Corporate default risk and stock returns : evidence from the UK

(1)

Corporate Default Risk and Stock Returns:

Evidence from the UK

August 2015

Master in International Finance Thesis

Student: Şule Ergϋt

(2)

Abstract

This paper analyses whether the default risk is correctly priced in the UK stock market by measuring the relationship between company default risk and stock returns and using Altman Z-score and Ohlson’s O-score models as default risk parameters. CAPM and Fama-French-Carhart asset pricing models are used for measuring the relationship between default risk and stock performance. The empirical analysis examines all listed companies in the London Stock Exchange, during the period 2000-2014. This study shows that Altman model performs better in pricing the default risk in line with the asset pricing theories and therefore can be used as a predictor for default risk in UK. The empirical results are interesting since they can be used by company management for financing decisions, by regulatory authorities and by portfolio managers in stock selection.

(3)

According to modern finance theory, there should be high returns for bearing the elevated risk that is associated with financial distress, bankruptcy, default and idiosyncratic volatility; however, this is not always the case. The relationship between default risk and stock returns, namely default risk premium, has been a subject of intense debate in the literature. Many studies investigated this relationship between default risk and stock returns using different default risk measurement techniques; whereas some supported evidence to a positive relationship and some were negative. A recent article by Kevin Aretz, Chris Florackis and Alexandros Kostakis (2014) that is entitled "Do stock returns really decrease with default risk? New International Evidence" also investigates this issue. The paper states that since the vast majority of defaults occur during recessions (see also Campbell et al., 2011; Moody’s, 2011), standard asset pricing theory predicts that highly distressed stocks should yield higher premia relative to less distressed ones. However, in contrast, prior empirical studies of the U.S. market usually report a flat, negative, or even hump-shaped relation between stock returns and several well-established proxies for default risk. Only few recent studies have reported a significantly positive relation. The puzzling relation between distress risk and stock returns is often called the "distress anomaly".

According to Chava and Purnanamdan (2010) the relationship between default risk and stock returns has important implications for the risk–reward trade-off in financial markets. If default risk is systematic, then investors should demand a positive risk-premium for bearing this risk. Contrary to what the CAPM predicts, recent empirical studies (Dichev 1998 and Campbell et al. 2008, among others) document a negative relationship between default risk and realized stock returns in the post-1980 period.

Among the first studies which examined the pricing of default risk was Dichev (1998), who used Altman’s (1968) Z-score and Ohlson’s (1980) O-score, two accounting-based proxies and showed that these measures are not positively related to stock returns. Several other empirical analyses examined this relationship and measured the effect of

(5)

default risk on the stock returns in the past years including Griffin and Lemon (2002) and George and Hwang (2010) who also used O-score.

This study will investigate the relationship between default risk and stock returns and if the default risk is correctly priced in the UK in line with the asset pricing models using accounting based metrics of Altman’s Z-score and Ohlson’s O-score for measuring the default risk. There are not many studies up to date which addressed this relationship based on using pre-crisis (before 2008) and post-crisis (2008 – 2014) data in UK which is a major market economy. UK is particularly interesting as London Stock Exchange is the third largest stock exchange in the world and the largest in Europe with a market cap of $6.06 trillion. The study will construct a dataset of all the publicly traded firms in UK, and will try to shed a new light on the largely debated cross sectional relation between default risk and stock returns.

2. Literature Review

The relationship between risk and return has been a long time debated issue and particularly interesting for many different parties from academia to various investors. Starting from 60s academicians published their works on different methods which were evolved for the measurement of the default risk. The methods developed and diversified as the advancements in the statistical tools and progress in technology.

2.1 Earlier studies on bankruptcy prediction

Beaver (1966), an accounting researcher, was among the first to investigate financial ratios as predictors of business failure by an empirical study. Using matched samples of "failed" firms versus "non-failed" firms, Beaver found that several easily available financial ratios were good predictors of failure, while others, probably more widely used, were mediocre predictors. Specifically the criterion ratios cash flow/total assets, net income/total assets, total debt/total assets and particularly cash flow/ total debt were good predictors of failure, the latter even up to five years before the event, while

(6)

such widely used ratios as the "current" ratio were of only mediocre value until the final year before failure, and even then inferior to the aforementioned ratios.

Altman (1968) assessed the quality of traditional ratio analysis as an analytical technique for measuring bankruptcy risk. Altman's work built upon research by Beaver and others. Beaver's work (1966 and 1968) was the first to apply a statistical method, t-tests to predict bankruptcy for a pair-matched sample of firms. Beaver applied this method to evaluate the importance of each of several accounting ratios based on univariate analysis, using each accounting ratio one at a time. Altman's primary improvement was to apply a statistical method, discriminant analysis, which could take into account multiple variables simultaneously. He investigated a set of financial and economic ratios in a bankruptcy prediction context wherein a multiple discriminant analysis (MDA) is employed. He questioned which ratios are most important in detecting bankruptcy potential, what weights should be attached to those selected ratios, and how should the weights be objectively established. The initial Z-score formula for predicting bankruptcy was published in 1968. The formula can be used to predict the probability that a firm will go into bankruptcy within two years. Z-scores are still used to predict corporate defaults and an easy-to-calculate control measure for the financial distress status of companies in academic studies. Altman’s Z-score rating model will be used as one of the distress risk measures in this study.

Altman, Haldeman and Narayanan (1977) developed a new bankruptcy classification model to improve and extend upon the statistical models which were published in the literature in the previous decade. The new model was called “Zeta’’ model considered recent developments like the increasing number of business failures and utilized a sample of bankrupt firms essentially covering the period 1969-1975. The study also incorporated current enhancements in the utilization of discriminant statistical techniques. Another addition was the inclusion of retailing companies into the new model. This new model was effective in identifying bankrupt companies up to five years prior to failure on a sample of corporations consisting of manufacturers and retailers.

