the above-mentioned literature, the cash flow ratio of profitability is more appropriate to measure the firm’s operating profitability, and we thus expect that the related ratios of OCF not only have a significant relation with the prediction of financial distress but to also have a significant association with the survival of financial distress. Hence, we focus on investigating whether the effect of OCF may benefit firms on resolving financial distress to enhance the likelihood of their survival. We thus hypothesize:

Hypothesis 1 (H1).

The greater the adequacy, stability, and growth of OCF, the higher the likelihood that a distressed firm survives.

As mentioned earlier, while some prior literature has shown how measures of OCF performance are associated with the prediction of financial distress and are important to the likelihood of survival, their effectiveness and efficacy on whether or not these measures influence the duration of distress is limited. Firms should respond to financial distress by taking corrective action or restructuring to recover from it [24]. Previous research suggests that the duration factor might be taken into account because the timeframe must be considered a key variable in firm failure or survival predictions [15]. Hence, the other interest of ours concerns the duration with which a firm returns from a loss position to a profitable position. A related previous paper examines the determinants that help explain variations in the time to survival of marginally distressed firms [2]. Its results suggest that the speed of a distressed firm’s recovery changes over time as the relevant managerial responses change. In addition, other empirical findings indicated that survival time is longer among new firms entering the market with higher levels of OCF [35]. Firms having high OCF means that they have sources of funds to implement their operating activities. The more the OCF generated by firm has increased, the less likely the firm will be to experience financial distress [4]. The OCF ratios are able to quantify the success of the firm because the survival and growth can be determined by the capacity to generate cash into business [36]. Thus, we expect that OCF might have a negative effect on the duration of firm survival (i.e., less time spent on recovery) as the more OCF generated, the better ability of the firm to bear and repay debts for survival. This study differs from previous studies; in particular, we focus on examining the impact of OCF on the duration of a distressed firm moving from loss to profitability for survival. We thus hypothesize:

Hypothesis 2 (H2).

The greater the adequacy, stability, and growth of OCF, the shorter the recovery time for a distressed firm.

In line with the theoretical considerations, in this paper, we adopted the cash-flow-based variables and propose the construction of OCF indicators such as the adequacy, stability, and growth of OCF as the main explanatory variable, which is justified by the following: (1) The adequacy of the changes in OCF that is a measure of a firm’s ability to repay debts with assets generated by the firm’s main and recurring activity [4]. In addition, this indicator is often mentioned as a powerful predictor, especially in relation to total debts [37]. (2) The stability of the changes in OCF that presents the changes in OCF to its volatility, considering the continuous and stable OCF being an important guarantee for the normal operation and risk avoidance of firms. (3) The growth of the changes in OCF that shows a firm’s growth of OCF, reflecting a firm’s capacity to repay its debts in future periods.

While there are many measures of financial distress, there is no theoretical guidance that provides insights into the factors on the likelihood of a distressed firm’s survival or on the duration of distressed firms’ survival [2,3]. In accessing the determinants on the likelihood of firm survival from a status of financial distress, the above studies have argued that a firm’s assets, size, earnings prospects and management commitment [4,38], solvency and liquidity [39,40], and profitability [40,41] are significantly associated with firm emergence from financial distress. On the other hand, a number of other studies suggest that firm size and profitability are not relevant factors in a firm’s emergence from financial distress [13,42]. In summary, the effect of firm characteristics on predicting the likelihood of firm survival remains an open issue because the results are mixed. Hence, this paper attempts to document empirical regularities and provide evidence on determinants that might interpret variations in the probability and the time to survival of firms experiencing financial distress. Our choice of the predictors for examining the determinants of firm survival mainly relies on three main factors: (1) indicators of OCF such as the adequacy, stability, and growth of changes in OCF; (2) pre-distress firm characteristics such as the profitability, liquidity, leverage, and growth of firms; and (3) control factors, including firm size, economic recession, and industry.

3. Data and Methodology

3.1. Data and Sample Selection

Poor profits may point to a firm that is financially unhealthy and may be a sign of financial distress. In this paper, we followed a previous study [2] that defines a “marginally financial distressed firm” as a firm that has negative operating profits (hereafter “loss”) for two consecutive years. Further, we define a “survival firm” as a firm that returns to profitability by reporting positive operating profits (hereafter “profitability”) for at least three consecutive years after the year of initially becoming a marginally distressed firm. It is worth explaining that the significance of adopting operating profits (also known as operating net income) to define a marginally distressed firm is to mainly consider that non-operating income and non-operating expenditures are manipulable and cannot truly represent the operation results of the firm for a period [6,8,9,10]. Additionally, the operating profit is to represent the profit before the non-operating income and expenditure as well as the revenue after the total revenue deducts the cost of goods sold, the management and marketing expenses, and the R&D expenditures, which reflect the ability of the firm to generate profits from the operation of the business. We therefore use this definition as the sample selection criterion.

This study utilized a two-step sample selection procedure. First, a total of 337 marginally distressed firms were screened according to the definition of a marginally distressed firm that a firm has reported negative operating profits for at least two consecutive years in the period 2003–2017. In the case of a firm with multiple distresses during the research period, the study only retained the first distress to avoid a potential econometric bias caused by several forms of distress being linked to the same firm. Second, this study further identified firms that returned to profitability by reporting positive operating profits for at least three consecutive years by tracing the sample firms until the year 2021. Thus, we eliminated nine firms whose operating profits never returned to a positive value and then entered to bankruptcy proceedings. In addition, six firms that belong to the financial industry were excluded due to the specialty of the finance industry, and 13 firms whose related financial data were missing were also eliminated. Financ.ally, a sample of 309 marginally distressed firms were selected for empirical analysis, consisting of 218 (70.55%) survival firms and 91 (29.45%) non-survival firms over the period 2003–2017.

According to the definitions of marginally distressed firms and survival firms in this study, we selected a marginally distressed firm that had negative operating profits for two consecutive years by observing the sample firms two years earlier than the first year of the research period (i.e., since from year 2001). Additionally, we selected a survival firm with positive operating profits for at least three consecutive years so that the sample firms needed to be observed for more than three years after the last year of the research period (i.e., until to year 2021). Hence, in this study, we collected all necessary research data from 2001 to 2021 and then utilized these data to examine and analyze the firms’ financial positions in the period between 2003 and 2017. The research data belongs to the censored data because this study only traced the sample firms until to year 2021. Data for most financial variables were obtained from the Taiwan Economic Journal (TEJ).

