Evaluation of Financial Distress Costs of Innovative Companies 37

Российский журнал менеджмента Russian Management Journal

16 (1): 37–62 (2018) 16 (1): 37–62 (2018)

Evaluation of Financial Distress Costs

of  Innovative Companies
E. Y. Makeeva*
National Research University Higher School of Economics, Russiaa
M. O. Khugaeva
Financial University under the Government of the Russian Federation, Russiab

Over the last few decades investments in research and development (R&D) projects have
risen significantly. Many experts have found a positive relationship between R&D expenses
and a company’s value. However, investments in innovations are characterized by a high level
of uncertainty. Often innovations generate income only in long run. These factors increase the
probability of firms’ financial distress considerably. Thus, the aim of the present paper is to
evaluate financial distress costs at companies that invest in research and development proj-
ects. The study covers 389 innovative firms over the period from 2006 to 2015. The analysis is
based on a direct and indirect costs evaluation. The first are estimated through panel logistic
regression. Indirect costs are calculated as unexpected losses or profits relative to industry
indicators. The paper confirms the existence of relation between R&D expenses and financial
instability. The results of the research show that the total financial distress costs of innovative
companies amount to 7.95% of total assets or 11.78% of the value of a company on average.
Keywords: financial distress costs, bankruptcy, innovative companies, panel logistic regression.
JEL: G32, G33.

One of the most important factors determin- that a huge amount of investments are fo-
ing a firm’s profitable growth is scientific cused on research and development projects
and technological progress. Innovations are in different sectors of the economy such
vitally important for companies to compete as software and programming, biotechno-
with one another. Global statistics shows logical products, capital goods, beverages,
* Corresponding author: [email protected]
a
Postal Address: 20 Myasnitskaya ul., National Research University Higher School of Economics, Moscow,
101000, Russia.
b
Postal Address: 49 Leningradsky prosp., Financial University under Government of Russian Federation,
Moscow, 125993, Russia.
© E. Y. Makeeva, M. O. Khugaeva, 2018
https://doi.org/10.21638/11701/spbu18.2018.102
38 E. Y. Makeeva, M. O. Khugaeva

accessories, restaurants, retail, hotels and Empirical studies outline two types of
motels, and so on. financial distress costs: direct and indirect.
However, R&D investments are charac- Direct costs include different administra-
terized by high uncertainty, so not every tive costs connected with the bankruptcy,
company can afford such risky investments. reorganization, liquidation process, or other
Most small and medium enterprises (SMEs) bankruptcy costs. Indirect costs are defined
are examples of the latter type. Besides ex- as alternative costs or missed opportunities.
ternal uncontrollable factors such as crises, The paper [Warner, 1977] was one of the
disasters, political instability, and armed first papers devoted to financial distress
conflicts, each firm has its own internal costs evaluation. The author examines direct
problems. Taken together this can lead to costs at 11 railroad companies in the US
financial distress or even failure of a firm, in 1933–1955 using the following method.
and each additional risky asset can increase He takes expenditures connected with the
the probability of insolvency. bankruptcy process and divides this figure by
Thus, the relevance of the present paper the amount of a firm’s total market value.
is that each company has to find the optimal However, the important assumption was
tradeoff between investments in innovations that a company’s value is equal to equity
and risk of financial distress. The solution and debt issued in the seven years preced-
to this problem can improve a company’s ing the bankruptcy. The results of Warner’s
efficiency and lead to its growth. And vice research showed that direct financial costs
versa, an unsuccessful selection can result are equal to 1.0–5.3% of firm’s value on
even in the failure of a firm. average (depending on the chosen base year).
The purpose of the present paper is to Therefore, it was concluded that these costs
evaluate financial distress costs of innovative do not influence financial decisions as they
companies. The subject of the investigation are much smaller than tax benefits.
is financial distress costs. Our study focuses The studies followed after [Warner, 1977]
on companies that spent at least $200 mil- did not yield significant results for capital
lion on research and development in 2015. structure decisions. The question arises: if
Data are collected from Bloomberg and the distress costs are not important, then why
financial reports of companies. The novelty do firms not use debt financing alone? Thus
of the present paper is in suggested model the experts decided to find the indirect costs
for direct and indirect cost evaluation of of financial distress. The first expert who
innovative companies. tried to calculate them was E. Altman. In
[Altman, 1984] he tried to find the optimal
Literature Review capital structure by comparing bankruptcy
costs with the benefits provided by financial
Academic papers define financial distress leverage and, in the case when the present
as the condition experienced by a company value of insolvency is higher than tax gains,
that cannot pay its obligations to finance its the company has to decrease its debt to eq-
debt. Analysts differentiate it from bank- uity ratio. The results of Altman’s analy-
ruptcy, explaining this by the fact that sis show that bankruptcy costs matter: in
bankruptcy is one of the outcomes of insol- the three years preceding bankruptcy, total
vency. However, the latest financial crisis bankruptcy costs were equal to 12.4–16.7%
showed that a corporation can go bankrupt of the firm’s value on average. Also, the au-
very unexpectedly. That is why it is neces- thor finds that present value of bankruptcy
sary to be on the lookout even for “healthy” costs is higher than the present value of
companies and observe their financial dis- tax benefits at most of studied firms and,
tress costs in order to take action before consequently, the firms’ debt level is much
their failure. higher than their optimal one.
RMJ 16 (1): 37–62 (2018)
Evaluation of Financial Distress Costs of Innovative Companies 39