(7)

Ohlson (1980) developed an alternative scoring method which is called O-Score for measuring bankruptcy using conditional logit. Similar to the Z-score, the O-score is usually described as a statistical bankruptcy indicator generated from a set of balance sheet ratios. It differs from Altman’s original in its application of a much larger sample of corporate successes and failures to inform the model. The wider pool of over 2000 companies gives it a more robust sample for basing the scaling factors applied to its nine variables with the intention of increasing its accuracy. The difference in this sample size is especially apparent when compared to Altman’s original whose statistical technique of pair matching limited him to just 66 companies. Subsequent studies have generally found the O-score to be a better forecaster of bankruptcy than the Z-score. This study will use also O-score as distress risk measurement technique and the both scores results will be compared.

Hillegeist, Keating, Cram, and Lundstedt (2004) show that both O-score and Z-score are limited in their forecasting power and advocate the use of a measure based on the Black and Scholes (1973) and Merton (1974) option pricing framework (called Prob), similar to the EDF measure provided commercially by MKMV. They show BSM-Prob provides significantly more information than either of the two-accounting-based measures. The Z- and O-Scores are calculated using fiscal-year-end data and are used to measure the risk of bankruptcy over the twelve-month period beginning four months after a firm’s fiscal year end. The Z- and O-scores are computed as the fitted values using the original coefficients. Authors report that computed this way, the accounting-based measures do not represent bankruptcy probabilities, but can be turned into probabilities using the logistic transformation.

Agarwal and Taffler (2008) compares the performance of two alternative formulations of market-based models based on Black and Scholes (1973) and Merton (1974); one following Hillegeist et al. (2004) and the other Bharath and Shumway (2004), with a well-established UK-based Z-Score model for the prediction of bankruptcy. Their results showed that in terms of predictive accuracy, there is little difference between the market-based and accounting models. Similar to Hillegeist et al. (2004), the authors

(8)

argued that neither market-based models nor the accounting-ratio-based model is a sufficient statistic for failure prediction and both carry unique information about firm failure. They concluded that despite extensive criticism of traditional accounting-ratio-based credit risk assessment approaches, in practice such conventional approaches are robust and not dominated empirically by KMV-type option-based models. According to authors, the accounting based approach produces significant economic benefit over the market-based approach.

2.2 Studies on the relationship between default risk and stock returns using Altman and Ohlson Models and developments in Altman´s model

Dichev (1998) presents a comprehensive analysis about the relationship between bankruptcy risk and systematic risk by measuring bankruptcy risk through existing models of Altman (1968) and Ohlson (1980). He investigates the importance of firm distress risk factor and its relation to firm size and book-to-market effects. Dichev highlights that a natural proxy for firm distress is bankruptcy risk. If bankruptcy risk is systematic, one would expect a positive relationship between bankruptcy risk and subsequent realized returns. However, results demonstrate that bankruptcy risk is not actually rewarded by higher returns. Hence, a distress factor is unlikely to account for the firm size and book-to-market effects. Remarkably, firms with high bankruptcy risk earn lower than average returns since 1980. He concludes that a risk-based explanation cannot fully explain the unusual evidence.

Altman (2000) discusses the developments in his two accounting based metric models; Z-Score (1968) and Zeta (1977) credit-risk model for assessing the distress of industrial corporations. The purpose of Altman’s article is understanding the recent characteristics of business failures in order to specify and quantify the variables which are indicators and predictors of corporate distress.

Griffin and Lemmon (2002) explores the relationship between book-to-market equity, distress risk and stock returns. They use Ohlson’s O-score to proxy firms with the highest distress risk and find that the difference in returns between high and low

(9)

book-to-market securities is more than twice as large as that in other firms. The authors states that high book-to-market equity firms are assigned a higher risk premium because of their greater risk of distress. Dichev’s (1998) bankruptcy risk measurement by Ohlson and Altman model revealed that high likelihood of financial distress results in low average stock returns. Authors found that the negative relationship between default risk and realized stock returns documented by Dichev (1998) is strong in growth firms. Dichev’s results are also inconsistent with the author´s view that firms with high book-to-market ratio earn high returns as a premium for distress risk.

Gerantonis, Vergos and Christopoulos (2009) analyse whether Altman Z-score models, can predict correctly company failures in Greece during the period 2002-2008. The authors investigated whether Z-score models can predict bankruptcies for a period up to three years earlier. The study shows that Altman model performs well in predicting failures. This study follows a similar methodology to their study in terms of data selection.

2.3 Latest studies of default risk prediction based on more advanced estimation techniques

Along with the development of more advanced bankruptcy measurement techniques academics in this field started to contribute to the literature with new type of studies with more advanced methodologies. One example is Campbell, Hilscher and Szilagyi (2008) who explored the determinant factors of corporate failure and the pricing of financially distressed stocks by presenting a model of corporate failure in which accounting and market-based measures forecast the likeliness of future financial distress. To measure the probability that a firm enters either bankruptcy or failure, they estimate a dynamic panel model using a logit specification, following the studies of Shumway (2001) and Chava and Jarrow (2004). They extend the previous literature by considering a wide range of explanatory variables, including both accounting and equity-market variables, and by explicitly considering how the optimal specification varies with the horizon of the forecast. Their best model is more accurate than leading

(10)

alternative measures of corporate failure risk. Their study revealed that firms with higher leverage, lower profitability, lower market capitalization, lower past stock returns, more volatile past stock returns, lower cash holdings, higher market-book ratios, and lower prices per share are more likely to file for bankruptcy, be de-listed, or receive a D rating. When predicting failure at longer horizons, the most persistent firm characteristics, market capitalization, the market-book ratio, and equity volatility become relatively more significant.

Garlappi, Shu and Yan (2007) examine the relationship between default probability and stock returns using the Expected Default Frequency (EDF) of Moody’s KMV, which is widely used by financial institutions as a predictor of default probability. Their study focus on the role of default risk in explaining some of the anomalies in the cross section of equity returns. Authors report that higher default probabilities are not associated with higher expected stock returns. However, investors demand a positive premium for holding stocks of firms that face high probability of default. Their theoretical analysis and empirical evidence indicate that ‘shareholder advantage’ has a direct impact on the equity risk of firms with high default probability and helps account for much of the observed cross-sectional variations in the relationship between default probability and stock returns, in addition to the known effects of size, book-to-market ratio and momentum.

