Search in book...
Toggle Font Controls
Create new playlist

Name your new playlist

Playlist description (optional)
Sign In

Email address

Password

Forgot Password?

or

Continue with Facebook

Continue with Google
Sign Up

Full Name

Email address

Confirm Email Address

Password

or

Continue with Facebook

Continue with Google

Chapter 11

Is Bankruptcy a Systematic Risk? Evidence From Vietnam

T. Chaiyakul^*

K. Bangassa^**

M. Iskandrani^†
^*    Kasetsart University Sriracha Campus, Faculty of Management Sciences, Chonburi, Thailand
^**    University of Liverpool, Management School, Liverpool, United Kingdom
^†    University of Jordan, Faculty of Business, Amman, Jordan

Abstract

There is conflicting evidence reported in the existing academic literature regarding the pricing of bankruptcy risk. Also, existing studies concentrate on developed and emerging markets (to some extent), but not on frontier markets. We aim to investigate the pricing of bankruptcy risk by using the DLI (default likelihood indicator) bankruptcy estimation model developed by Vassalou and Xing [Vassalou, M., Xing, Y., 2004. Default risk in equity returns. J. Finance 59 (2), 831–868]. The most recent global financial crisis (GFC) was chosen as an ideal test period for this study. Firms listed on the Ho Chi Minh City Stock Exchange in Vietnam during the precrisis period of Jul. 1, 2007 to Sep. 14, 2008; the crisis period of Sep. 15, 2008 to Mar. 31, 2009; and the postcrisis period of Apr. 1, 2009 to Dec. 31, 2010 were analyzed. The DLI bankruptcy risk measures were assessed using both regression and mimicking portfolio analysis. We observed a significantly positive relationship between stock returns and bankruptcy risk with portfolio analysis, by conducting analysis in which the portfolios were formed using a medium BM (book-to-market) factor and controlling for DLI-sorted stocks. As expected, we also found a significant relationship between stock returns and bankruptcy risk during the post-GFC period, as measured by the average default likelihood indicator (ADLI) developed by Vassalou and Xing [Vassalou, M., Xing, Y., 2004. Default risk in equity returns. J. Finance 59 (2), 831–868].

Keywords

bankruptcy risk

equity returns

Vietnamese stock market

global financial crisis

JEL classification

G11

G12

G13

G14

G33

1. Introduction

The Vietnamese stock market is an MSCI frontier market. Vietnam has been identified as an investable economy in need of more capital and market liquidity, as opposed to a more developed emerging market. The Vietnamese stock market has been growing aggressively. According to the World Bank, the average Vietnamese stock market capitalization as a percentage of GDP for the period from 2003 to 2012 was 12.25%, with a minimum of 0.36% in 2003 and a maximum of 25.24% in 2007. As a rule of thumb, stock market capitalization of 50% or more as a percentage of GDP indicates a developed market. According to HSBC (2012), the HOSE (Ho Chi Minh City Stock Exchange) is skewed toward financials, whereby 50 companies operate in the financial sector with more than 49% market capitalization, followed by consumer staples (19%), utilities (12%), and materials (6%).

Vietnam is ruled by one political party. The Vietnamese equities market consists of two major stock exchanges, the HOSE and the Hanoi Stock Exchange (HNX). The HOSE was the first stock exchange in Vietnam; it was started in 2000 with the government’s decision to privatize state-owned enterprises (SOEs) and support the raising of capital needed for the economic development of Vietnam. The HOSE has grown up rapidly. In 2006 the HOSE had 34 listed companies with a market capitalization of USD 1.1 billion. In 2012 there were 303 listed companies with a market capitalization of USD 29.9 billion. The HOSE is larger than HNX. The indices tracking the two markets are the VN-INDEX and HNX-INDEX, respectively.

Starting in Jul. 2003, foreign investors became able to invest in up to 30% of the total value of the SOEs that had been privatized. This amount was raised to 49% toward the end of 2005, with the aim of promoting the privatization process. To participate in the market, foreign investors are required to open an account in one of the local brokerage firms. There is currently a quota of 60% foreign room, which limits the maximum amount foreign investors can buy in 1 day.

The products available for trading in both markets include Vietnamese equities, and government and corporate bonds. The HOSE mainly deals with equity trading, whereas the HNX deals with equities, bonds, and over-the-counter (OTC) securities.

As mentioned previously, the first stock market was established in 2000 with the government’s decision to privatize SOEs and support raising the capital needed for the economic development of Vietnam. The Vietnamese government aims to build a market economy in Vietnam within the general framework of a socialist sociopolitical system and socialist principles. Nguyen et al. (2014) argue that privatization can be useful in the process of developing the Vietnamese stock market and that providing market access to well-established institutions is important for the long-term development of the market. They also question whether Vietnam has laid down the basic elements needed to establish a sustainable stock market. The position of the World Bank (1992) regarding the privatization which is embraced by many nations around the globe (eg, Western Europe, Latin America, Southeast Asia, Russia, Eastern Europe, Africa, etc.) is that “privatization is not a blanket solution for the problems of the poorly performing SOEs. It cannot in and of itself make up totally for lack of competition, for weak capital markets, or for the absence of an appropriate regulatory framework.”