3.2. Methodology and Variables

This study used a logistic regression to identify how potent influencing variables affect the likelihood of firm survival. In addition, a duration analysis was used to determine how factors affect the time length of a distressed firm in successfully surviving. More specifically, to identify whether the marginally distressed firms survive from their losses, we estimated logistic regression models. The dependent variable in the regression equaled one if a firm survived from a status of marginal distress and zero otherwise. For the independent variables, there were two types of theory-based variables that came from four indicators of changes in OCF and four ratios of firm characteristics to determine the likelihood of the distressed firm’s survival. The control variables consisted of the firm size, economic recession dummy, and three industry dummies. We specified a logistic regression model as follows:

$$Survival=\mathit{ln}\left(\frac{Prob\left(Surv\right)}{1-Prob\left(Surv\right)}\right)={\beta }_0+\sum {\beta }_i\Delta OC{F}_i+\sum {\beta }_{j}Firm Characteristic{s}_j+\sum {\beta }_{k}Contro{l}_k+\epsilon$$

(1)

where Prob (Surv) represents the probability of successful survival of a distressed firm, and Surv is a binary dependent variable that was equal to one if the firm survived from its distress and zero otherwise. In Equation (1), the theory-based variables came from the cash-flow-based variables of the changes in OCF and the variables of the pre-distressed firm characteristics. In addition, there were four control variables, and ε is the random error item.

Specifically, this study constructed four OCF indicators as well as variables of OCF by tracing a distressed firm’s changes in OCF, which were included as the adequacy, stability, and growth of OCF in the post-distress period. There were two variables of the adequacy of OCF. One was the amount of changes in OCF to total assets (ΔOCF_adequacy1) that represented the ability of the firm’s total assets to generate cash in a certain period of time. The other was the number of changes in OCF to total liabilities (ΔOCF_adequacy2), which reflected the size of OCF that the firm can bear, representing the ability of the firm to repay debts. In particular, these two OCF indicators were the ratios of changes in OCF to total assets and total liabilities, respectively. These two OCF indicators were calculated by taking two given years, i.e., the last year of the three consecutive years (i.e., the firm survival year) in which operating profit returned to a positive value, and the year before two consecutive years (i.e., the pre-distress year) in which operating profit was a negative value, respectively. However, if the firm was a non-surviving firm, the firm survival year would be set for the year 2021, as closely as possible to the end of this research period. The stability of changes in OCF (ΔOCF_stability) captured the volatility risk of the firm by using a three-year rolling time window to calculate this indicator value. The three-year rolling window denoted the three consecutive years in which operating profits returned to a positive value. The growth of changes in OCF (ΔOCF_growth) showed the growth of the firm’s OCF by adopting a three-year window to calculate this indicator value. Importantly, all four OCF indicators showed that the higher the individual value of these four variables, the greater the adequacy, stability, and growth of changes in OCF.

In this study, there were four variables of the pre-distress firm characteristics, namely, profitability (return on assets), liquidity (short-term liquidity ratio), leverage (debt ratio), and growth (sales growth ratio) of the firm in the first year before the negative operating profit for two consecutive years. The control variables were firm size (the nature log of total assets) in the first year before the negative operating profits for two consecutive years, and the recession dummy (during the period 2007–2009 of the global economic contraction) and three industry dummies. In particular, the distressed firms of the sample were categorized into 16 industries on the basis of the security code of the Taiwan Security Exchanged Council if all different industries were represented in the study, which would have required numerous categorical independent variables, one for each industry. This would have seriously reduced our degree of freedom in estimating the parameters of the statistical estimation models [43]. Hence, this study picked three industry dummies to control for the effect of the sample industry on the likelihood of firm survival, such as the industry1 dummy for the manufacturing firms, industry2 dummy for the construction firms, and industry3 dummy for the “new economy” (e.g., the electronics and high-tech industries).

Table A1. Definition of variables.