Vassalou and Xing (2004) construct a metric for default probability to mimic the EDF measure. They find that high-default-probability firms with a small market capitalization and a high book-to-market ratio earn higher returns than their low-default probability counterparts and conclude that low-default risk is systematic and positively priced in stock returns. This result is contrary to the other evidence in the literature and has been challenged on the ground of return attribution.

Chen, Cholette and Ray (2009) challenged the link between distress risk and idiosyncratic volatility by examining the twin puzzles of anomalously low returns for high idiosyncratic volatility stocks and high distress risk stocks, documented by Ang et

(11)

al. (2006) and Campbell et al. (2008). Authors document that these puzzles are empirically connected, and can be explained by a simple, theoretical, single-beta CAPM model. They investigated the link between idiosyncratic volatility effect and the distress effect by sequential sorting and proxied for firms distress risk by Altman’s (1968) Z-score and Ohlson’s (1980) O-score. In their primary exercise, they controlled for the distress effect by first sorting stocks into quintiles according to their Z-score and O-score, then within each quintile, sorting again into portfolios based on firms’ idiosyncratic volatility. They found out that stocks with high idiosyncratic volatility earn significantly lower returns than low idiosyncratic volatility stocks, mainly in extreme quintiles: those with the lowest and highest distress risk.

Chava and Purnanamdam (2010) found a positive cross-sectional relationship between expected stock returns and default risk, contrary to the negative relationship estimated by prior studies. Whereas prior studies use ex post realized returns to estimate expected returns, authors use ex ante estimates based on the implied cost of capital. The results reveal that investors expected higher returns for bearing default risk, but they were negatively surprised by lower-than-expected returns on high default risk stocks in the 1980s. The authors also extended the sample compared with prior studies and found that the evidence based on realized returns is considerably weaker in the 1952–1980 period. They also examined the impact of default risk on stock returns over a long period of time. Consistent with Campbell et al. (2008), they find that high default risk stocks significantly underperform in the post-1980 period; however, there is no evidence that these stocks underperformed during 1952–1980. Their results show that the post-1980 underperformance of high default risk stocks does not represent an asset pricing anomaly; rather, it is due to surprisingly low realized returns during the 1980s.

Campbell, Hilscher and Szilagyi (2010) extended their initial study and used their measure of financial distress to examine the performance of distressed stocks from 1981 to 2008. They find that distressed stocks have highly variable returns and high market betas and that they tend to underperform safe stocks by more at times of high

(12)

market volatility and risk aversion. However, investors in distressed stocks have not been rewarded for bearing these risks. Instead, distressed stocks have had very low returns, both relative to the market and after adjusting for their high risk. The underperformance of distressed stocks is present in all size and value quintiles. Based on the results of the study, they suggest that investors should stay away from investing in distressed stocks.

Aretz, Florackis and Kostakis (2014) followed Campbell et al.’s (2008) methodology for assessing bankruptcy risk and for estimating default risk probabilities brought a new standpoint to the relationship between default risk and stock returns but this time with a larger sample of non-US firms in 14 developed markets. Their study supported the existence of an economically and statistically significant positive default risk premium in international markets. Their approach was an inspiration to conduct a similar study with less advanced techniques by focusing on a non-US country (in this case UK).

3. Theoretical Framework

This section discusses the theoretical aspects and practical methodology employed in the data selection for this study. The theoretical aspects of data selection employed for the most part in this study follow very closely the paper by Altman (1968) and Ohlson (1980), while in other parts necessary deviations are present.

3.1 Variable description

The sample selection process starts with assessing and filling the requirements of Altman´s and Ohlson´s model. Both scores are accounting based measures which include several accounting and financial ratios with a different weighting factor determined based on the extensive calibration studies of Altman and Ohlson.

(13)

3.1.1 Z score

Altman (1968) considered five ratios jointly doing the best job in bankruptcy prediction, which belonged to the following categories: liquidity, profitability, leverage, solvency, and activity. Since this study firstly uses the model advocated by Altman, the initial variable selection is in line with Altman's method. The discriminant Z-score function in this study is constructed as follows:

Z score bankruptcy model: Z = 1.2T1 + 1.4T2 + 3.3T3 + 0.6T4 + .999T5 Original z-score component definitions variable definition weighting factor:

T1 = Working Capital / Total Assets

T2 = Retained Earnings / Total Assets

T3 = Earnings Before Interest and Taxes / Total Assets

T4 = Market Value of Equity / Total Liabilities

T5 = Sales/ Total Assets

Zones of Discrimination:

Z > 2.99 -“Safe” Zones: Considered financially healthy

1.81 < Z < 2.99 -“Grey” Zones: Could go either way

Z < 1.81 -“Distress” Zones: Risk that company will go bankrupt within two years

Z score components:

X1 - Working capital/Total assets. The first ratio advocated by Altman (1968) is a liquidity ratio. Its numerator, working capital, is the difference between current assets and current liabilities. Thus looking at the variables that this ratio is made of, one may correctly note that this ratio is taking into account liquidity and size aspects. More specifically, it is a liquidity ratio because it reflects the net liquid assets in relation to

(14)

the total capitalization. With this ratio one would expect a negative relation with bankruptcy or a positive relation with non-bankruptcy, as decreasing working capital relative to total assets is a liquidity indicator of operating losses which in turn a affects a firm negatively.

X2 - Retained earnings/Total assets. The second ratio advocated by Altman (1968) is a (cumulative) profitability ratio. Interestingly, this is a new ratio that was proposed by Altman himself. Next to profitability, age is also taken into account when this ratio is used. The implicitly considered age characteristic of this ratio can be understood better with the example put forth by Altman: A lower Retained earnings/Total assets ratio is expected for younger firms because they did not have time to grow and build up their cumulative profits. Hence one may correctly expect that younger firms will more probably be classified as bankrupt relative to older firms. With this ratio one would expect a negative relation with bankruptcy or a positive relation with non-bankruptcy, as negative profits, which affect a firm negatively, decrease retained earnings relative to total assets.