Between 1989 and 2008 Vietnam achieved an average growth rate of 7.4% per annum. However, while the GDP per capita of Vietnam increased to US $1000 in 2008 (which indicated a very favorable achievement), according to the General Statistics Office of Vietnam, 2008 was a gloomy year for the Vietnamese economy (Pincus, 2009), and ended with a significant drop in performance compared to the achievements registered during the previous year. Pincus argues that huge capital inflows in 2007 and early 2008 accelerated economic “overheating,” increases in inflation, and the building of a significant current account deficit. The HOSE lost 66% of its value that year, becoming the worst performer at a global level. In 2008 besides domestic challenges to the economy, Vietnam was also affected by the GFC, which started with the US subprime mortgage crisis and was further intensified by the unexpected collapse of Lehman Brothers in 2008. Big economies in the world—such as those of the United States, the United Kingdom, and many euro zone countries—suffered from the financial crisis (Das and Shrestha, 2009). Increases in inflation beyond 25%, rises in the cost of housing, a significant drop in the Vietnamese dong against the US dollar, falling wages in real terms as a result of rallying inflation, employee strikes across the nation, etc., gave further evidence that the Vietnamese economy was hit hard by a national and the global financial crisis. As a result of the GFC, Vietnam’s economy was affected mainly in the areas of trade, investment, capital mobility, and the financial markets. The economy of Vietnam was directly and indirectly affected by the financial crisis and economic recession in the economies of its major trading partners, such as the United States, the European Union, Japan, etc.

Inflation figures of double digits were recorded in 2007 and reached 28.32% in Aug. 2008. Furthermore, fiscal and trade deficits of up to 5% and over 20% of GDP, respectively, were recorded. Demand for manufactured export goods, such as footwear and furniture, dropped significantly. Other areas of the Vietnamese economy which were also affected significantly included properties and construction materials, chemicals used in agriculture and industry, etc. Companies listed on the HOSE and HNX were affected by the financial crisis and as a result laid off a number of employees. Multinational corporations were globally affected by the financial crisis, and consequently the level of foreign direct investment (FDI) in Vietnam dropped significantly, leading to delays in and cancelations of investment projects.

Furthermore, investor psychology was affected by the GFC, which in turn had impacts on domestic capital markets. In 2008 the VN-INDEX and the HNX-INDEX lost 66.9 and 67.2% of their values, respectively. According to the figures reported by 41 out of 63 provinces and municipalities in Vietnam, 66,700 workers lost their jobs in 2008, increasing the unemployment rate to 4.65%. Based on these figures, it is estimated that the total number of workers who lost their jobs in 2008 due to the all-around severe effects of the global and national financial crisis on Vietnam’s economy could have been well over 80,000. The General Statistics Office of Vietnam reported a fall in the Consumer Price Index (CPI) amounting to 0.68% in Dec. 2008 compared to Nov. 2008, which was preceded by falls of 0.76 and 0.19% in Nov. and Oct. 2008, respectively. As a result of these negative impacts of the financial crisis on the Vietnamese economy, the overall economic growth of the nation declined from 8.48% in 2007 to 6.23% in 2008.

There is an ongoing and unresolved debate over whether bankruptcy risk is a stated variable in explaining equity returns. For example, Fama and French (1996) consider that the existence of anomalous size and value premiums can arise as a result of financial distress. We are going to assess the extent to which, if any, bankruptcy risk explains, in a systematic manner, equity returns above the risk premiums which can be empirically identified as important for market, size, and BM factors.^a The probability of bankruptcy reflects the extent of a firm’s financial distress, while the eventual outcome from severe financial distress can be bankruptcy. Altman (1993) provides an early review of literature on bankruptcy prediction and measures of ex ante bankruptcy risk. Accordingly, the existing evidence that bankruptcy risk is systematic is related to a distress factor explanation for the market, size, and BM effects. Griffin and Lemmon (2002) and Campbell et al. (2008) argue that it is not difficult to identify a mechanism for bankruptcy risk that can be priced if we accept that financially distressed firms have prices that comove and are difficult to diversify due to arbitrage restrictions. Lang and Stulz (1992) and Denis and Denis (1995) report evidence which implies that bankruptcy risk could be positively related to systematic risk. Shumway (1996) finds evidence suggesting that the risk of default is systematic. Vassalou and Xing (2004) also argue that default risk is systematic and therefore can be priced in the cross-section of equity returns.