The paper [Thorburn, 2000] examined total financial distress costs were equal to
263 Swedish bankrupt firms. The author 15.9–22.4% of a company’s value. This fact
estimates direct bankruptcy costs (such as also supports the hypothesis that distress
lawyer, consulting, & administrative costs) costs matter. Later, in [Bhabra, Yao, 2011],
and finds that these costs represent 6.4% the US bankrupt firms in 1997–2004 were
of pre-filing assets on average, and 19.1% examined. It was found that the costs of
for bankruptcy auctions. The results are financial distress divided by a company’s
significant, although the author only ex- value amounted 6.09%, 9.71% and 17.43%
amines the direct costs of financial dis- relative to the year prior to bankruptcy. For
tress. Another view on distress costs was the developing markets [Wijantini, 2007]
put forward by [Opler, Titman, 1994]. They covered Indonesian financially distressed
tried to analyze the link between insolvency companies in the period from 1997 to 2002.
and corporate performance in distressed The found mean value of indirect costs was
industries rather than individual distressed equal to 3–11% of total sales. Recently,
companies. The authors compared whether [Bu­lot, Salamudin, Abdoh, 2014] tried to es-
highly leveraged companies lose more oper- timate indirect costs of financial distress in
ating profit than more conservative ones. the trading and services sector in Malaysia.
Unfortunately, Opler and Titman did not The analysis finds that indirect costs are sig-
evaluate the costs of financial distress, but nificant and amounted 3.1–21.39% of firms’
they did come to the conclusion that con- value. Malaysian firms were also examined in
nection between debt ratio and performance [Bulot, Salamudin, Aziz, 2017]. The analysis
is stronger for companies with huge R&D of 190 financially distressed companies in
investments and in the case of more con- 2001–2011 revealed that indirect distress
centrated industries. costs are equal to 21.6% on average.
The idea of financial distress costs anal- Another method of measuring indirect
ysis, without any reference to economic costs of financial distress was used in [Hor­
costs, was developed in the paper [Andrade, taçsu et al., 2013]. The authors analyze all
Kaplan, 1998]. They analyze 31 financially public automobile companies in the period
distressed, highly leveraged transactions between 1985 and 2010. They examine in-
(HLT) and find that the net costs of financial direct financial costs using credit default
distress amount about 10–20% of firm’s swap spreads and find that marginal indirect
value. These results explain why firms be- costs are equal to 2–39% of firm’s value.
have themselves quite conservatively. The researchers paid attention to Ford and
Although the aforementioned papers are General Motors, and compared their distress
rather old, there are many studies that use costs to tax benefits. The analysis showed
the methodologies proposed there. For exam- that indirect costs in some years exceeded
ple, the indirect costs calculation approach tax benefits.
of Altman was recently repeated by [Kwansa, Many papers devoted to direct financial
Cho, 1995; Pham, Chow, 1989; Bhabra, Yao, costs are based on another Altman study,
2011; Bulot, Salamudin, Abdoh, 2014] and the Z-score model. The article was written
others. in 1968, but it remains very popular still.
In [Kwansa, Cho, 1995] authors covered The aim of the paper was to develop a model
bankrupt restaurant firms’ indirect costs that can predict a firm’s bankruptcy with
and found that they should be taken into the help of financial and economic ratios
account when making financial decisions [Altman, 1968].
(on average, 7% of firm’s value). In their The other method for calculating direct
turn, Pham & Chow (1989) analyzed bank- costs was used in [Mansi, Maxwell, Zhang,
rupt Australian companies in 1978–1983. 2012]. Their study covers 120 608 monthly
The authors came to the conclusion that observations on 1752  companies during the
RMJ 16 (1): 37–62 (2018)
40 E. Y. Makeeva, M. O. Khugaeva