X3 - Earnings before interest and taxes/Total assets. The third ratio advocated by Altman (1968) is a solvency ratio. Insolvency takes place, from a bankruptcy point of view, when the total liabilities are higher than the fair valuation of a company's assets, with value established by the earnings power of the assets. It basically measures the productivity of a company's assets, ignoring leverage and taxes effects. EBIT in this setting provides the earnings power assessment of the assets, which is a determinant of a company's ultimate existence. Hence this ratio can be appropriately used in assessing a firm's continuation. With this ratio one would expect a negative relation with bankruptcy or a positive relation with non-bankruptcy, as decreasing EBIT relative to total assets is a solvency indicator of decreasing earnings power of the assets which in turn affects a firm negatively.

X4 - Market value of equity/Total liabilities. The fourth ratio advocated by Altman (1968) is a leverage ratio. This ratio can be broken down into its variables: liabilities

(15)

and market value of equity, which is calculated as the total market value of all shares of stock (common and preferred). It basically shows when the firm will become insolvent, in terms of how much a company's value (market value of equity plus liabilities) can decrease before the company value will be exceeded by liabilities. With this ratio, one would expect a negative relation with bankruptcy or a positive relation with non-bankruptcy, as decreasing market value of equity relative to total liabilities is a leverage indicator of decreasing solvency which in turn affects a firm negatively.

X5 - Sales/Total assets. The fifth ratio advocated by Altman (1968) is an activity ratio. This ratio, usually employed by company managements, is key in handling competitive situations and rejects the sales yielding capability of a company's assets. With this ratio one would expect a negative relation with bankruptcy or a positive relation with non-bankruptcy, as decreasing sales to total assets is an activity indicator of decreasing sales which in turn a affects a firm negatively.

3.1.2 O-Score

The Ohlson O-Score is the result of a 9-factor linear combination of coefficient-weighted business ratios which are readily obtained or derived from the standard periodic financial disclosure statements provided by publicly traded corporations. The nine different approximate measures of a firm's default risk are used to determine firm size, leverage, working capital, liquidity, profitability, change in net income, and debt financing. Two of the factors utilized are widely considered to be dummies as their value and thus their impact upon the formula typically is 0. Together, these nine variables build an O-Score where the probability of failure is EXP(O-Score) divided by 1+EXP(O-score). Results greater than >.5 indicate a firm with a high chance of default.

(16)

The calculation for Ohlson’s O-Score appears below: where; TA = total assets TL = total liabilities WC = working capital CL = current liabilities CA = current assets · X = 1 if TL > TA, 0 otherwise · NI = net income

· FFO = funds from operations

· Y = 1 if a net loss for the last two years, 0 otherwise

3.2 Methodology and hypothesis

This section discusses the methodology that is employed in this thesis. In this study, two widely used asset pricing methodologies is used to test the relationship between default risk and stock returns; namely capital asset pricing model (CAPM) and Fama-French three factor model. Firstly we estimate alpha from the CAPM:

Ri,t– Rft= αi+ βi,MKT (Rm,t– Rft) + εi,t

where Ri,t is the return of portfolio i in month t, Rft is the risk-free rate for month t and MKTt (Rm,t– Rft) is the excess market portfolio return, in month t.

(17)

Secondly, we compute Fama–French-Carhart alpha, i.e. the intercept of the four-factor Fama-French-Carhart (1997) model:

Ri,t– Rft= αi+ βi,MKT(Rm,t– Rft) + βi,SMBSMBt+ βi,HMLHMLt+ βi,UMDUMDt+ εi,t

where Ri,t is the return of the portfolio i in month t, Rft is the risk free rate for month t, (Rm,t – Rft) is the excess market portfolio return in month t, βi,MKT is the beta of portfolio i, SMBt1, HMLt2 and UMDt3 stand for the size premium, value premium and momentum factors respectively, while βi,SMB, βi,HML and βi,UMD denote the corresponding factor loading of portfolio i.

SMB, which stands for Small Minus Big, refers to additional return that investors historically received by investing in stocks of companies with relatively small market capitalization. This additional return is often referred as "size premium". HML, which

1_{Fama and French (1992) found that value and size factors are the most significant two factors outside}

of market risk for explaining the realized returns of publicly traded stocks. To represent these risks, they constructed two factors: SMB and HML. SMB, which stands for Small Minus Big, is designed to measure the additional return investors have historically received by investing in stocks of companies with relatively small market capitalization. This additional return is often referred to as the “size premium.”

2Fama and French (1992) found that value and size factors are the most significant two factors outside

of market risk for explaining the realized returns of publicly traded stocks. To represent these risks, they constructed two factors: SMB and HML. HML, which is short for High Minus Low, has been constructed to measure the “value premium” provided to investors for investing in companies with high book-to-market values (essentially, the value placed on the company by accountants as a ratio relative to the value the public markets placed on the company, commonly expressed as B/M)

3 Carhart (2009) four-factor model is an extension of the Fama–French three-factor model including a

momentum factor (UMD), also known in the industry as the MOM factor (monthly momentum). Momentum in a stock is described as the tendency for the stock price to continue rising if it is going up and to continue declining if it is going down. UMD can be calculated by subtracting the equal weighted average of the highest performing firms from the equal weighed average of the lowest performing firms, lagged one month.

(18)

stands for High Minus Low, has been constructed to measure the "value premium" provided to investors for investing in companies with high book-to-market (book value to market value ratio) values. There are several studies which suggest that effects of firm size and book-to-market ratio could be related to some sort of a firm distress risk factor.