On the other hand, Asquith et al. (1994) find that bankruptcy is mostly due to idiosyncratic factors, implying that it is unrelated to systematic risk. Vassalou and Xing (2004) report that debt default risk is related to the size and BM characteristics of a firm, but that the evidence regarding its impacts on equity returns is mixed and is likely to be related to the strength of shareholder bargaining power in extracting rents from competing claim holders in a distress situation (Aretz, 2011; Zhang, 2012; Garlappi et al., 2008).

Our contribution in this context can then be straightforwardly stated: we assess bankruptcy risk as a latent risk factor conditioned on economic states which provide a hypothesis for its fleeting existence. In so doing, we attempt to revisit some of the conflicting findings reported to date and offer out-of-sample evidence.

In sum, we are able to more clearly specify the relationship between bankruptcy risk and equity returns. Our test basis is the most recent GFC, which supplies a set of useful economic circumstances that provide the potential to clearly present out-of-sample evidence for the relationship between bankruptcy risk and equity returns.

We are also motivated by the recommendations of Lo and MacKinlay (1990), who argue that empirical findings should be examined out-of-sample to ensure that any findings reported are not a product of data snooping. The Vietnamese market, which is a frontier market, is tested in this study. The data set we examine is not investigated with appropriate scope in existing studies (eg, for studies of the United States see Lang and Stulz, 1992; Opler and Titman, 1994; Asquith et al., 1994; Dichev, 1998; Griffin and Lemmon, 2002; Vassalou and Xing, 2004; for studies of the United Kingdom see Hussain et al., 2001; Agarwal and Taffler, 2008).^b Hence, we examine the relationship between bankruptcy risk and equity returns by conducting both portfolio and cross-sectional analyses.

In this paper we investigate the relationship between bankruptcy risk and equity returns using the GFC and the HOSE as a test basis. This study is organized as follows: Section 2 reviews the existing literature. Section 3 describes research design, data, and methodology. Section 4 discusses the empirical evidence, and Section 5 concludes the study.

2. Literature Review

The relationship between bankruptcy risk and equity returns is examined by small group of prior researchers using different bankruptcy risk measures. For instance, Dichev (1998) employed models developed by both Altman (1968) and Ohlson (1980) to measure bankruptcy risk. Similarly, Griffin and Lemmon (2002) employ the O-score developed by Ohlson (1980) as the proxy for bankruptcy risk. The study from Vassalou and Xing (2004) is the first one to apply the model developed by Merton (1974).^c They argue that the models designed by Altman (1968) and Ohlson (1980) are naturally backward looking, since financial statements tend to present a firm’s past performance rather than its future prospects. The Merton (1974) option-pricing model uses market-value data to compute the bankruptcy risk of firms. Bystrom et al. (2005) also apply the Merton (1974) model to measure bankruptcy risk and investigate the relationship between bankruptcy risk and equity returns.

Existing studies that investigate the relationship between bankruptcy risk and equity returns report dissimilar findings. For instance, Lang and Stulz (1992) study the effect of bankruptcy announcements on the equity value of a bankrupt firm’s competitors and conclude that bankruptcy announcements decrease the value of competitors, implying a positive relationship between bankruptcy risk and equity value. Vassalou and Xing (2004) analyze the effects of bankruptcy risk on stock returns by using the Merton (1974) option-based model and conclude that bankruptcy risk in the United States is a systematic risk. More recently, Chava and Purnanandam (2010) reported a positive cross-sectional relationship between default risk and expected stock returns.

On the other hand, there are other studies that disagree with the evidence that bankruptcy risk is a systematic risk. For example, Opler and Titman (1994) study the effect of financial distress on corporate performance and report that the stock returns of more leveraged firms in distressed industries are lower than those of less leveraged firms, and hence, by implication, bankruptcy risk is not a systematic risk. Asquith et al. (1994) analyze financially distressed firms that try to avoid bankruptcy through public and private debt restructuring, asset sales, mergers, and capital expenditure reductions. They find no evidence which supports the expectation that the better performing companies in their sample are those that are more successful in dealing with financial distress, since such firms are as likely to go bankrupt as the other firms. Therefore, they argue that bankruptcy risk is an idiosyncratic factor. Dichev (1998) uses the models from Altman (1968) and Ohlson (1980) to investigate whether bankruptcy risk is a systematic risk by using all industrial firms on the New York Stock Exchange (NYSE), American Stock Exchange (AMEX), and OTC markets from 1981 to 1995, and they report that bankruptcy risk is not rewarded by higher returns. In actuality, firms with a high bankruptcy risk are reported to earn substantially lower than average returns.