Variables	Definition
Indicators of changes in OCF:
Adequacy of changes in OCF (ΔOCFadequacy1)	[(OCF/total assets) in period t − (OCF/total assets) in period t − x)]/ ABS(OCF/total assets) in period t − x where the period t refers to the last year of the three consecutive years (i.e., firm survival year) in which operating profits return to positive value, and the period t-x refers to the year before the two consecutive years (i.e., the pre-distress year) in which operating profit is a negative value. However, if the firm is a non-surviving firm, the firm survival year would be set for the year 2021, as closely as possible to the end of this research period. Additionally, the denominator takes an absolute value. A higher value shows a greater size of OCF generated from a firm’s assets, with a higher ability to repay debts.
Adequacy of changes in OCF (ΔOCFadequacy2)	[(OCF/total liabilities) in period t − (OCF/total liabilities) in period t − x)]/ ABS(OCF/total liabilities) in period t − x where the period t refers to the last year of the three consecutive years (i.e., firm survival year) in which operating profits return to positive value, and the period t-x refers to the year before the two consecutive years (i.e., the pre-distress year) in which operating profit is a negative value. However, if the firm is a non-surviving firm, the firm survival year would be set for the year 2021, as closely as possible to the end of this research period. Additionally, the denominator takes an absolute value. A higher value represents a greater amount of OCF for a distressed firm to repay its debts.
Stability of changes in OCF (ΔOCFstability)	(average OCF)/σ(OCF) where it adopts a three-year rolling time window to calculate average OCF and σ(OCF). σ(OCF) denotes the standard deviation of OCF. The three-year rolling window begins from the first of the three consecutive years in which operating profit returned to a positive value. However, if the firm is a non-surviving firm, the firm survival year would be set for the year 2021, as close as possible to the end of this research period. A higher value suggests a higher level of stability in the OCF of the firm, or alternatively speaking, a lower exposure to the liquidity risk of the firm.
Growth of changes in OCF (ΔOCFgrowth)	(OCF of period t − OCF of period t + 2)/ ABS(OCF of period t) where it adopts a three-year window to calculate the OCF growth ratio for a given year. The period t refers to the first year of the positive operating profit for three consecutive years. However, if the firm is a non-surviving firm, the firm survival year would be set for the year 2021, as close as possible to the end of this research period. The denominator must take an absolute value. A higher value represents a higher growth of OCF for a distressed firm.
Firm characteristic variables:
Firm profitability	Return on assets, i.e., net income over total assets in the year before the negative operating profit for two consecutive years.
Firm liquidity	Liquidity ratio, i.e., current assets over current liabilities in the year before the negative operating profit for two consecutive years.
Firm leverage	Total labilities over total assets in the year before the negative operating profit for two consecutive years.
Firm growth	Sales growth ratio for a given year as well as the year before the negative operating profit for two consecutive years, defined as follows: (net sales of period t − net sales of period t − 1)/ (net sales of period t)
Control variables:
Firm size	The natural log of total assets in the year before the negative operating profit for two consecutive years.
Recession dummy	A dummy variable takes the value of one if the financial distress occurred during the period 2007–2009 (the global economic contraction).
Industry1 dummy	A dummy variable takes the value of one if a distressed firm belongs to the manufacturing industry.
Industry2 dummy	A dummy variable takes the value of one if a distressed firm belongs to the construction industry.
Industry3 dummy	A dummy variable takes the value of one if a distressed firm belongs to the “new economy”, e.g., the electronics and high-tech industries
Robustness test variables:
Changes in OCF to equities (ΔOCFequity)	[(OCF/equities) in period t − (OCF/equities) in period t − x)]/ ABS(OCF/equities) in period t − x where the period t refers to the last year of the three consecutive years (i.e., firm survival year) in which operating profit returns to positive value, and the period t-x refers to the year before the two consecutive years (i.e., the pre-distress year) in which operating profit is a negative value. However, if the firm is a non-surviving firm, the firm survival year would be set for the year 2021, as close as possible to the end of this research period. The denominator takes an absolute value. A higher value shows a greater amount of OCF.
Changes in OCF to net income (ΔOCFnet income)	[(OCF/net income) in period t − (OCF/net income) in period t-x)]/ ABS(OCF/net income) in period t − x where the period t refers to the last year of the three consecutive years (i.e., firm survival year) in which operating profit returns to a positive value, and the period t-x refers to the year before the two consecutive years (i.e., the pre-distress year) in which operating profit is a negative value. However, if the firm is a non-surviving firm, the firm survival year would be set for the year 2021, as close as possible to the end of this research period. The denominator takes an absolute value. A higher value shows a greater amount of OCF.
Changes in OCF to sales (ΔOCFsales)	[(OCF/sales) in period t − (OCF/sales) in period t-x)]/ ABS(OCF/sales) in period t − x where the period t refers to the last year of the three consecutive years (i.e., firm survival year) in which operating profits return to a positive value, and the period t-x refers to the year before the two consecutive years (i.e., the pre-distress year) in which operating profit is a negative value. However, if the firm is a non-surviving firm, the firm survival year would be set for the year 2021, as close as possible to the end of this research period. Additionally, the denominator takes an absolute value. A higher value shows a greater amount of OCF.
Stability of changes in OCF (ΔOCFstability_5yrs)	(average OCF)/σ(OCF) where it adopts a five-year rolling time window to calculate average OCF and σ(OCF). σ(OCF) denotes the standard deviation of OCF. The five-year rolling window begins from the first of the three consecutive years in which operating profit returned to a positive value. However, if the firm is a non-surviving firm, the firm survival year would be set for the year 2021, as close as possible to the end of this research period. A higher value suggests a higher level of stability in the OCF of the firm.
Growth of changes in OCF (ΔOCFgrowth_5yrs)	(OCF of period t − OCF of period t + 4)/ ABS(OCF of period t) where it adopts a five-year window to calculate the OCF growth ratio for a given year. The period t refers to the first year of the positive operating profit for three consecutive years. However, if the firm is a non-surviving firm, the firm survival year would be set for the year 2021, as close as possible to the end of this research period. The denominator must take an absolute value. A higher value represents a higher growth of OCF for a distressed firm.

in Appendix A lists the definitions of the theory-based and control variables.

Additionally, this study further examined the factors of the duration of firm survival among 218 firms that suddenly encountered financial difficulties and then returned to a profitable position. We applied a survival analysis model to examine the effect of OCF on the duration of firm survival. Hence, the dependent variable was measured as the length of time needed (in years) for a firm to enhance its performance to positive operating profit for at least three consecutive years after suddenly encountering financial distress. The indicators of changes in OCF, firm characteristics and control variables, were independent variables as well as time-varying variables. A key feature of a survival analysis model is the hazard function, which determines the probability that a switch will occur, conditional on the spell T surviving through time t, and is defined as follows:

$$h\left(t\right)=\underset{\Delta t\to 0}{\lim }\frac{P(t\le T\le t+\Delta t|T\ge t)}{\Delta t}=\frac{f\left(t\right)}{1-F\left(t\right)},\begin{array}{cc}& 0\le t\end{array}<\infty$$

(2)

where f(t) is the density function associated with the distribution of spells. The incidence of events at duration t is equal to the density of events at t divided by the probability of surviving to that duration without experiencing the event. The hazard function offers a suitable method for summarizing the relationships between spell length and the probabilities of switching. When h(t) is increasing (decreasing) in t, the hazard function is considered to display positive (negative) duration. The reason is that the probability of ending the spell increases (decreases) as the spell lengthens, i.e., as the time passes. However, the constant duration dependence indicates the lack of a relation between h(t) and t.

We used a proportional hazard specification to estimate hazard functions as well as to examine the effects of independent variables on the duration of the survival period, such that

$$h(t,X(t),\beta )=\underset{\Delta t\to 0}{\lim }\frac{P(t\le T\le t+\Delta t|T\ge t,X(t),\beta )}{\Delta t}=h_0(t)\exp (\beta \text{'}{X}_t),$$

(3)

where X_t is a set of observables as well as possibly time-varying independent variables, β is a vector of unknown parameters associated with the independent variables, h₀(t) is the baseline hazard function, and $\exp (\beta \text{'}{X}_t)$ is selected because it is non-negative and generates an attractive explanation for the coefficients β. The logarithm of h₀(t, X(t), β) is linear in X_t. Thus, β reflects the partial effect of each variable in X on the logarithm of the estimated hazard rate.

The baseline hazard h₀(t) determines the shape of the hazard function relative to time. Equation (3) can be estimated without specifying the functional form of the baseline hazard. The model of partial likelihood that estimates β on the basis of the ordering of the duration spells is proposed [44]. This study refers to the Cox partial likelihood model as “semi-parametric” because it does not specify the shape of h₀(t). In addition, the most commonly used parametric specifications for the baseline hazard is the Weibull distribution [45,46,47]. The Weibull specification assumes $h_0(t)=\lambda {\alpha }_{}{t}^{\alpha -1}$, and the shape parameter (α) captures the behavior of the hazard over time. When α > 1 (α < 1), there is a positive (negative) duration dependence in the data. Specifically, a positive duration dependence in the whole duration implies that the instantaneous probability of the successful survival of distressed firms increases over time. Hence, using the maximum likelihood method, this study estimated hazard functions selecting to employ the Weibull specification. Furthermore, the present study also employed the Cox’s model of the partial likelihood [44] to check the robustness of the results obtained from the Weibull model. On the basis of hypotheses development and methodology description, Figure 1 presents the research framework and hypotheses.