In line with CAPM and Fama French asset pricing models, the hypothesis test will be as follows:

H1: Interference, if the relationship between default risk and stock returns is negative in UK, which means if the stock returns decrease as the portfolio gets riskier

H2: Alignment, if the relationship between default risk and stock returns is positive in UK, which means if the stock returns increase as the portfolio gets riskier

H3: Neutral, if there is no statistically significant relationship between default risk and stock returns in UK

3.3 Data

Having established the needed ratios and method to be used in the empirical research, the next step in order to make a viable study for measuring the relationship between distress risk of the companies in UK and their stock returns, is determining the period for collecting the data observations. As there are not many similar published studies performed in UK using Z-scoring and O-scoring methodology for assessing the relationship between default risk and stock returns, the period was chosen based on mutual judgement with the supervisor of this thesis to have a broad horizon covering both the pre-crisis and post-crisis periods. As a result, the period of analysis is determined as 2000 to 2014. After the examination period is chosen, a sample population is formed from all companies which are traded in FTSE All Share Index in each of the years over the period 2000 to 2014.

(19)

In order to measure default risk based on Altman´s Z-scoring and Ohlson´s O scoring method, company financials are essential to make a computation of accounting ratios. Having decided upon the time period the next step was gathering the financial statements of publicly traded UK companies. The financial statements of all the companies which are listed on FTSE (London Stock Exchange) has been extracted for a period of 14 years starting from 2000 year end to 2014 year end using WRDS Compustat Global Fundamentals Database. Out of 1,497 corporate (manufacturing) and financial services companies were publicly traded and had published financials statements during 2000 - 2014, 340 were financial services companies and therefore eliminated from the dataset. The data used in this paper are limited to 1,157 manufacturing corporations in line with Altman’s classical research. Out of the remaining 1,157, only 196 were publicly traded during the whole period (15 years from 2000 to 2014) whereas the rest 961 were active during different time frames.

This initial data set already included all the variables needed for the calculation of Z-score and O-Z-score except market value of equity. The variables extracted in the initial dataset were receivables, current assets, current liabilities, working capital, total assets, total liabilities, retained earnings, sales, EBIT, net income/loss, outstanding shares issued and funds from operations. To be able to compute the market value of equity (4th_{component in Z-score calculation), the year-end closing share prices are}

extracted from Thomson DataStream database and integrated into the data. Year-end share prices were multiplied with the number of outstanding shares as of year-end. Then default risk proxies (i.e. Z-scores and O-scores) for each company for each year were calculated based on the existing methodology which were explained in detail in the related section above. Overall for 1,157 companies in the sample, 8,273 Z-scores and 8,671 O-scores were calculated spanning the years 2000-2014. Calculated Z-scores are less than the O-Z-scores due to lack of either share price or number of outstanding shares information for calculating market value of equity. For some descriptive statistics regarding the key variables of the financial statements of these

(20)

companies like mean, median and standard deviations please refer to Table 2 in the Appendix.

Finally, to analyse the relationship between the default risk and stock returns, month end stock closing prices of 1,157 UK companies for the related period (2000-2014) has been extracted from Thomson DataStream database. To make a double check the same data has also been extracted from Compustat Global Securities Daily database and compared with Thomson´s output. The both data were in line.

4. Empirical Research

The main empirical question that this paper investigates is the existence of a relationship between default risk and stock returns based on CAPM and FFC (Fama-French-Carhart four-factor) models in which the default risk is measured by Altman’s Z-score and Ohlson’s O-Z-score. In order to do this empirical investigation, upon integrating the Z-score and O-score of the companies for the whole 14 years period and the monthly stock prices, the data has been transformed into a panel data so that average excess returns can be calculated and linear regression analysis can be performed to estimate CAPM and FFC four factor model alphas.

To begin with, monthly returns per individual company stocks have been calculated for all the years which the company was actively traded. As previously stated some companies were not actively traded during the whole period and therefore they also did not have updated monthly stock prices in their latest trading year. For these companies, returns are 0% in the dataset starting from the month which their stock prices are no longer updated.

After the monthly return calculation of individual stocks, the data has been sorted based on Z-scores and O-scores separately per each individual year. Based on Z-scores and O-Scores separately, data was proxied. In order to make a comparative analysis both portfolios are categorized into deciles starting from lowest risk (portfolios

(21)

containing less-distressed stocks) to highest risk (portfolios containing distressed stocks). A third categorization was also made based on Altman’s (1968) methodology which is called Z-score band. This categorization classifies the scores into three different groups as safe, grey and distress as explained in the related section above. After the separation of the data into deciles and Z-score band categories, monthly UK market factors including risk free rate (rf), market return (rm), HML, SMB and UMD has also been integrated into the panel data for performing CAPM and Fama French Carhart four factor analysis (source : University of Exeter Business School). Since data for monthly factors was available only until end September 2014, the company stock returns that was extracted for the last three months of 2014 has been removed from the analysis. After the integration of the factors into panel data, monthly returns of each portfolio per deciles and per Z-score band has been calculated. At this stage there were three different categorization of the whole portfolio of stocks; namely Z-score per deciles, Z-score per bands and O-score deciles and therefore three separate working files to calculate the excess return per portfolio level.

Subsequent to the creation of three separate working files, firstly equal weighted average of excess returns at portfolio level has been calculated per each decile or band in the working files. In order to transform the data into a statistical tool, the working files has been separated into separate working files for each portfolio. At this stage, as portfolio level returns were the same for all the companies which are in the same decile or category, duplicate data per portfolio was deleted from the working files. At the end there were only one monthly return per a certain year and month, and therefore only 177 (12*14+9*1) observations per each decile or band. Then, regressions were run individually per decile and band portfolio of stocks. For each portfolio, regressions were run per two different asset pricing methodology; CAPM and Fama French Carhart models separately to estimate the alphas and betas of the portfolios. Results based on the categorization of portfolios are explained as follows.