Hussain et al. (2001) study the behavior of relative financial distress by applying the capital asset pricing model (CAPM) and the Fama and French (1993) three-factor model and using UK data for the period 1980–99. Their results show that there is no difference between returns of financially distressed and nonfinancially distressed firms, implying that bankruptcy risk is not systematic. Griffin and Lemmon (2002) examine the relationship between bankruptcy risk and stock returns. They employ the model from Ohlson (1980) to calculate the probability of bankruptcy and show that bankruptcy risk is negatively related to equity returns. Agarwal and Taffler (2008) use the Taffler (1983, 1984)^d bankruptcy prediction model to measure bankruptcy risk and test whether bankruptcy risk is a separately priced risk factor. Their study illustrates that financially distressed stocks earn lower returns than nonfinancially distressed stocks. Their findings are consistent with the evidence reported by Dichev (1998). More recently, Da and Gao (2010) argued that the abnormal returns on high default risk stocks found by Vassalou and Xing (2004) are mainly due to short-term return reversals rather than systematic default risk.

Gharghori et al. (2007) employ data from the Australian market during the period from Jan. 1996 to Dec. 2004 to investigate whether default risk is priced in the cross-section of equity returns. They augment the Fama and French (1993) three-factor model with their new default-risk factor (developed by applying the Merton (1974) option-based model) and report that default risk is not priced in equity returns. Bystrom et al. (2005) analyze data of 50 financially distressed companies from the Stock Exchange of Thailand. They also apply the Merton (1974) model to measure the probability of bankruptcy and argue that bankruptcy risk is not a systematic risk. Samad et al. (2009) find a significant inverse relationship between distress risk and the BM equity measure. This is a finding consistent with Lang and Stulz (1992), Denis and Denis (1995), Shumway (1996), and Chen and Zhang (1998). Furthermore, Avramov et al. (2009) find that firms with low credit risks obtain higher returns compared to high credit risk firms due to mispricing by retail investors and the inability of arbitrageurs to play a role because of prohibitive factors such as illiquidity of low-rated stocks. Also, Gharghori et al. (2009) examine Australian companies and find that default probability is negatively related to returns. They also report that the claim by Fama and French (1996) that size and BM factors explain the cross-sectional variation in equity returns is not consistent with their finding from Australian data, and they thereby recommend further research in different markets to gain more understanding about the default risk hypothesis.

Chen et al. (2010) examine the relationship between financial distress and idiosyncratic volatility. They use the Altman (1968) and Ohlson (1980) models to measure the risk of bankruptcy. Applying the Ferguson and Shockley (2003) approach and the CAPM, they show that the stocks with high volatility and high distress risks have low returns. They argue that the presence of idiosyncratic volatility, which is connected to distress risk, explains why some high distress stocks do not receive high returns, since idiosyncratic risks are not priced but diversified. Gutiérrez et al. (2012) analyze firms’ total value performance in relation to different bankruptcy systems and find that the total value of firms in financial distress is low, possibly due to a decline in efficiency. Garlappi and Yan (2011) develop and test a model that explains lower returns for financially distressed stocks, among other findings. Aretz and Shackleton (2011) also argue that asset-return correlations are more important determinants of betas than default risk, since the relation between leverage and beta bias can be diversified away by including a large number of securities in a portfolio.

Lin et al. (2012) employ data from the Taiwan stock market during the period from Jan. 1996 to Dec. 2007 to investigate the relationship between default risk and equity returns. They use both the Merton (1974) and the compound-option models^e used by Geske (1977, 1979), Delianedis and Geske (2003), and Lin and Chang (2009) to measure the default risk. They apply Fama and French (1993) three-factor model to test the influences of firm’s size, BM, and default risk on equity returns. Moreover, Lin et al. (2012) use the two-pass regression of Fama and MacBeth (1973) to test the influences of size, BM, and default factors on equity returns. The portfolio analysis reveals that both the firm’s size and its BM ratio cannot proxy the default risk. With respect to the default risk effect, they show that no significant differences exist in average returns of portfolios with high and low default risks. Nevertheless, by employing the Fama and French (1993) three-factor model they find that default risk is part of systematic risk, but that still the firm’s size and BM ratio cannot proxy for default risk.

Kim (2013) employs data from the US market during the period 1978–2007 to analyze the relationship between bankruptcy risk and subsequent returns. For bankruptcy risk he uses the O-score (Ohlson, 1980) and the BSM-prob model (Hillegeist et al., 2004). Kim (2013) argues that using two proxies of bankruptcy risk increase the possibility to detect the source of anomaly and reduce the bias of results that may occur because of differences in sample periods. His findings reveal that there is a negative relationship between the O-score and subsequent returns, whereas there is no relationship between BSM-prob and subsequent returns. In particular, he finds that funds from operations divided by the total liabilities is the only component that predicts returns, whereas the relationship between funds from operations divided by the total liabilities and subsequent returns diminishes once he controls for cash flows from operations divided by average total assets. He concludes that the funds from operations divided by total liabilities follow those of cash flows from operations divided by average total assets.