4. Empirical Results

4.1. Sample Distribution, Descriptive Statistics, and Correlation Analysis

Table 1. Distribution of the sample by year.

Panel A: Distribution of the Sample by Year of Firm Survival (n = 309)
Year		Distressed Firms (and %)				Surviving Firms (and %)			Successful Survival (%)
2003		13 (4.21)				10 (4.59)			76.92
2004		15 (4.85)				11 (5.05)			73.33
2005		20 (6.47)				15 (6.88)			75.00
2006		22 (7.12)				17 (7.80)			77.27
2007		27 (8.74)				17 (7.80)			62.96
2008		40 (12.94)				25 (11.47)			62.50
2009		30 (9.71)				18 (8.26)			60.00
2010		17 (5.50)				12 (5.50)			70.59
2011		21 (6.80)				16 (7.34)			72.73
2012		19 (6.15)				14 (6.42)			73.68
2013		20 (6.47)				14 (6.42)			70.00
2014		14(4.53)				11 (5.05)			78.57
2015		18 (5.83)				14 (6.42)			77.78
2016		19 (6.15)				14 (6.42)			73.68
2017		14 (4.53)				10 (4.59)			76.92
Total		309 (100.00)				218 (100.00)			70.55
Panel B: Distribution of Firms by Years Required to Return to Profitability (n = 218)
		Years Required to Eliminate Distress
Year	Firms	3	4	5	6	7	8	9	10	11	Avg. Length for Recovery (Year)
2003	10	5	2	1	2	0	0	0	0	0	4.00
2004	11	6	2	0	1	1	1	0	0	0	4.27
2005	15	3	3	1	3	2	2	1	0	0	5.53
2006	17	1	2	2	4	3	2	3	0	0	6.41
2007	17	1	1	2	5	1	3	2	1	1	6.88
2008	25	1	2	2	6	3	3	5	2	1	7.12
2009	18	2	3	2	4	1	1	3	2	0	6.33
2010	12	6	2	1	0	2	0	0	1	0	4.58
2011	16	7	3	1	2	2	0	1	0	0	4.56
2012	14	5	3	2	0	1	1	2	0	0	5.00
2013	14	6	4	1	1	0	2	0	0	0	4.36
2014	11	5	2	1	2	1	0	0	0	0	4.27
2015	14	5	2	4	3	0	0	0	0	0	4.36
2016	14	5	4	5	0	0	0	0	0	0	4.00
2017	10	6	4	0	0	0	0	0	0	0	3.40
Total	218	64	39	25	33	17	15	17	6	2	5.25

illustrates the distribution of the sample by year. On the basis of the distribution of the sample by year of distressed firms in Panel A, a large number of distressed firms (i.e., 97 firms and 31.39%) were found in the years 2007 to 2009 in particular. This is consistent with the timing of the global financial crisis (in 2007–2008) and the subsequent European sovereign debt crisis (in 2009), as it is expected that there would be more financially distressed firms during severe macroeconomic conditions. Among 309 marginally distressed firms, 218 firms experienced their financial distress and then subsequent survival. However, in the global economic recession, the average survival rate of 61.82% (i.e., 60 surviving firms in the period over 2007–2009) was lower than the overall average of 70.55%.

Panel B reports the distribution of surviving firms by year of initial distress and the years required for the firms to return to profitability. As is known, the years required to recover to profitability ranges from 3 years to 11 years, and a substantial proportion of firms recovered to profitability within three or four years (103 firms for 47.25%), but for 26.15% of the firms (57 firms), it took seven or more years to return to profitability. Moreover, the average length of all 218 firms that experienced financial distress and then recovered to profitability was 5.25 years, ranging from 3.40 years to 7.12 years by the individual year, presenting considerable differences in the time needed for the firm managers in identifying strategies for survival.

Table 2. Descriptive statistics and correlations.

Variables	Mean	St. Dev.	1	2	3	4	5	6	7	8	9	10	11	12	13	14
1. Dependent variable (Survival = 1)	0.71	0.46	1.00
2. ΔOCFadequacy1	1.08	1.61	0.41	1.00
3. ΔOCFadequacy2	1.27	1.37	0.35	0.84	1.00
4. ΔOCFstability	0.74	1.04	0.44	0.40	0.30	1.00
5. ΔOCFgrowth	0.11	0.61	0.42	0.28	0.25	0.20	1.00
6. Profitability	1.36	1.06	0.20	0.09	0.10	0.17	0.21	1.00
7. Liquidity	2.19	1.63	0.19	0.11	0.07	0.13	0.13	−0.02	1.00
8. Leverage	0.42	0.17	−0.12	−0.05	−0.01	−0.10	−0.16	−0.08	−0.48	1.00
9. Growth	0.28	0.21	0.33	0.11	0.08	0.03	0.32	−0.07	0.28	−0.18	1.00
10. Firm size	14.59	1.65	0.19	0.10	0.08	0.19	−0.05	0.29	−0.28	0.31	−0.12	1.00
11. Recession	0.31	0.46	0.06	0.02	−0.01	−0.07	−0.10	0.02	−0.03	0.06	0.03	0.08	1.00
12. Industry1	0.09	0.29	−0.09	−0.05	−0.04	−0.01	−0.17	−0.05	−0.08	0.05	−0.09	0.13	0.01	1.00
13. Industry2	0.17	0.38	−0.05	0.05	0.01	−0.00	0.01	−0.03	0.03	−0.07	0.04	−0.11	0.02	−0.02	1.00
14. Industry3	0.13	0.34	−0.11	−0.04	−0.02	−0.08	−0.02	−0.02	0.01	0.16	0.15	0.17	0.17	−0.09	−0.18	1.00

provides means, standard deviations, and correlations of variables for the full dataset used in our analysis. As is seen in this table, we found that four relevant OCF variables, including the adequacy (ΔOCF_adequacy1 and ΔOCF_adequacy2), stability (ΔOCF_stability), and growth (ΔOCF_growth) of changes in OCF, exhibited considerable heterogeneity because their standard deviations were larger than their means. All other variables did not show much heterogeneity. The correlations of all variables are also shown in Table 2. In particular, there was a strong positive correlation of 0.84 between the adequacy of changes in OCF on total assets (ΔOCF_adequacy1) and the adequacy of changes in OCF on total labilities (ΔOCF_adequacy2) because in general, a larger ratio of OCF to total assets will also have a larger ratio of OCF to total liabilities for a firm. We therefore separately match these two variables in empirical models to avoid multicollinearity. All other variables exhibited moderate Pearson correlations, ranging from −0.48 to 0.44.