As Z-scores were sorted from low to high scores and the stocks were allocated into deciles based on the highest score representing the lowest risk as 1 and lowest score

(22)

representing the highest risk as 10, another portfolio, which is called spread portfolio, has been formed to observe the difference between the excess returns of the lowest risk and the highest risk portfolios. This difference portfolio has been named as P10-P1 and refers to spread strategy to long the decile portfolio with highest risk stocks (P10) and short the decile portfolio with the lowest default risk (P1). Excess returns per decile were calculated based on equal weighted average of the excess returns of each portfolio including spread portfolio. Regressions were run for each decile for determining CAPM alpha and FFC alpha separately. Finally, average excess returns and alphas are annualized before comparison. From the results it is observed that return of the portfolio is increasing with the increase in the default risk. There is a considerably high difference between the highest risk (P10) and the lowest risk portfolio (P10) which is in line with the existing methodology of CAPM and Fama-French-Carhart Models. The difference between the lowest and highest risk portfolio (spread portfolio) was 23.1% for excess return, 23% for CAPM alpha and 19.9% for FFC alpha. When significance test was performed on the difference portfolio, t-statistics stood at 4.24 and 3.53 respectively for CAPM and FFCM and were significant at 1%, 5% and 10% significance levels. The results of the regressions of Z-scores which depicts CAPM, Fama French Carhart alphas and average excess returns of each decile portfolio can be observed in Table 3 in the appendix section.

Similarly Z-scores bands portfolios were classified in line with the existing literature explained above and the portfolios were divided into categories as Safe, Grey and Distress. The difference portfolio has been calculated to observe the difference between Distress and Safe portfolios and named as PDistress-PSafe. This portfolio refers to spread strategy to long the distressed portfolio (PDistress) and short the safe portfolio (PSafe). Excess returns per category were calculated based on equal weighted average of the excess returns of each portfolio including spread portfolio. Regressions were run for each band to determine CAPM alpha and FFC alpha separately. Finally, average excess returns and alphas are annualized before comparison. From the results it is also observed that excess return of the portfolio is increasing with the increase in

(23)

the default risk. There is a significant difference between the Distress and the Safe portfolio which is in line with the existing methodology of CAPM and Fama-French-Carhart Models. The differences between the distress and the safe portfolio was 11.7% for excess return, 6% for CAPM alpha and 4.9% for FFC alpha. When significance test was performed on the difference portfolio, t-statistics stood at 2.55 and 1.98 respectively for CAPM and FFCM and were significant at 1%, 5% and 10% significance levels. The results of the regressions for Z-scores which reveal CAPM, Fama French Carhart alphas and average excess returns of each portfolio category can be observed in Table 4 in appendix.

Finally O-scores were analysed after sorting from high to low scores as this is an inverse formulation compared to Z-score and any score above +0.5 represents a high risk of default. Therefore, stocks were allocated into deciles based on the lowest score representing the lowest risk as 1 and the highest score in this case representing the highest risk as 10. Similar to Z-score, an additional portfolio which is called spread portfolio has been created to observe the difference between the excess returns of lowest risk and highest risk portfolios. This difference portfolio has been named as P10-P1 and refers to spread strategy to long the decile portfolio with highest risk stocks (P10) and short the decile portfolio with the lowest default risk (P1). Excess returns per decile were calculated based on equal weighted average of the excess returns of each portfolio including spread portfolio. Regressions were run for each decile for determining CAPM alpha and FFC alpha separately. Finally, average excess returns and alphas are annualized before comparison. However, O-score results depicted no significant relationship between risk and return. Inversely, lowest risk portfolio has produced the highest average excess return by 7.1% and the highest risk portfolio produced an excess return of 3.1%, which is 3% lower than the lowest risk portfolio. The difference between the highest risk and the lowest risk portfolio was -4% for excess return, -4.4% for CAPM alpha and -8.8% for FFC alpha. When significance test was performed on the difference portfolio, t-statistics stood at 0.78 and -1.51 respectively for CAPM and FFCM and were insignificant. The results of the regressions

(24)

for O-scores which reveal CAPM, Fama French Carhart alphas and average excess returns of each decile portfolio can be observed in detail in Table 5 in the appendix section.

5. Conclusion

Previous studies have extensively documented the relationship between default risk and stock returns using either accounting based measures of Altman or Ohlson as default risk parameter or recently in the last decades using more advanced calculation techniques of market based measures. Some of these studies proved to be in line with the existing theory of CAPM and FFC models showing a positive relationship between default risk and stock returns and some were proven to be against the existing theory. Most of these studies were performed with US data as US provides the most extensive sample with quality information including bankruptcy filings. The purpose of this thesis was to provide empirical evidence on the relationship between default risk and stock returns in UK and to investigate if the default risk is correctly priced in UK. The study is significant in the sense that it contributes to the existing literature with a country analysis like UK which is not much previously covered by using Altman’s Z score and Ohlson’s O score default risk calculation methods at the same time while covering a broad time horizon to perform such an empirical analysis.

To perform the study, as already mentioned two different accounting based default metrics used for the same analysis; Z-score and O-score. The results of the empirical study suggested that Z-score successfully measures the relationship between default risk and stock returns as results were in line with capital asset pricing model and the highest risk stocks produced the highest excess returns during the period covering 2000-2014. Results showed that portfolios containing highest default risk stocks significantly outperform portfolios containing lowest default risk ones. The results of the Z-score in two different categorizations based on deciles and original Z-score bands led to statistically significant results. On the other hand, although usually

(25)

regarded as a more successful default risk predictor in the literature, surprisingly O-score categorization did not provide a similar output to lead any strong inferences and did not produce any statistically significant test results. Therefore it is hard to conclude that O-score methodology is a suitable measure for default risk categorization or distress risk measurement in UK. An alternative explanation could be that UK stocks are not really a proper sample for score calculation. Another reason for the vain score results could be the quality of the financial information for the calculation of O-scores. Due to time and scope limitations no further investigations were possible to conduct to dig into this problem.