Chen and Lee (2013) test the effect of firm’s size, BM ratio, liquidity, and default risk on portfolio returns and investigate the impact of a momentum factor on portfolio returns. They use the data from the Taiwanese stock market during the period from Jan. 1986 to Dec. 2008. They follow the methodology of Vassalou and Xing (2004) and Bharath and Shumway (2008) in order to calculate a firm’s expected default frequency. For market liquidity, they use the Amihud (2002) illiquidity ratio and Pastor and Stambaugh (2003). Their results reveal three interesting findings, which are as follows:

1. The effect of default risk on equity returns exists only when they control for market liquidity; the opposite is true once they control for the firm’s size and BM ratio.

2. Under the asset pricing regression, the results show that there is a relationship between the default risk and equity returns. However, after they control for the firm’s size, BM ratio, and momentum, they notice that default risk cannot act as systemic risk in asset pricing.

3. With respect to the timing of the distress, they find that a short-term reversals in portfolios have high default risks.

Simlai (2014) analyzes the variability of average portfolio returns in relation to firm-level characteristics for US distressed firms. He follows Campbell et al. (2008) to build a financial-distress measure. In particular, Simlai (2014) studies the effect of firm size, BM ratio, and momentum on average portfolio returns with application of the Fama and French (1993) three-factor model and the Carhart (1997) four-factor model, using US data for the period 1972–2008. His results show that average excess returns for small firms are negatively associated with systematic risk and high-growth firms are negatively related to the probability of financial distress. With respect to the momentum effect, the results show that high-distressed firms have negative momentum, while the low-distressed firms have positive momentum. Simlai (2014) concludes that firm size and BM ratio effects are not associated with distress risk, and that the momentum factor is not a proxy of distress risk.

3. Research Design, Data, and Methodology

If the coefficient of the bankruptcy risk factor is statistically significant in a regression analysis, bankruptcy risk can be considered as systematic risk. Hence, the degree of the variability of returns of distressed firms with respect to asset pricing factors is expected to be higher than that for nondistressed firms, if the risk of bankruptcy is a systematic risk. This could be well demonstrated by investigating the relationship between asset returns and the measures for bankruptcy risk over a changing economic environment. The GFC created a suitable condition to investigate the significance of bankruptcy risk in asset pricing. We use the DLI bankruptcy risk measure developed by Vassalou and Xing (2004), which is an application of the Merton (1974) option-pricing model. This measure is based on market-value data, unlike the Altman (1968) and Ohlson (1980) models that use historical data from financial statements.

We apply portfolio and regression analyses to investigate the relationship between equity returns and risk of bankruptcy. In portfolio analyses, quintile portfolio returns are formed based on the DLI bankruptcy-risk measure, size-bankruptcy risk measure, and BM-bankruptcy-risk measure. We look for the presence of a monotonic pattern that displays the level of equity returns as the value of bankruptcy risk increases or decreases. Return differences between high and low bankruptcy portfolios that are controlled by firm characteristics are examined. A null hypothesis

H_{0} : {\bar{X}}_{1} - {\bar{X}}_{2} = 0

$H_{0} : {\bar{X}}_{1} - {\bar{X}}_{2} = 0$

, where

{\bar{X}}_{1}

${\bar{X}}_{1}$

is the population mean return for high bankruptcy risk portfolio and

{\bar{X}}_{2}

${\bar{X}}_{2}$

is the population mean return for low bankruptcy risk portfolio, is tested. The alternative hypothesis is

H_{1} : {\bar{X}}_{1} - {\bar{X}}_{2} \neq 0

$H_{1} : {\bar{X}}_{1} - {\bar{X}}_{2} \neq 0$

. The null hypothesis H₀ will be rejected in favor of the alternative hypothesis H₁ if the population means are statistically different. The test statistic for the difference between two mean returns is conducted following Lomax (2007).^f

In conducting regression analysis we divide the study period into precrisis, crisis, and postcrisis periods. The precrisis period is from Jul. 1, 2007 to Sep. 14, 2008; the crisis period is from Sep.15, 2008 to Mar. 31, 2009; and the postcrisis period is from Apr. 1, 2009 to Dec. 31, 2010.

The proxies of market returns, size, and BM ratio are augmented by the bankruptcy risk factors used in the cross-sectional regression analysis. In previous studies on the relationship between bankruptcy risk and equity returns—such as Griffin and Lemmon (2002), Vassalou and Xing (2004), and Gharghori et al. (2007)—the market return, size, and BM ratio factors are incorporated to investigate the explanatory power of bankruptcy risk in pricing equity returns, since these factors have shown significant ability to explain equity returns and are often used in investigations of asset pricing. Following Fama and French (1993, 1996), this study employs excess market returns and the return on a zero-net investment portfolio for size and the BM equity ratio in the cross-sectional regression analysis.