4.2. Univariate Analysis

Table 3. Differences in mean (median) of independent variables.

Variables	Mean (Median)			Differences of Surviving and Non-Surviving Firms
	Surviving Firms	Non-Surviving Firms	Differences	t-Test t-St. (p-Value)	Wilcoxon Test z-St. (p-Value)
ΔOCFadequacy1	1.77 (1.53)	−0.57 (−0.27)	2.34 (1.80)	7.91 (0.00 ***)	7.42 (0.00 ***)
ΔOCFadequacy2	2.02 (1.47)	−0.53 (−0.11)	2.55 (1.58)	6.43 (0.00 ***)	6.70 (0.00 ***)
ΔOCFstability	1.13 (0.99)	−0.22 (−0.12)	1.35 (1.11)	8.59 (0.00 ***)	8.95 (0.00 ***)
ΔOCFgrowth	0.54 (0.46)	−0.92 (−0.63)	1.46 (1.09)	7.98 (0.00 ***)	7.87 (0.00 ***)
Profitability	2.16 (2.45)	−0.54 (1.20)	2.70 (1.25)	3.63 (0.00 ***)	4.09 (0.00 ***)
Liquidity	2.38 (1.75)	1.71 (1.64)	0.67 (0.11)	3.36 (0.00 ***)	3.64 (0.00 ***)
Leverage	0.41 (0.42)	0.45 (0.47)	−0.04 (−0.05)	−2.16 (0.03 **)	−2.18 (0.03 **)
Growth	0.41 (0.22)	−0.03 (−0.03)	0.44 (0.25)	6.21 (0.00 ***)	8.98 (0.00 ***)
Firm size	14.79 (14.78)	14.11 (14.19)	0.68 (0.59)	3.33 (0.00 ***)	3.34 (0.00 ***)
Recession	0.33 (0.00)	0.26 (0.00)	0.07 (0.00)	1.15 (0.19)	1.07 (0.28)
Industry1	0.07 (0.00)	0.13 (0.00)	−0.06 (0.00)	−1.63 (0.10)	−1.62 (0.10)
Industry2	0.16 (0.00)	0.08 (0.00)	0.08 (0.00)	1.87 (0.06 *)	1.86 (0.06 *)
Industry3	0.16 (0.00)	0.20 (0.00)	−0.04 (0.00)	−0.89 (0.37)	−0.90 (0.37)
Sample size	218	91

***, **, and * represent significance at the 1%, 5%, and 10% levels, respectively.

documents the differentiated analysis of the samples. We divided the sample into two groups, i.e., 218 surviving firms and 91 non-surviving firms. Afterwards, we implemented the two-sample mean comparison test for these two groups of samples, finding that the adequacy, stability, and growth of changes in OCF were significantly distinct between the two groups at the 1% level. These differences showed that the adequacy, stability, and growth of changes in OCF were higher in surviving firms than those in non-surviving firms. This result suggests that the surviving firms with the adequacy, stability, and growth of cash flows from operation increased the likelihood of survival for distressed firms, which was consistent with H1. For surviving firms, the selected firm characteristics such as profitability, liquidity, growth, and size were significantly greater than those in non-surviving firms (at the 1% level), except for the leverage variable, which was significantly smaller than that in non-surviving firms (at the 5% level). In addition, the one of industry dummies, i.e., construction industry, presented significant differences between the two groups only at the 10% level. Accordingly, a nonparametric Wilcoxon signed-rank test showed similar results.

4.3. Multiple Regression Results

There were two distinct estimation methodologies used in analyzing the determinants of the likelihood and duration of successful survival for marginally distressed firms. First, we employed logistic regression models to illustrate how influencing variables affect the probability of firm survival from financial losses. Next, the duration model was used to determine how variables affect the length of time from which a firm survives from a loss position to a profitable position.

4.3.1. Determinants of Survival for Marginally Distressed Firms

The analysis in Section 4.2 has presented several differences between surviving and non-surviving firms in univariate comparisons. We further estimated logistic regression models where the dependent variable was equal to 1 if a firm survived from financial distress and 0 otherwise.

Table 4. Determinants of the likelihood of survival for marginally distressed firms—logistics regression.

Variables	(1)	(2)	(3)
Intercept	−9.154 *** (2.585)	−9.304 *** (2.559)	−8.913 *** (2.572)
Changes in OCF to total assets (ΔOCFadequacy1)	0.306 *** (0.114)
Changes in OCF to total liabilities (ΔOCFadequacy2)		0.194 ** (0.092)
Intensity of changes in OCF (ΔOCFintensity)			0.564 *** (0.210)
Stability of changes in OCF (ΔOCFstability)	0.863 *** (0.245)	0.954 *** (0.243)	0.895 *** (0.245)
Growth of changes in OCF (ΔOCFgrowth)	0.404 *** (0.150)	0.396 *** (0.145)	0.397 *** (0.147)
Firm profitability	0.054 (0.040)	0.052 (0.038)	0.053 (0.039)
Firm liquidity	0.492 ** (0.236)	0.515 *** (0.255)	0.499 ** (0.253)
Firm leverage	−0.523 (1.399)	−0.408 (1.380)	−0.508 (1.389)
Firm growth	4.152 *** (0.851)	4.188 *** (0.842)	4.157 *** (0.846)
Firm size	0.562 *** (0.166)	0.564 *** (0.164)	0.564 *** (0.165)
Recession dummy	0.631 (0.439)	0.705 (0.436)	0.673 (0.439)
Industry1 dummy	−0.499 (0.636)	−0.548 (0.636)	−0.529 (0.637)
Industry2 dummy	0.437 (0.680)	0.303 (0.653)	0.365 (0.662)
Industry3 dummy	−0.589 (0.543)	−0.573 (0.541)	−0.584 (0.542)
Sample size	309	309	309
Pseudo R-sq.	0.530	0.526	0.530

Numbers in the parentheses are standard errors. ***, ** represent significance at the 1% and 5% levels, respectively.

reports the results of logistic regressions. The estimated coefficients on the adequacy of changes in OCF (ΔOCF_adequacy1 and ΔOCF_adequacy2) were positive (0.306 and 0.194) and significant (at the 1% and 5% levels) in regressions (1) and (2), respectively. Additionally, in the regressions (1) and (2), the coefficients on the stability (ΔOCF_stability) and growth (ΔOCF_growth) of changes in OCF were highly significant and with correct signs, respectively. Since there was a high correlation (0.84) between ΔOCF_adequacy1 and ΔOCF_adequacy2, this study built up a principal component, namely, strength of changes in OCF (ΔOCF_strength), to deal with the problem of multicollinearity. In regression (3), the coefficient on the strength of changes in OCF (ΔOCF_strength) still showed a highly positive and significant effect on the likelihood of survival for distressed firms. These results indicated that a distressed firm with the adequacy, strength, stability, and growth of incremental cash flows from operation significantly increased the likelihood of distressed firms returning to profitability for survival.