period from 1980 to 2006. The researchers ability through a logistic regression model.
made an effort to determine bankruptcy The sample consisted of 105 bankrupt com-
probability according to firms’ bond ratings. panies and 2058  non-bankrupt firms during
The results showed average figures for AA, the period between 1970 and 1976. Ohlson
A, BBB, BB and B rated companies being tested nine variables and only four of the
equal to 0.206, 0.255, 0.353, 0.433 and characteristics showed significant results:
0.535, respectively. size, financial structure, performance, and
In [Zhang, 2015] the author also uses current liability.
Z-score. He examines whether there is any There are many papers that based their
connection between R&D investments and research on Ohlson’s methodology. One of
a company’s distress risk measured by Alt­ them is [Ho et al., 2013]. The authors ana-
man’s Z-score. The results show a positive lyzed 122  North American public companies
relationship, which is especially high dur- in the period from 1990 to 2009. The sam-
ing economic downturns [Zhang, 2015]. The ple includes 12 bankrupt firms. Two models
author introduces and estimates four models were proposed. The first one includes nine
with four R&D indicators: R&D expenses variables according to Ohlson’s methodology
scaled by total assets, sales, number of em- and the second one includes nine factors,
ployees, and R&D capital divided by total and also real GDP growth and the real inter-
assets. The present paper refers to Zhang’s est rate of two-year Treasury bonds as mac-
R&D index evaluation for further investiga- roeconomic proxies. As expected, the ex-
tion of distress risk costs in the innovative tended model predicts bankruptcy better
sector. than the original.
Findings of all covered studies are briefly Recent papers also show a preference
described in Table 1. for binary choice models, and the results
In order to estimate direct and indirect give a very high level of prediction. For
costs, analysts used different methodologies. example, [Jones, Hensher, 2004] analyzed
Altman in his paper of 1968 estimated direct 3032 firm-year observations in the period
financial distress costs through bankruptcy. between 1996 and 2000. The authors used
He used multiple discriminant analysis to a mixed logit model and came to the con-
determine variables that influence the like- clusion that the model has a forecasting
lihood of insolvency. This investigation has accuracy of 99.16% on the pooled data. In
disadvantages: it involves a firm’s charac- [Laitinen, Suvas, 2016] authors gathered
teristics, but does not consider country- a large sample of 1 255 768 non-bankrupt
specific indicators. That is why it should and 22 594 bankrupt firm-year observa-
not be used for developing countries. Be­ tions from 26  countries during 2007–2010.
sides, it is not very good for studying long- The authors examined the influence of Hof­
term forecasts. stede’s cultural characteristics on financial
After Altman’s paper, multiple discrim- distress prediction and found that some
inant analysis became very popular for bank- of them demonstrate a significant impact.
ruptcy prediction. Many modifications of Russian researchers also prefer to use
the Z-score equation were developed with this methodology. In [Zhdanov, Afanasyeva,
different independent variables; their num- 2011] the aviation industry was examin­
ber amounted to 2–47. However, multiple ed. The authors analyze 20 non-bankrupt
discriminant analysis has a huge drawback: and 20  bankrupt companies in 2001–2010.
there is no opportunity to predict the prob- Forty variables are divided into four catego-
ability of bankruptcy for firms that are in ries: profitability, financial sustainability,
the “grey area”. business activity, and liquidity. The re-
Another method was suggested in [Ohlson, sults show that the model accuracy for the
1980]. He tried to estimate bankruptcy prob- Russian aviation industry amounts 75%.
RMJ 16 (1): 37–62 (2018)
Evaluation of Financial Distress Costs of Innovative Companies 41