Unlike previous studies (to our knowledge), this study initially uses the ADLI as the proxy for bankruptcy risk. The reason for this is that if assets are priced rationally, bankruptcy risk must proxy as a nondiversifiable risk factor in returns. Following Vassalou and Xing (2004), the DLI formula developed from the Merton (1974) option-pricing model is shown as follows:

$D L I_{t} = N (- D D_{t})$ $D L I_{t} = N (- D D_{t})$

(11.1)

$D D_{t} = \frac{\ln (V_{A} / X_{t}) + (r - (1 / 2) σ_{A}^{2}) (T)}{σ_{A} \sqrt{T}}$ $D D_{t} = \frac{\ln (V_{A} / X_{t}) + (r - (1 / 2) σ_{A}^{2}) (T)}{σ_{A} \sqrt{T}}$

(11.2)

where DLI_t is the DLI, N is the cumulative density function of the standard normal distribution, DD_t represents the number of standard deviations that a firm deviates from the mean for bankruptcy to occur, V_A is the market value of the firm’s equity, X_t is the total amount of the firm’s debts, r is the risk-free rate, σ_A is the volatility of the firm’s asset returns, and T is the time to maturity of the firm’s debt. The larger the value of DD, the smaller the probability of bankruptcy risk; a higher DLI indicates a higher probability of bankruptcy. The ADLI is constructed from a simple average of the DLI of all firms.

The model choice follows previous studies in the literature, that is, Vassalou and Xing (2004) and Bystrom et al. (2005). Following Vassalou and Xing (2004), the daily data are aggregated in order to obtain monthly observations. Daily market values for firms (V_A) are employed, while annual data are used for the book value of debt (D_t), calculated using “short-term debt and the current portion of long-term debt” (Datastream item WC03051) plus half the “long-term debt” (Datastream item WC03251). This study includes long-term debt in the calculations because firms need to deal with their long-term debts, and these interest payments are part of their short-term liabilities. Following Vassalou and Xing (2004) and Moody’s KMV, which is a credit-rating company, this study uses 50% of long-term debt encounters in the calculations. Moody’s KMV argues that this choice is sensible and adequately captures the financing constraints of firms. Vassalou and Xing (2004) also found that having a different proportion of long-term debt included in the DLI calculations does not lead to a significant change in the results. Monthly equity volatilities (σ_A) were estimated using 12-month historical sample volatilities. The risk-free rate for each market is the Vietnamese discount rate-middle rate. Following Bystrom et al. (2005), the time to maturity of debt (T) is always assumed to be 1 year.

To investigate whether bankruptcy risk is priced in equity returns, we construct the model as follows:

$R_{t} = a + b E M K T_{t} + s S I Z E_{t} + h B M_{t} + g A D L I_{t} + ɛ_{t}$ $R_{t} = a + b E M K T_{t} + s S I Z E_{t} + h B M_{t} + g A D L I_{t} + ɛ_{t}$

(11.3)

where R_t and EMKT_t refer to the excess return of a stock and the market, respectively. The Vietnamese discount rate-middle rate is used as a proxy for the risk-free rate. The variable SIZE_t^g refers to return on the zero-investment portfolio, which is long on stocks with a small market capitalization (size) and short on large-sized stocks, as in the Fama and French (1993) model. The variable BM_t refers to the BM equity-value ratio of each stock. The variable ADLI_t is the simple average of the DLI of all firms.

The data in this study come from all firms in the HOSE for the period Jan. 2007 to Dec. 2010. All data were collected from the Datastream database. A return is the monthly average return calculated from the percentage change of the monthly return index. The monthly DLIs (Vassalou and Xing, 2004) are employed as monthly bankruptcy risk variables. Size, or the market capitalization, is the share price multiplied by the number of ordinary shares in issue. The daily market values of firms are employed to calculate the monthly average size of firms. Each month, the BM equity ratio of a firm is the last fiscal year’s book value of the equity divided by the current month’s market value of the equity. Following previous literature such as Vassalou and Xing (2004) and Griffin and Lemmon (2002), firms with a negative BM are excluded from this study because it is difficult to interpret BM portfolios. The stocks in low BM portfolios refer to those with the highest growth potential; however, many negative BM stocks face financial difficulties.