Among the firm characteristic variables, the liquidity and growth of firms were positively and significantly related to the survival firms, implying that the distressed firms with a higher short-term liquidity and a greater sales growth before the onset of financial distress were more likely to survive. However, in regressions (1)–(3), the profitability and leverage of firms were insignificant in spite of their coefficients with the correct signs. A possible cause for this result was that the liquidity effect probably overshadowed the leverage effect on the likelihood of survival due to a slightly high negative correlation (−0.48) between each other. Furthermore, the coefficient of firm size was positive and significant, which implies that a larger distressed firm has a higher incidence of survival while suddenly suffering financial distress because it is more likely to be economically viable [2]. Although we also controlled for the recession dummy and industry dummies, these factors were not significant.

Overall, our empirical evidence suggests that a distressed firm is more likely to return to profitability for survival if it can generate abundant operating cash flow after suddenly encountering financial distress, consistent with H1. Thus, two arguments are worth noting here. First, cash flows from operation provide the important information for the correct classification of a sample of distressed firms between survival firms and non-survival ones. Second, a distressed firm’s survival depends on its ability to return to profitability, which is strongly associated with its incremental magnitude of OCF.

4.3.2. Factors of Duration in Firm Survival

This study took a further step to look into the time dependence of the firms’ survival after post-distress among 218 surviving firms. The dependent variable was duration of time (in years) that elapsed between the year of the distress in which the firm had negative operating profits for two consecutive years and the year of the survival in which the firm returned to positive operating profits for at least three consecutive years. The independent variables were the same as in the preceding logistic regression model. The signs of coefficients for independent variables were expected to be opposite to those of the logistic regressions because the dependent variable (i.e., the length of time required to return to profitability by years) is an inverse measure of successful survival: the quicker the survival, the shorter the time for the firm to return to profitability. Following the previous empirical studies [45,47,48], the specification of the duration model was run by a parametric survival model with the Weibull distribution. In addition, we also adopted the Cox semi-parametric model to check the robustness of the results obtained by the Weibull model.

In

Table 5. Factors of the duration of survival for marginally distressed firms.

Variables	Weibull Model			Cox’s Model
	(1)	(2)	(3)	(4)	(5)	(6)
Intercept	2.025 *** (0.280)	2.010 *** (0.277)	2.044 *** (0.279)
Changes in OCF to total assets (ΔOCFadequacy1)	−0.027 *** (0.011)			1.064 ** (0.032)
Changes in OCF to total liabilities (ΔOCFadequacy2)		−0.022 *** (0.007)			1.049 *** (0.016)
Intensity of changes in OCF (ΔOCFintensity)			−0.051 ** (0.222)			1.120 ** (0.054)
Stability of changes in OCF (ΔOCFstability)	−0.053 *** (0.020)	−0.056 *** (0.020)	−0.055 ** (0.021)	1.121 ** (0.058)	1.132 ** (0.057)	1.127 ** (0.056)
Growth of changes in OCF (ΔOCFgrowth)	−0.042 ** (0.017)	−0.039 ** (0.015)	−0.042 *** (0.016)	1.093 ** (0.040)	1.098 *** (0.038)	1.103 *** (0.039)
Firm profitability	−0.012 ** (0.005)	−0.010 ** (0.005)	−0.013 *** (0.004)	1.026 ** (0.012)	1.023 * (0.012)	1.028 ** (0.010)
Firm liquidity	−0.042 *** (0.013)	−0.041 *** (0.013)	−0.040 *** (0.013)	1.088 *** (0.031)	1.087 *** (0.032)	1.081 *** (0.031)
Firm leverage	0.067 (0.214)	0.069 (0.213)	0.056 (0.214)	0.995 (0.776)	0.881 (0.693)	0.853 (0.520)
Firm growth	−0.106 *** (0.031)	−0.101 *** (0.032)	−0.099 *** (0.031)	1.261 *** (0.089)	1.247 *** (0.091)	1.237 *** (0.088)
Firm size	−0.001 (0.021)	−0.001 (0.020)	−0.006 (0.021)	1.032 (0.995)	1.002 (0.993)	1.010 (0.046)
Recession dummy	0.130 *** (0.050)	0.125 ** (0.050)	0.125 ** (0.050)	0.731 ** (0.092)	0.743 ** (0.093)	0.742 ** (0.092)
Industry1 dummy	0.111 (0.097)	0.107 (0.096)	0.115 (0.093)	0.801 (0.826)	0.826 (0.838)	0.789 (0.775)
Industry2 dummy	0.092 (0.078)	0.079 (0.077)	0.077 (0.076)	0.840 (0.530)	0.864 (0.549)	0.867 (0.513)
Industry3 dummy	0.080 (0.065)	0.075 (0.064)	0.083 (0.066)	0.834 (0.569)	0.845 (0.657)	0.833 (0.628)
Scale (σ)	1.099 *** (0.048)	1.087 *** (0.046)	1.097 *** (0.047)
Log likelihood	−91.339	−90.382	−92.820	−972.371	−971.645	−982.770
Sample size	218	218	218	218	218	218

Numbers in parentheses are standard errors. ***, **, and * represent significance at the 1%, 5%, and 10% levels, respectively.

, we present coefficient estimates of β, with a positive β indicating a longer time spent in returning to profitability from loss. In regressions (1) and (2), the results show that the adequacy (ΔOCF_adequacy1 and ΔOCF_adequacy2), stability (ΔOCF_stability), and growth (ΔOCF_growth) of changes in OCF significantly shortened the time needed for a distressed firm returning to profitability. In addition, in regression (3), the evidence shows that the strength of incremental OCF (ΔOCF_strength) also significantly reduced the time needed to achieve survival for a distressed firm. Hence, the greater the adequacy, strength, stability, and growth of OCF in the distressed episode, the shorter the time required for a distressed firm returning to profitability for survival, consistent with H2. For the firm characteristics, the better profitability, liquidity, and growth of firms significantly shortened the time required to terminate the distressed episode, implying that firms with higher performance in the pre-distress period are better equipped to shorten the duration of the distress.