Perhaps the most important weakness of the thesis is that it relies solely on the Z-Score and O-Z-Score models as the only two means of default risk estimation. Including market-based measures of financial distress with more advanced techniques like logit model, BSM-Prob and Moody’s EDF model would have strengthened the results. There are also several authors who also criticized the reliability of these methods in the literature. Begley, Ming and Watts (1996) indicated that while Altman and Ohlson measures performed relatively well when they were initially estimated, they do not perform well in more recent periods (in particular, 1980s) even when the coefficients were re-estimated. The authors found that Ohlson’s model displays the strongest performance when they compared both model’s re-estimated versions. Zmijewski (1984) also discussed methodological issues related to the estimation of financial distress measures due to sampling with two potential biases as choice-based sampling which results from “oversampling’’ of distressed firms and sample selection bias which results from using a “complete data” sample selection criterion. On the other hand, it is also possible that this study has a geographic limitation, in the sense that the findings related to the enterprises listed on FTSE might not be directly transferable to other settings. For instance, it may be that certain conditions in UK market are decisive for the findings. Testing the validity for some of the results on broader stock exchanges may help to clarify this potential restriction.

(26)

The empirical results of this study are interesting for both portfolio managers and investors interested in UK market. By using Altman’s approach, this study provides strong evidence supporting the existence of an economically and statistically significant positive default risk premium in UK market. Default risk is correctly priced in UK market. Therefore portfolio managers and investors can benefit from this by a spread strategy to form a portfolio by longing stocks with highest risk and shorting stocks with lowest default risk. The conclusions are mainly in line with the findings of the other researches in the field. However, since the sample used in this paper consists of the U.K.-based non-financial services corporations, the obtained results should be applied with caution to assess the business sustainability of the firms from other countries or specific industries. It should be acknowledged that the conclusions may vary a lot depending on the sector or the location of the enterprise, as well as the choice of metric for assessing default risk. The identification of the most robust metric and the development of the model with high predictive power could be a basis for another research in the future.

(27)

Appendix

Table 1 Definitions

Z score (discriminant score) formula Z = 1.2T1 + 1.4T2 + 3.3T3 + 0.6T4 + .999T5

T1 = Working capital/Total assets liquidity ratio. Its numerator, workingcapital, is the difference between current assets and current liabilities.

T2 = Retained earnings/Total assets (cumulative) profitability ratio T3 = Earnings before interest and taxes/Total assets solvency ratio

T4 = Market value of equity/Total liabilities leverage ratio T5 = Sales/Total assets activity ratio

Safe zone Z > 2.99

Grey zone 1.81 < Z < 2.99

Distress zone Z < 1.81

O score formula

CAPM Capital Asset Pricing Model

(28)

Table 2

Summary Statistics

This table represents summary statistics (means, medians and standard deviation) of key variables from the financial statements of 1,157 non-financial services companies used in the computation of Z-score (Altman, 1968) and O-score (Ohlson 1980) which are the default risk proxies in this study. Figures for financial statement balances are in thousand EURs.

Statistics regarding the key variables used in Z-score and O-score computation

Mean Median DeviationStandard Financial statement variables

Receivables 298 6 3,647 Current Assets 737 20 8,702 Working Capital 100 3 2,847 Total Assets 2,200 49 22,434 Current Liabilities 637 12 7,181 Retained Earnings 535 - 5,987 Total Liabilities 1,268 19 12,061 Market Value of Equity 1,804 44 25,718 Sales/Turnover (Net) 2,045 33 22,222 EBIT 243 1 3,083 Net Income (Loss) 117 0 1,173 FFO (Funds from Operations) 83 0 922 Ratios

T1 - Working Capital/Total Assets

(Z-score and O-score) - 0.1 0.1 7.7 T2- Retained Earnings/Total Assets

(Z-score) - 6.3 0.0 207.1 T3 - EBIT/Total Assets (Z-score) - 0.2 0.0 3.9 T4 -Market Value of Equity/Total Liabilities

(Z-score) 207.6 1.8 5,903.4 T5 - Sales/Total Assets (Z-score) 1.0 0.7 2.3 Total liabilities/Total assets (O-score) 1.2 0.5 35.1 Current Liabilities/ Current assets

(O-score) 1.4 0.7 9.4 Net income/Total Assets (O-score) - 0.3 0.0 6.9 FFO/Total Liabilities (O-score) 0 0 24

(29)

Table 3

Distribution of default risk per Z score deciles

This table reports average excess returns, CAPM alphas, and four factor alphas from the Fama-French-Carhart asset pricing model (FFC alphas) for portfolios constructed on the basis of Z-scores of Altman’s default risk proxies of non-manufacturing companies which were publicly traded in UK during 2000 -2014. Z-scores were sorted from low to high and the stocks were allocated into deciles based on the highest score representing the lowest risk (1) and lowest score representing the highest risk (10). P10-P1 represents the difference (spread) portfolio which is the difference between the returns of highest risk and lowest risk portfolios and refers to spread strategy to long the decile portfolio with high risk stocks (P10) and short the decile portfolio with the lowest default risk (P1). Excess returns were calculated based on equally weighted average of the excess returns of each portfolio. Regressions were run for each decile for determining CAPM alpha and FFC alpha separately. Average excess returns and alphas are annualized and represented in percentages. Significance test was made based on the difference portfolio. *** represents statistical significance at 1% level.

Deciles per Z-Score

1 2 3 4 5 6 7 8 9 10 P10-P1 Excess return -14.3% -3.0% -0.2% 6.0% 7.4% 5.1% 3.3% 6.5% 4.8% 8.8% 23.1% CAPM alpha -13.3% -2.4% 1.1% 7.0% 8.1% 5.7% 4.0% 6.8% 6.1% 9.7% 23.0% T -statistic 4.24*** FFC alpha -12.3% -1.2% 1.4% 7.7% 8.4% 6.5% 4.7% 7.3% 6.2% 7.6% 19.9% T -statistic 3.53***

(30)

Table 4

Distribution of default risk per Z score bands

This table reports average excess returns, CAPM alphas, and four factor alphas from the Fama-French-Carhart asset pricing model (FFC alphas) for portfolios constructed on the basis of Z-score bands of Altman’s default risk proxies of non-manufacturing companies which were publicly traded in UK during 2000 -2014. Z-scores were sorted from low to high and the stocks were allocated into bands based on Altman’s grading as Safe, Grey and Default. PDistress –PSafe represents the difference (spread) portfolio which is the difference between the returns of distress and safe risk portfolios and refers to spread strategy to long the decile portfolio with default stocks (PDefault) and short the decile portfolio with the safe (PSafe). Excess returns were calculated based on equally weighted average of the excess returns of each portfolio. Regressions were run for each band for determining CAPM alpha and FFC alpha separately. Average excess returns and alphas are annualized and represented in percentages. Significance test was made based on the difference portfolio. ** represents significance at 5% level.