4. Results of Analysis and Discussion of Findings

In Table 11.1 stocks are sorted into five portfolios based on the Vassalou and Xing’s (2004) DLI bankruptcy risk measure. The average returns of each portfolio sorted by the bankruptcy risk measure are calculated. A high DLI bankruptcy risk measure is related to the stocks which carry a high probability of becoming bankrupt, whereas a low bankruptcy risk is related to stocks which have a low probability of becoming bankrupt. A bankruptcy risk premium (discount) prevails when the difference between the returns of the highest and lowest bankruptcy risk portfolios are positive (negative) and statistically significant. The average returns for all portfolios shown in Table 11.1 are negative over the whole study period, that is, from Jul. 1, 2007 to Dec. 31, 2010—indicating that the stocks did not perform well during this period of time. The difference in average returns between high DLI–sorted portfolios and low DLI–sorted portfolios is showing that the average returns for high DLI–sorted portfolios are lower than average returns for low DLI–sorted portfolios. However, the difference is statistically insignificant. There is no particular pattern observed in average returns of the bankruptcy risk–sorted portfolios according to the results presented in Table 11.1. We expect a monotonic pattern that displays the changes in the level of equity returns as the value of bankruptcy risk increases or decreases if there exists a relationship between bankruptcy risk and portfolio returns.

Table 11.1

Average Returns of Portfolios Sorted by Bankruptcy Risk Measures

Average returns
DLI-sorted portfolios
Low	2	3	4	High	High–low	t-values
−0.06804	−0.06221	−0.07691	−0.01777	−0.06894	−0.00089	−0.0637

Stocks are sorted into five portfolios by their levels of bankruptcy risk using DLI models (Vassalou and Xing, 2004). The average returns of each portfolio are then computed. When stocks are sorted by DLI, Portfolio 5 contains the stocks with highest bankruptcy risk. “High–low” is the return difference between the high and low bankruptcy risk portfolios. Significance at the 1 and 5% levels is indicated by ** and *, respectively.

Table 11.2 presents average returns of quintile portfolios, which are sorted by the DLI bankruptcy risk measure while being controlled for the size factor (small, medium, and large). Size is represented by the market capitalization of the stocks. Therefore, fifteen bankruptcy-size sorted portfolios are formed. Fourteen out of fifteen portfolios formed in this way have negative average returns, and there is no particular observable pattern both size and bankruptcy risk-wise. We expect a monotonic pattern that displays the changes in the level of equity returns as the value of bankruptcy risk increases or decreases. The average returns calculated for small and medium portfolios with low bankruptcy risk are greater than returns for small and medium portfolios with high bankruptcy risk. Thus, the high minus low average returns for these portfolios are negative. Though the size-returns dimension of these results is consistent with the existing literature—in which small-sized firms are widely known for greater levels of returns compared to large-sized firms—the findings are inconsistent with the risk-return relationship, wherein higher return is normally expected for higher risk.

Table 11.2

Average Returns of DLI-Sorted Portfolios Controlled by Size

Average returns
DLI-sorted portfolios
Size	Low	2	3	4	High	High–low	t-values
1 Small	−0.07751	−0.02779	−0.07906	0.01357	−0.08318	−0.00567	−0.2677
2	−0.06496	−0.07021	−0.07671	−0.05695	−0.06859	−0.00363	−0.0974
3 Large	−0.06277	−0.10894	−0.07589	−0.06512	−0.03927	0.02350	1.428

Stocks are sorted into three portfolios by their levels of market capitalization (sizes). In each size-sorted portfolio, stocks are then sorted into five portfolios by their DLIs. Next, the average returns of DLI size–sorted portfolios are computed. When stocks are sorted by size, Portfolio 3 contains the biggest stocks. When stocks are sorted by DLI, Portfolio 5 contains the stocks with highest bankruptcy risk. “High–low” is the return difference between the high and low bankruptcy risk portfolios. Significance at the 1 and 5% levels is indicated by ** and *, respectively.

Table 11.3 presents average portfolio returns of stocks which are formed by sorting stocks using the levels of their BM ratios (low, medium, and high) and quintile DLI bankruptcy risk measures. BM is calculated as the ratio of the book value of the stock divided by the market value of the stock for the purpose of convenience. Hence, 15 bankruptcy-BM portfolios are formed. The portfolio formed by taking the intersection of medium BM–ratio stocks and high bankruptcy risk (high DLI) stocks produce an interesting finding, whereby the difference in average returns between high and low portfolios in this category generate a value of 6.38%, which is statistically significant at 1% level of significance. However, the average returns for all 15 portfolios are negative. Furthermore, there is no particular pattern observed in average returns of the 15 bankruptcy risk–sorted portfolios which were formed by controlling for the BM factor. We expect a monotonic pattern that displays the changes in the level of equity returns as the value of bankruptcy risk increases or decreases.