However, the economic recession significantly lengthened the time of the firm’s distressed period and slowed the ultimate survival, as presented in regressions (1)–(3), suggesting that macroeconomic contractions may aggravate the financial deterioration of financially distressed firms [49]. However, the leverage and size of firms turned out to be insignificant in this duration analysis, although they had the correct signs. Overall, our duration analysis results suggest that firms generated more OCF in the post-distress period and possessed higher profitability, liquidity, and sales growth in the pre-distress period; moreover, in better macroeconomic conditions, they took significantly less time to resolve financial distress for survival. In regressions (4)–(6), the hazard ratios estimated from the Cox proportional hazards model were similar in magnitude and significance to the coefficients estimated from the Weibull model. Therefore, the Cox’s model estimator provided estimates that were robust to the Weibull model for the duration analysis.

Our empirical results provide practical implications for managers, investors, policymakers, and lenders who intend to promote firm financial sustainability and performance. The identification of factors that determine the probability of success or failure and the speed of recovery would be a useful tool for managers making decisions to turnaround distressed firms in the future. Knowledge of the associated explanatory determinants of firm survival will be useful to investors in judging whether and when to buy or sell positions on loss making firms, consistent with the implications that are documented by a previous study [2]. This analysis also has implications for strategy making to policymakers and potential benefit to lenders.

4.4. Robustness Test

In this study, we only traced the sample firms until to year 2021. However, at that truncated point in time, a lot of distressed firms whose operating profits have never returned to positive profits for at least three consecutive years are not survival firms and need to be continuously observed. Thus, our research data belongs to right-censoring data. The present study further employed the Tobit censored regression [50] to be helpful for checking the robustness of the results estimated by the logistic regressions.

Table 6. Determinants of the likelihood of survival for marginally distressed firms—Tobit censored regression.

Variables	(1)	(2)	(3)
Intercept	−1.386 ** (0.704)	−1.433 ** (0.705)	−1.304 * (0.704)
Changes in OCF to total assets (ΔOCFadequacy1)	0.088 *** (0.027)
Changes in OCF to total liabilities (ΔOCFadequacy2)		0.057 *** (0.021)
Intensity of changes in OCF (ΔOCFintensity)			0.166 *** (0.052)
Stability of changes in OCF (ΔOCFstability)	0.218 *** (0.052)	0.244 *** (0.052)	0.225 *** (0.052)
Growth of changes in OCF (ΔOCFgrowth)	0.208 *** (0.047)	0.207 *** (0.047)	0.205 *** (0.047)
Firm profitability	0.005 (0.010)	0.004 (0.010)	0.005 (0.011)
Firm liquidity	0.151 ** (0.076)	0.159 ** (0.077)	0.153 ** (0.076)
Firm leverage	−0.229 (0.387)	−0.222 (0.389)	−0.233 (0.388)
Firm growth	1.099 *** (0.209)	1.133 *** (0.211)	1.108 *** (0.209)
Firm size	0.147 *** (0.046)	0.149 *** (0.058)	0.148 *** (0.046)
Recession dummy	0.166 (0.132)	0.189 (0.133)	0.177 (0.182)
Industry1 dummy	−0.172 (0.182)	−0.181 (0.184)	−0.179 (0.182)
Industry2 dummy	0.164 (0.207)	0.118 (0.207)	0.138 (0.206)
Industry3 dummy	−0.191 (0.158)	−0.189 (0.158)	−0.191 (0.158)
Sample size	309	309	309
Pseudo R-sq.	0.378	0.373	0.378

Numbers in the parentheses are standard errors. ***, **, and * represent significance at the 1%, 5%, and 10% levels, respectively.

documents the determinants of the likelihood of survival for marginally distressed firms by the Tobit censored regressions. In regressions (1)–(3), the empirical evidence illustrates that the indicators of OCF, such as the adequacy, strength, stability, and growth of changes in OCF, as well as the liquidity, growth, and size of firms have statistically significant positive effects on the likelihood of survival for distressed firms. These results suggest that the firms with the greater amount of incremental OCF, better performance at liquidity and growth, and larger size might increase the likelihood of survival when they suddenly encounter financial distress. This estimated result confirms the results of the logistics regression earlier reported in Table 4.

In order to check the robustness of our estimation results, we further replaced the OCF measures by using a number of alterative OCF indicators, such as the amount of changes in OCF to equities (ΔOCF_equity), the amount of changes in OCF to net income (ΔOCF_{net income}), and the amount of changes in OCF to sales (ΔOCF_sales), which were commonly employed in many prior studies [4,6]. Moreover, we alternatively adopted a five-year window to recalculate the stability of changes in OCF (ΔOCF_{stability_5yrs}) and the growth of changes in OCF (ΔOCF_{growth_5yrs}). We re-estimated these five OCF indicators for the robustness test by using the logistic regression model and the duration analysis model.

The estimated results are presented in

Table 7. Robustness test: alternative indicators of OCF.