Z-Score Band

Safe Grey Distress Distress- Safe

Excess return -3.7% 5.4% 8.1% 11.7%

CAPM alpha -0.1% 5.5% 5.9% 6.0%

T -statistic 2.55**

FFC alpha 0.7% 5.8% 5.6% 4.9%

(31)

Table 5

Distribution of default risk per O score deciles

This table reports average excess returns, CAPM alphas, and four factor alphas from the Fama-French-Carhart asset pricing model (FFC alphas) for portfolios constructed on the basis of O-scores of Ohlson’s default risk proxies of non-manufacturing companies which were publicly traded in UK during 2000 -2014. O-scores were sorted from low to high and the stocks were allocated into deciles based on the lowest score representing the lowest risk (1) and the highest score representing the highest risk (10). P10-P1 represents the difference (spread) portfolio which is the difference between the returns of highest risk and lowest risk portfolios and refers to spread strategy to long the decile portfolio with high risk stocks (P10) and short the decile portfolio with the lowest default risk (P1). Excess returns were calculated based on equally weighted average of the excess returns of each portfolio. Regressions were run for each decile for determining CAPM alpha and FFC alpha separately. Average excess returns and alphas are annualized and represented in percentages. Significance test was made based on the difference portfolio. Results were insignificant.

Deciles per O-Score

1 2 3 4 5 6 7 8 9 10 P10-P1 Excess return 7.1% 2.5% 4.5% 3.8% 2.0% 3.6% 3.1% 2.7% -6.4% 3.1% -4.0% CAPM alpha 7.1% 2.4% 4.9% 4.6% 2.6% 5.3% 4.1% 3.9% -5.3% 2.6% -4.4% T –statistic - 0.78 FFC alpha 9.5% 2.9% 5.2% 4.8% 2.7% 6.3% 4.4% 4.0% -5.9% 0.7% -8.8% T –statistic - 1.51

(32)

References

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

Altman, E. I., 1968. Financial ratios, discriminant analysis and the prediction of corporate bankrupt-cy. The Journal of Finance 23, 589–609.

Altman, E.I., Haldeman, R.G., Narayanan, P., 1977. ZETA analysis: A new model to identify bankruptcy risk of corporations. Journal of Banking and Finance 1, 29–51. Altman, Edward I., 2000. Predicting Financial Distress of Companies: Revisiting the Z-Score and Zeta Models.

Aretz, Kevin, Florackis, Chris and Kostakis, Alexandros, Do Stock Returns Really Decrease with Default Risk? New International Evidence, September 2014

Beaver, W., 1967, Financial Ratios as Predictors of Failure, Empirical Research in Accounting: Selected Studies, Supplement, Journal of Accounting Research 5, 71-127. Beaver, W., 1968, Alternative Financial Ratios as Predictors of Failure, Accounting ReviewXLIII, 113-122.

Begley, J., J. Ming and S. Watts. (1996). ‘‘Bankruptcy Classification Errors in the 1980’s: An Empirical Analysis of Altman’s and Ohlson’s Models.’’ Review of Accounting Studies 1, 267–284.

Campbell, J. Y., J. Hilscher, and J. Szilagyi. 2008. In Search of Distress-Risk. Journal of Finance 63:2899–2939.

Campbell, J. Y., Hilscher, J., Szilagyi, J., 2011. Predicting financial distress and the performance of distressed stocks. Journal of Investment Management 9, 14-34.

Chava, S., Purnanandam, A. K., 2010. Is default risk negatively related to stock returns? Review of Financial Studies 23, 2523–2559.

Chen, J., Cholette, L., Ray, R., 2009. Financial distress and idiosyncratic volatility: An empirical investigation. Journal of Financial Markets 13, 249-267.

Dichev, Ilia D., Is the Risk of Bankruptcy a Systematic Risk?, The Journal of Finance Volume 53, Issue 3, 1131-1147, June 1998

Fama, Eugene F., and Kenneth R. French, 1992, The cross-section of expected stock returns, Journal of Finance 47, 427–465.

(33)

Garlappi, L., T. Shu, and H. Yan. 2008. Default-Risk, Shareholder Advantage, and Stock Returns. Review of Financial Studies 21:2743–78.

George, T. J., Hwang, C.-Y., 2010. A resolution of the distress risk and leverage puzzles in the cross section of stock returns. Journal of Financial Economics 96, 56– 79.

Griffin, John M. and Lemmon, Micheal L., Book-to-Market Equity, Distress Risk, and Stock Returns, 2002

Gerantonis, Nikolaos, Vergos, Konstantinos and Christopoulos, Apostolos G., Can Altman Z-score Models predict business failures in Greece, 2009

Hillegeist, S., Keating, E., Cram, D., Lundstedt, K., 2004. Assessing the probability of default. Re-view of Accounting Studies 9, 5–34.

Merton, Robert C., 1974, On the pricing of corporate debt: The risk structure of interest rates, The Journal of Finance 29, 449-479

Olhson, James A., 1980, Financial Ratios and the Probabilistic Prediction of Bankruptcy, Journal of Accounting Research 18, 109-131

Shumway, T., 2001, Forecasting Bankruptcy More Accurately: A Simple Hazard Model, The Journal of Business 74, 101-124

University of Exeter Business School

(http://businessschool.exeter.ac.uk/research/areas/centres/xfi/research/famafrench/fil es/)

Vassalou, M., Xing, Y., 2004. Default risk in equity returns. The Journal of Finance 59, 831–868.

Zmijewski, M. (1984). ‘‘Methodological Issues Related to the Estimation of Financial Distress Prediction Models.’’ Journal of Accounting Research Supplement, 59–86.

Corporate default risk and stock returns : evidence from the UK