Table 1
Summary of financial distress costs’ studies

Author Data Period Results

Direct costs
Warner (1977) 11 US railroad firms 1933–1955 1.0–5.3% of firm’s value on average
Altman (1984) 18 US firms: 11 retail- 1974–1978 8.1–10.5% of firm’s value
ers, 7 other industrial
firms
Pham, Chow 55 Australian firms September On average 2.5% and 3.6% of firm value at
(1989) 1978–May the year before bankruptcy and in the year
1983 when bankruptcy occurred
Weiss (1990) 37 US firms November On average from 1% to 6.6% of market capi-
1979–De­cem­ talization of shares
ber 1986
Andrade, Kaplan 31 financially distressed 1980–1989 10–20% of firm’s value
(1998) firms with highly lever-
aged transactions
Thorburn (2000) 263 Swedish firms 1988–1991 6.4% of pre-filing assets on average, and
19.1% for bankruptcy auctions
LoPucki, 48 US firms 1998–2002 1.4% of firm’s assets
Doherty (2004)
Bris, Welch, 286 firms in Arizona 1995–2001 8.1% of firm’s pre-bankruptcy assets on
Zhu (2006) and New York average after liquidation; 1.6% of firm’s
pre-bankruptcy assets on average after reor-
ganization
Mansi, Maxwell, 1752 companies 1980–2006 Bankruptcy probability for AA, A, BBB,
Zhang (2012) BB, and B bonds on average is 0.206, 0.255,
0.353, 0.433, and 0.535, respectively
Charalambakis, UK and Indian compa- 1980–2011 Expected default frequency for UK market is
Garrett (2016) nies equal to 1%, and for Indian firms — 2%
Indirect costs
Altman (1984) 18 US firms: 11 retail- 1974–1978 4.3–6.2% of firm’s value
ers, 7 other industrial
firms
Pham, Chow 55 Australian firms September On average 13.4% and 18.7% of firm value
(1989) 1978–May at the year before bankruptcy and in the
1983 year when bankruptcy occurred
Kwansa, Cho US restaurant firms 1980–1992 7% of firm’s value on average
(1995)
Wijantini (2007) 28 Indonesian firms 1997–2002 3–11% of total sales on average
Bhabra, Yao 62 bankrupt firms in 1997–2004 2%, 6.2% and 14.9% of firm value in the
(2011) USA year prior to bankruptcy
Hortaçsu et al. Automobile manufac- 1985–2010 2–39% of firm’s value
(2013) turers in the US
Bulot, 48 Malaysian listed 15 February 3.1–21.39% of firms’ value
Salamudin, companies from trad- 2001–31  De­­
Abdoh (2014) ing and services sector cember 2011
Bulot et al. (2017) 190 Malaysian firms 2001–2011 21.6% of total sales on average

RMJ 16 (1): 37–62 (2018)

42 E. Y. Makeeva, M. O. Khugaeva

Table 2
Advantages and disadvantages of different methods for evaluating the probability of bankruptcy

Method Advantages Disadvantages

Multiple Simple calculation. Assumptions of normality.
Discriminant Use of several accounting indicators simultane- “Grey zone”.
Analysis ously Complicated interpretation of results
Binary High level of accuracy. Does not work well with a large number
Model Simple calculation. of categorical variables.
Ability to take into account industry specific Multicollinearity can lead to overestima-
factors tion of results
Neural Does not take non-normality or correlation Sophisticated methodology/code writing.
Networks between variables into account. Too sensitive for chosen sample.
Works with incomplete samples or those that Complicated interpretation of results
are not clearly defined

Another interesting paper is [Fedorova, total costs of misclassification, and the lo-
Gilenko, Dovzhenko, 2013], which studied gistic regression and neural networks yielded
the Russian manufacturing industry. The the poorest results.
sample involves 3001 non-bankrupt and The paper [Makeeva, Bakurova, 2012]
504 bankrupt firms during the period from presents the prediction of bankruptcy at
2007 to 2011. Both logit-regression method European oil & gas companies in the period
and probit approach were used. However, from 2000 to 2010. The authors used two
the results show that the logit approach is approaches: logistic regression and neural
better suited. The accuracy of the model network. They came to the conclusion that
was estimated using two methods and dem- logistic regression does not work as well as
onstrated a predictive accuracy of 87.14% neural networks since the accuracy of mod-
and 84.7%, which proves the high predictive els is equal to 76% and 98%, respectively.
power of the model. Although the second method predicts bet-
Many research studies devoted to bank- ter, [Makeeva, Bakurova, 2012] emphasized
ruptcy probability prediction suggested an- that using this technique greatly depends
other methodology, i. e., the neural networks on the chosen sample, which can lead to
approach. This approach has advantages over biased results.
the others since it does not take non-nor- Unfortunately, there is no perfect meth-
mality or correlation between variables into od that suits every research study. So, in
account; it does not have the same limita- order to put together the pros and cons
tions possessed by a multiple discriminant of the most popular approaches we place
analysis or binary model when a sample is Table 2.
incomplete or not clearly defined. That is Empirical studies do not have many tech-
why some authors think that this technique niques for evaluating indirect financial costs.
can predict bankruptcy better then statisti- Probably the oldest approach was suggested
cal methodology. However, previous research in [Altman, 1984]. He estimated indirect
does not always prove it. For example, [Chen costs as the profit losses/gains relative to
et al., 2006] examined 940  non-bankrupt and industrial figures. It is important to mention
89 bankrupt Chinese firms between 1990 that this method was used by most research-
and 2003. The authors used four models ers who tried to evaluate indirect distress
to predict bankruptcy: linear discriminant costs to present. That is why the methodol-
analysis, logistic regression, decision trees, ogy of the present paper will follow it, too.
and a neural network. In order to find which A detailed, step-by step explanation will be
model works best, the authors estimated the provided later.
RMJ 16 (1): 37–62 (2018)
Evaluation of Financial Distress Costs of Innovative Companies 43