Panel A: Determinants of the Likelihood of Survival for Marginally Distressed Firms—Logistic Regression
Variables	(1)		(2)		(3)
Intercept	−8.262 *** (2.552)		−8.435 *** (2.658)		−8.893 *** (2.659)
Changes in OCF to equitis (ΔOCFequity)	0.209 ** (0.083)
Changes in OCF to net income (ΔOCFnet income)			0.181 ** (0.073)
Changes in OCF to sales (ΔOCFsales)					0.147 ** (0.072)
Stability of changes in OCF (ΔOCFstability_5yrs)	0.852 *** (0.284)		0.984 *** (0.292)		0.851 *** (0.287)
Growth of changes in OCF (ΔOCFgrowth_5yrs)	0.442 *** (0.135)		0.479 *** (0.133)		0.478 *** (0.134)
Firm profitability	0.028 (0.029)		0.022 (0.028)		0.019 (0.029)
Firm liquidity	0.465 ** (0.217)		0.554 ** (0.225)		0.512 ** (0.218)
Firm leverage	−0.160 (1.519)		−0.164 (1.428)		−0.117 (1.517)
Firm growth	4.181 *** (0.976)		4.463 *** (0.957)		4.488 *** (1.131)
Firm size	0.505 *** (0.157)		0.498 *** (0.169)		0.547 *** (0.164)
Recession dummy	0.435 (0.396)		0.467 (0.383)		0.410 (0.397)
Industry1 dummy	−0.684 (0.610)		−0.830 (0.662)		−0.840 (0.669)
Industry2 dummy	0.456 (0.609)		0.498 (0.558)		0.327 (0.601)
Industry3 dummy	−0.298 (0.433)		−0.339 (0.412)		−0.321 (0.422)
Sample size	309		309		309
Pseudo R-sq.	0.466		0.468		0.466
Panel B: Factors of the Duration of Survival for Marginally Distressed Firms
Variables	Weibull Model			Cox’s Model
	(1)	(2)	(3)	(4)	(5)	(6)
Intercept	2.090 *** (0.262)	2.080 *** (0.266)	2.076 *** (0.261)
Changes in OCF to equities (ΔOCFequity)	−0.024 ** (0.012)			1.087 ** (0.027)
Changes in OCF to net income (ΔOCFnet income)		−0.021 ** (0.011)			1.069 ** (0.021)
Changes in OCF to sales (ΔOCFsales)			−0.026 ** (0.014)			1.044 ** (0.013)
Stability of changes in OCF (ΔOCFstability_5yrs)	−0.071 * (0.038)	−0.072 * (0.038)	−0.065 * (0.040)	1.183 ** (0.100)	1.189 ** (0.102)	1.171 * (0.104)
Growth of changes in OCF (ΔOCFgrowth_5yrs)	−0.043 *** (0.015)	−0.044 *** (0.015)	−0.040 *** (0.015)	1.096 *** (0.038)	1.099 *** (0.038)	1.091 ** (0.037)
Firm profitability	−0.010 ** (0.005)	−0.011 ** (0.005)	−0.012 ** (0.006)	1.011 ** (0.011)	1.011 ** (0.011)	1.011 ** (0.011)
Firm liquidity	−0.045 *** (0.014)	−0.045 *** (0.013)	−0.044 *** (0.014)	1.096 *** (0.034)	1.098 *** (0.034)	1.095 *** (0.034)
Firm leverage	0.046 (0.237)	0.045 (0.238)	0.047 (0.235)	0.970 (0.605)	0.978 (0.611)	0.946 (0.608)
Firm growth	−0.095 *** (0.032)	−0.095 *** (0.031)	−0.101 *** (0.031)	1.231 *** (0.090)	1.227 *** (0.088)	1.243 *** (0.091)
Firm size	−0.011 (0.020)	−0.010 (0.020)	−0.010 (0.020)	1.016 (0.045)	1.016 (0.044)	1.013 (0.044)
Recession dummy	0.132 *** (0.050)	0.132 *** (0.051)	0.134 *** (0.050)	0.730 ** (0.091)	0.731 ** (0.093)	0.727 ** (0.091)
Industry1 dummy	0.133 (0.084)	0.130 (0.085)	0.129 (0.086)	0.740 (0.147)	0.741 (0.148)	0.747 (0.153)
Industry2 dummy	0.078 (0.070)	0.079 (0.070)	0.082 (0.073)	0.859 (0.139)	0.856 (0.138)	0.856 (0.143)
Industry3 dummy	0.062 (0.069)	0.065 (0.072)	0.067 (0.069)	0.876 (0.140)	0.876 (0.143)	0.866 (0.137)
Scale (σ)	1.077 *** (0.043)	1.078 *** (0.043)	1.080 *** (0.444)
Log likelihood	−99.137	−99.122	−98.406	−986.316	−986.347	−985.892
Sample size	218	218	218	218	218	218

Numbers in the parentheses are standard errors. ***, **, and * represent significance at the 1%, 5%, and 10% levels, respectively.

. As expected, the estimated coefficients of all alternative OCF indicators, as well as the liquidity, growth, and size of the firm were still significant and positive in the regressions (1)–(3), as the results show in Panel A. In addition, for the estimated results presented in Panel B, the evidence shows that alternative OCF indicators and the profitability, liquidity, and growth of the firm reduced the time needed to achieve survival. However, the economic recession extended the time of the distress episode and delayed an eventual survival, as presented in regressions (1)–(6) of Panel B in Table 7. The conclusion hence remained unchanged as those drawn from the preceding logistic regression of Table 4 and the Weibull and Cox’s models in survival analysis of Table 5. Overall, the results of the robustness test support our earlier findings that the effects of OCF significantly account for the success of survival and the time needed for recovery in financial distress, regardless of if the OCF indicators were alternatively proxied by either of the other alternative indicators of OCF.

5. Conclusions

In this paper, we investigated whether the effect of OCF affected the likelihood and duration of firm survival using 309 marginally distressed firms, of which 218 were survival firms and 91 were non-survival firms. To the best of our knowledge, we believe that the study firstly proposes an innovative cash-flow-based approach that provides new insights into the economic effects of three OCF indicators as well as the adequacy, stability, and growth of cash flows from operations on the likelihood and the duration of survival for the distressed firms.

Our main findings identify the economic factors that significantly affect the likelihood and the duration of a distressed firm’s survival. Specifically, it is not only the adequacy, stability, and growth of the OCF but also the liquidity, growth, and size of the firm that are important determinants in the likelihood of firm survival from financial distress. Likewise, with more OCF generated during the post-distress period, and better profitability, liquidity, and growth in the pre-distress period, distressed firms then take significantly less time resolving financial distress to survive. However, a macroeconomic contraction significantly lengthens the time needed for a distressed firm to successfully survive. Overall, the cash flow from operations is crucial to the successful survival of financially distressed firms, suggesting that our OCF indicators are reliable tools to predict the likelihood and the duration of distressed firms’ survival in the context of Taiwan.

Our research makes contributions to extant literature by providing additional information to expand the validity of the flow-based theory that is employed in estimating the survival of financially distressed firms. Many prior studies have found rich evidence that shows a negative significant relationship between cash flow ratios and financial distress. However, this study exploited the characteristics of OCF in predicting the likelihood and duration of survival for financially distressed firms. Our findings document that the greater the adequacy, stability, and growth of OCF, the higher the likelihood that distressed firms survive, and also the shorter the recovery time for surviving firms. Moreover, our results bear important practical implications for managers, investors, policymakers, and lenders who intend to promote firm sustainability and performance in the future.

Financ.ally, although this study was limited to a sample of Taiwan listed firms due to data availability, our findings call for further research on how much additional explanatory power can be attributed to the impact of OCF for explaining the likelihood and duration of firm survival when controlling for the essentially strategic actions of distressed firms.