Table 11.3

Average Returns of DLI-Sorted Portfolios Controlled by BM

Average returns
DLI-sorted portfolios
Size	Low	2	3	4	High	High–low	t-values
1 Small	−0.06794	−0.04014	−0.09414	−0.05854	−0.07791	−0.00996	−0.7566
2	−0.06524	−0.05655	−0.11871	−0.02531	−0.00140	−0.06383	5.8359**
3 Large	−0.06636	−0.04218	−0.04209	−0.03213	−0.08160	−0.01524	−0.9402

Stocks are sorted into three portfolios by levels of their BM ratios. In each BM-sorted portfolio, stocks are then sorted into five portfolios by their DLIs. Next, average returns of the DLI BM–sorted portfolios are computed. When stocks are sorted by BM, Portfolio 3 contains the stocks with the highest BM. When stocks are sorted by DLI, Portfolio 5 contains the stocks with the highest bankruptcy risk. “High–low” is the return difference between the high and low bankruptcy risk portfolios. Significance at the 1 and 5% levels is indicated by ** and *, respectively.

Table 11.4 presents results derived by applying regression analysis using the Fama and French (1993, 1996) factors augmented by the bankruptcy risk factor, that is, the ADLI (Vassalou and Xing, 2004). The coefficients for the market portfolio are significantly positive at the 1% level, whereas the coefficients for the size factor are significantly negative during the whole period, and the precrisis and crisis periods, but significantly positive during the postcrisis period. The coefficients for the BM factor are significantly negative during all periods of analysis. The ADLI factor is positively correlated to the stock return series during all four periods, except for the crisis period, but significantly positive only during the postcrisis period. If bankruptcy risk is clearly related to the level of equity returns consistent with the fundamental principle in finance which proclaims direct relationship between risk and return, we expect significantly positive coefficients for the ADLI factor during all periods of analysis.

Table 11.4

Results of Multifactor Regression Analysis During Various Economic States

Market	CONST	EMKT	SIZE	BM	ADLI	R² (adj.)
Whole	0.04661*	0.09046*	−0.2549*	−0.00004*	0.0096	40.86%
Precrisis	0.0336	0.09472*	−3.092*	−0.000053*	0.28	47.32%
Crisis	0.0286	0.0529*	−1.541*	−0.000017*	−0.745*	26.24%
Postcrisis	−0.049*	0.11224*	0.1845*	−0.000052*	0.63*	35.64%

This table presents the results from the test of size and BM factors along with bankruptcy risk measures. EMKT refers to the excess return on the stock market portfolio over the risk-free rate. Size refers to returns on the zero-investment portfolio, which is long on stocks with a small market capitalization (size) and short on large-sized stocks. BM refers to the BM ratio of each stock. ADLI presents the average DLI, which is a simple average of the DLIs for all firms. The estimation period is from Jan. 1996 to Dec. 2007. Significance at the 1% level is indicated by *.

The findings based on the portfolio analyses do not show an adequate level of support for the existence of a positive and significant relationship between bankruptcy risk and equity returns. The prevalence of below-zero values for portfolio returns of stocks sorted by the DLI measure indicate that these stocks have generally underperformed over the study period. Furthermore, a negative value for the difference between portfolios formed from stocks with a high DLI measure and portfolios formed from stocks with a low DLI measure is unexpected.

No monotonic and expected patterns of average portfolio returns were calculated for those stocks sorted by the DLI measure while being controlled for size and BM factors. According to Fama and French (1996), anomalous size and value premiums are expected to arise as a result of a financial distress factor. The DLI is expected to effectively stand as a proxy for bankruptcy risk, and hence we would expect the average portfolio returns to reveal a monotonic trend due to the fundamental relationship between risk and return, that is, the higher the risk, the higher would be the expected return. The expected relationship is only demonstrated by the medium portfolio sorted by the BM factor, whereby the difference between high BM–sorted and low BM–sorted portfolio return is positive and statistically significant at a 1% level.

The results of regression analysis show that the coefficient for the ADLI is negatively related to stock returns during the financial crisis period, unlike the positive and statistically significant coefficient during the postcrisis period. The coefficients for the ADLI during the whole period and the precrisis period are also consistent with the fundamental theory of finance at least in terms of direction.

5. Conclusions and Recommendations

We tested the relationship between bankruptcy risk and equity returns by conducting portfolio and regression analyses for stocks listed on the HOSE in order to expand the boundaries of research in this topical area by including a frontier market. Expected monotonic patterns were observed neither from the average returns of stocks sorted by the DLI measure nor from the average returns of stocks sorted by the DLI and size factors, as well as by the DLI and BM factors. Further analysis was conducted by applying the Fama and French (1993, 1996) three factors augmented by the ADLI in cross-sectional settings. The regression analysis was conducted for the whole study period, as well as for the subperiods, that is, precrisis, crisis, and postcrisis. Except for the crisis subperiod, the coefficients for the ADLI factor indicate a direct relationship between bankruptcy risk and stock returns in the HOSE. The coefficient for the analysis conducted during the postcrisis period is also statistically significant. We recommend repeating the analysis using the Altman (1968) and Ohlson (1980) bankruptcy risk measures.