Hypotheses tive equity value. Ohlson suggests that the

coefficient may have different signs, so its
Against this presented background of theo- influence is indeterminate.
retical and empirical studies we can formu- It was suggested that direct financial
late the set of hypotheses. costs are estimated as the probability of
Performance characteristics of a firm. The bankruptcy multiplied by sales, which means
nine financial explanatory variables are taken that the probability of bankruptcy has di-
from Ohlson’s paper. The authors decided to rect impact on these costs. So, if the listed
forecast their influence on the probability of factors are significant in the likelihood of
bankruptcy in accordance with the original bankruptcy, they play the same role in an
paper [Ohlson, 1980]. Leverage is estimated evaluation of direct financial costs. Thus,
as the total debt to market capitalization. the following hypotheses are posited:
The higher the debt, the higher are the in- Hypothesis 1. Leverage has positive in-
terest expenses that negatively influence the fluence on the probability of bankruptcy.
interest coverage ratio. Dependent variable Hypothesis 2. The current liabilities to
is a dummy variable equal to 1 if interest current assets ratio has a positive influence
coverage is less than 1 and 0 otherwise. The on bankruptcy probability.
influence of current liabilities to current Hypothesis 3. The dummy for net income
assets and cash flow from operating activity has a positive impact on the probability of
to total debt has the same logic, so sign of bankruptcy.
these variables should be positive and nega- Hypothesis 4. The size of a company
tive, respectively. Working capital scaled by negatively influences the likelihood of bank-
total assets is expected to be negative, as the ruptcy.
numerator of the ratio is estimated as cur- Hypothesis 5. Working capital scaled by
rent assets minus current liabilities. Thus, total assets has negative influence on bank-
the higher the working capital, the lower ruptcy probability.
are the interest expenses, and the higher is Hypothesis 6. Return on assets has nega-
the interest coverage ratio, so the dummy tive impact on the probability of bankruptcy.
is not equal to 1. Hypothesis 7. Cash flow from operating
Variables that include net income have an activity to total debt has a negative influ-
impact on earnings before interest, taxes, ence on bankruptcy probability.
depreciation and amortization (EBITDA), Hypothesis 8. Net income growth has
which is the numerator of the interest cov- negative impact on the probability of bank-
erage ratio. Therefore, the return on assets ruptcy.
and changes in net income have a positive Market characteristics of a firm. Here
effect on the interest coverage ratio, but a book to market value of total assets and stock
negative impact on the dependent dummy. return volatility are analyzed. According to
The dummy for net income, if it was nega- [Zhang, 2015], they have a positive impact
tive over the last two years is equal to 1, upon the probability of distress.
and 0 otherwise. It should have a positive Hypothesis 9. The higher the book to mar-
impact as if it is equal to 1, then interest ket value of total assets, the higher are the
coverage is lower than 0, and the dependent direct financial distress costs.
dummy is equal to 1. The size of company Hypothesis 10. The higher the stock re-
is presented by a natural logarithm of total turn volatility, the higher are the direct
assets. Logically, the larger the company is, financial distress costs.
the lower is the likelihood of bankruptcy. Research & development characteristics
The dummy for the total debt to total of a firm. R&D expenses are very risky in-
assets ratio is included in the equation as vestments because of the uncertainty since
a correction for the companies with a nega- only a portion of R&D investments lead to
RMJ 16 (1): 37–62 (2018)
44 E. Y. Makeeva, M. O. Khugaeva

successful outcomes. That is why previous cyclical sector. The rest of the firms form
empirical papers have found a positive rela- about 20% of the total.
tionship between R&D expenses and financial It is important to mention that some
distress. companies from the sample appeared not
Hypothesis 11. The higher the R&D ex- so long ago, i. e., some of the firms are less
penses, the higher are the direct financial than 10 years old, which is why the sample
distress costs. is unbalanced.

Data Variables description

For further investigation, it is necessary
Sample to choose variables that can influence both
To evaluate financial distress costs of in- direct and indirect costs of financial dis-
novative companies, a sample comprised of tress. Direct costs are equal to bankruptcy
389 firms spending more than $200 million probability costs, and indirect costs are un-
a year on research and development in 2015 expected losses/profits.
was gathered. Information about companies Ohlson’s model was chosen as the basis
and industries was collected for the period for the methodology of the first part of the
from 2006 to 2015. Most data were taken present paper. The difference between the
from Bloomberg; the rest of indicators were present research and Ohlson’s paper is that
taken from the annual reports of the com- the first includes only non-bankrupt firms
panies themselves. while the second combines both bankrupt
The sample presents firms from 25  coun- and non-bankrupt companies. That is why
tries: 37% of the most innovative companies the interest coverage ratio was chosen as
are from the US, 19% from Japan, and 10% an indicator of financial distress. It is cal-
from China. The rest of the corporations culated according to the following formula:
are from Germany, South Korea, Switzer­
Interest covarage ratio =
land, Japan, Great Britain, France, Finland,
the Netherlands, Sweden, Ireland, Taiwan, Earnings before interest, taxes, (1)
Den­mark, Italy, Israel, Belgium, Spain, Ber­ depreciation and amortization
muda, Australia, Saudi Arabia Brazil, Rus­ = .
Interest expenses
sia, Canada, and India. The firms from both
developed and emerging economies are re­ Thus, dependent dummy variable y is
presented in the sample. The data is in mil- equal to 1 if the interest coverage ratio is
lions of US dollars, except stock price return lower than 1 and 0 otherwise.
and stock price volatility. Financial independent variables were tak-
The sample was compiled subject to the en from Ohlson’s paper. A more detailed
following limitation: a company’s R&D ex- explanation is given below in Table 3.
penses should have been more than $200  mil- The first factor is the Size. It is added
lion in 2015. So the data includes 389 in- in order to take into account the effect of
novative firms from nine industries: com- scale. The indicators that describe financial
munications; consumer, cyclical; technology; stability are financial Leverage, dummy vari-
consumer, non-cyclical; industrial; financial; ables that show whether debt exceeds total
energy; basic materials; and utilities. Most assets or not, and Cash flows provided by
of chosen companies operate in the industrial operations divided by total liabilities. The
sector, 88  holdings. A huge number of firms Dummy for leverage “serves as a discontinu-
are from technological and consumer non- ity correction for financial leverage” [Ohlson,
cyclical industries, 20% and 21%, respec- 1980]. Ohlson calculated leverage according
tively. Another 17% of corporations from to the total debt-to-total assets ratio, but
the sample are companies from the consumer this factor yielded insignificant results. So,
RMJ 16 (1): 37–62 (2018)
Evaluation of Financial Distress Costs of Innovative Companies 45

Table 3
Independent financial variables

Variable Label Measure

Size Size ln (Total  assets)
Leverage Leverage Total debt
Market capitalization of shares
Dummy for leverage Ldummy,  1 if Total debt exceeds Total assets,
L 
 0 if otherwise
Operating cash flow scaled by total debt Ocftl Operating cash flow
Total debt
Current ratio Clca Current liabilities
Current assets
Return on assets ROA Net income
Total assets
Working capital scaled by total assets Wcta Working capital
Total assets
Net income growth Nigrowth NI t − NI t −1
NI t + NI t −1
Dummy for net income Nn  1 if Net income was negative last two years,

 0 if otherwise

we decided to use the debt-to-market capi- add industry-specific characteristics and

talization ratio in this study. Liquidity is R&D investment ratios. Zhang suggested
presented by the Current ratio. Profitability four R&D characteristics, i. e., R&D expen­
ratios are represented by the Return on as­ ditures to total assets, to sales, and to the
sets and Working capital scaled by total as­ number of employees. The last indicator is
sets. Besides the listed factors, Ohlson ex- R&D capital to total assets [Zhang, 2015,
amined the following two factors, which are p. 98]. It is calculated as:
also taken into account in the present inves-
tigation, i. e., Net income growth and the RDCi, t = RDi, t + 0.8 ⋅ RDi, t −1 +
Dummy for net income and whether it was (3)
+ 0.6 ⋅ RDi, t −2 + 0.4 ⋅ RDi, t −3 + 0.2 ⋅ RDi, t −4 .
negative over the previous two years.
In addition to the chosen financial factors, These four factors are described in Table  4.
it was decided to examine variables accord- Thus, the analysis includes 15 indepen-
ing to [Zhang, 2015], which included more dent variables: financial and non-financial
firm characteristics and market indicators, firm characteristics as well as market and
stock return volatility and the book-to-mar- industry factors.
ket value of total assets, measured as:

BV of total assets Methodology

. (2)
BV of totalassets − Most papers devoted to evaluation of finan-
− (BV of equity + Market value of equity) cial distress costs use panel data in order
to obtain more accurate coefficients and
This paper is devoted to an examination clearer predictions. Today there are many
of the innovative firms, and we decided to proposed methodologies, but none can yield
RMJ 16 (1): 37–62 (2018)
46 E. Y. Makeeva, M. O. Khugaeva

Table 4
R&D indicators

Variable Label Measure

R&D expenditures scaled by total assets Rda R & D expenditures
Total assets
R&D expenditures scaled by sales Rds R & D expenditures
Sales
R&D expenditures scaled by number of employees Rde R & D expenditures
Number of employees
R&D capital scaled by total assets Rdcta R & D capital
Total assets

perfect results that contain each country, “grey zone” where the probability of bank-
industry or even firm. The most widespread ruptcy cannot be predicted. This is why the
approaches include multiple discriminant binary choice model, the logistic regression,
analysis, binary regressions, and neural net- was selected as the basis for this part of
works. After comparing the pros and cons the analysis.
of each method, we decided to opt for the Variable yi, t is the dependent variable
binary choice model in this investigation, of firm i in year t. It takes the value 1 if
in particular, a logit regression. Ohlson’s the interest coverage ratio is lower than 1
model modification will serve as the basis and 0 if otherwise. xi, t are the independent
for an evaluation of direct financial dis- variables. The function of the probability
tress costs, given that it seems necessary that yi, t takes a value of 1 is described by
to include macroeconomic and market vari- the formula (4):
ables in the analysis, as [Zhang, 2015] did.
Indirect financial distress costs will be es- (
P yi, t = 1 | xi, t , β, α i  = Λ α i + xi′, t β , (4) )
timated as profit losses/gains according to
Altman’s analysis. ( )
where Λ α i + xi′, t β is the logistic cumula-
y
Methodology of the evaluation of direct tive distribution function with Λ ( y) = e y
1+ e
financial costs [Cameron, 2005, p. 795]. Here yi, t can be
One of the assumptions of the present paper found as a function from the independent
is that direct distress financial costs are variables (5):
equal to bankruptcy costs, as there is no
common opinion on this term. yi, t = β′xi, t. (5)
A huge amount of previous empirical
studies have tried to find out which model The joint density function of company i is
predicts bankruptcy better than others, but
T
∏F (α i + xi′, t ⋅ β)
they did not reach a common conclusion. yi, t
f (yi | X i , β, α i ) = ×
In 2006, Aziz and Dar came to the conclu- t =1 (6)
sion that the most popular methods for
( ( ))
1− yi, t
calculating direct financial distress costs × 1 − F α i + xi′, t ⋅ β .
are statistical methods, i. e. multiple dis-
criminant analysis and binary choice mod- Furthermore, the analysis is based on
els are the most frequently used. However, panel data that gives an opportunity for
multiple discriminant analysis has a sig- using the extended version of the logit re-
nificant disadvantage — the existence of a gression, i. e., the panel logit regression,
RMJ 16 (1): 37–62 (2018)
Evaluation of Financial Distress Costs of Innovative Companies 47

which can be estimated through fixed effect After constructing the models, the one with
and random effect models. the most predictive power will be chosen.
The random effect binary model assumes It should be mentioned that
that the firm-specific effect is a random vari- 1) the probability of financial distress rises
able and that it is uncorrelated with the in- with y;
dependent variables. This approach supposes
that individual effects submit to a normal (
2) y = log P .
1− P )
distribution ε i, t  N (0, σ 2e ) , α i, t  N (0, σ 2α ) and As Ohlson remarks, these relations make
that the joint density function is given in the model easy to interpret [Ohlson, 1980,
the following formula: p. 118].
The quality of the models will be esti-
f (yi | X i , β, σ 2α ) = mated through the area under the receiver
−α2 (7) operating characteristic (ROC) curve. If it
1 e 2σ2α dα .
i

= ∫ (yi | X i , β, α i ) i
is more than 70%, then such a model has
2πσ 2α a high level of predictive power. So, the
models will be compared according to this
The fixed effect model assumes that firm- criterion.
specific effect is a random factor, but in In order to evaluate the direct financial
contrast to random effect model, it can be distress costs the following assumption was
correlated with independent factors. The taken: the derived probability should be
joint density for firm i of this model is multiplied by the sales of each firm-year
observation. The logic here is that if the
⋅ e (∑t i, t i, t ) . (8)
αi ∑ yi, t y ⋅ x′ ⋅β
f (yi | X i , β, α i ) = e
t bankruptcy happens, then sales will be equal
∏ t  1 + e αi + xi′, t ⋅β  to zero with the probability:

Thus, the functions for the dependent DDCi, t =

(10)
variable look as follows = Probability of bancruptcy′⋅ Salesi, t .
1) yi, t = α + xi, t ⋅ β + (ui + ϑi, t ) for the random
effect model; Methodology for the evaluation of indirect
2) yi, t = (α + ui ) + xi, t ⋅ β + ϑi, t for the fixed ef- financial costs
fect model. The next step of the methodology is evalu-
In our case, the dependent variable is ation of indirect financial costs. Previous
described by 12 independent factors: empirical research does not give such a wide
range of choices as in the case of direct
yi, t = α i + β1 ⋅ Sizei, t + β2 ⋅ Leveragei, t + costs. Intuitively the indirect costs of fi-
+ β 3 ⋅ Li, t + β 4 ⋅ Ocftdi, t + β5 ⋅ Clcai, t + nancial distress reflect unexpected profit or
losses. The first paper where it was esti-
+ β6 ⋅ ROAi, t + β7 ⋅ Wctai, t + β 8 ⋅ Nigrowthi, t + (9) mated was [Altman, 1984]. Nevertheless,
+ β 9 ⋅ Nni, t + β100 ⋅ Bmi, t + β11 ⋅ Voli, t + it is a relatively old approach that is still
popular among analysts.
+ β12 ⋅ R&D indicatori, t + ε i, t . For a better understanding, a more de-
tailed explanation is given below step by
The logic is that the original Ohlson model step. The sales of the firm were regressed on
will be regressed, and after that, market and total industry sales for the 10 prior years:
industry factors will be added one by one.
Special attention should be paid to indus- Si, t = α + β ⋅ SI , t , t = 10 years, (11)
try variables, i. e., four R&D expenditures
indicators are included, but the four models where Si, t — sales of firm i in period t; SI , t —
will be evaluated with each factor separately. aggregate sales of industry I in period t.
RMJ 16 (1): 37–62 (2018)
48 E. Y. Makeeva, M. O. Khugaeva

In this case, the sales of firm i at the mo- Unfortunately, there are many outliers.
ment t are the firm’s sales in 2015. Industry In order to get unbiased results they were
sales over previous 10 years are the sales of dropped from the sample. Final descriptive
the global industry in 2005–2014. statistics of the data is given in Table 5.
After that, Altman included aggregate Table 5 contains information about mean
industry sales and obtained estimated com- value, standard deviation, minimum and maxi-
pany sales: mum values of the variables. The overall num-
ber of observations is equal to 3714 and 3326
^ = α + β ⋅ S . (12)
S for net income growth. There is huge interval
i, t I, t
between minimum and maximum value for
Then the author suggests calculating aver- the factors. It can be explained by the 10-year
age profit margin for sales over the 10  years period of analysis that includes not only the
prior to the period under consideration. global financial crisis of 2007, separate re-
After that, if we multiply estimated sales cessions in different countries, but also the
of the firm by sales margin, we will obtain specific features of the innovative companies
estimated profit: since such firms invest a lot in R&D projects.
Special attention should be paid to the
P^i, t = S^ i, t ⋅ PM . (13) dependent variable y. It has a mean value
of 0.051 with a standard deviation equal to
In order to find the indirect costs, we 0.22. For a better understanding of what
have to compare real profit with estimated happens to the sample, the Table  6 was con-
profit, i. e., the difference between them is structed in Stata 13.0.
equal to indirect distress costs [Altman, It can be concluded that if a firm was
1984, p. 1073]: not in distress in the previous year, it will
be distressed this year with the probability
^i, t. (14)
IDC = ∆Pi, t = Pi, t − P 2.02%, and with a 97.98% chance that it
will become healthy. At the same time, the
Thus, for each firm we calculate indi- probability that a company will be distressed
rect and direct costs of financial distress this year if it was distressed last year is
for 2015. The total value of distress costs 62.2%, with a 37.8% chance that it will
is equal to the sum of these two values: become healthy.
The variables were also checked for mul-
TDC = DDC + IDC . (15) ticollinearity, and a VIF was constructed for
this purpose. It is lower than 5, so it can be
Results concluded that there is no multicollinearity
between the factors (see Appendix).
Descriptive statistics The second part of the research performs
Before moving to our calculations, it is nec- a calculation of unexpected profit or losses
essary to describe the collected data and as it was suggested by Altman. The follow-
check variables for multicollinearity. ing variables are included: a firm’s sales
The initial number of observation is equal for 2015, the aggregate industry sales for
to 3729 for 389  companies, so the sample is 2005–2014, the average profit margin in
unbalanced. There are firms in the sample 2006–2015 for each company, and net in-
that were founded between 2006 and 2015, come in 2015. All factors are in absolute
and they have fewer observations present in terms, except for average profit margin,
the sample. The calculation of net income which is calculated as the sum of net in-
growth also limits the number of observa- come scaled by sales during the 10 years
tions as they are computed with indicators and divided by 10. Descriptive statistics is
from previous years. presented in Table 7.
RMJ 16 (1): 37–62 (2018)
Evaluation of Financial Distress Costs of Innovative Companies 49

Table 5
Descriptive statistics for direct distress costs model

Variable Observations Mean Std. Dev. Min Max

y 3714 0.05 0.22 0 1
Size 3714 9.67 1.32 7.17 12.02
Leverage 3714 0.44 1.07 0 27.98
Ldummy 3714 0.11 0.31 0 1
Wcta 3714 0.18 0.16 –0.07 0.54
Clca 3714 0.65 0.28 0.2 1.23
ROA 3714 0.06 0.06 –0.06 0.17
Ocftl 3714 1.86 4.18 –0.01 18.25
Nn 3714 0.05 0.21 0 1
Nigrowth 3326 0.02 0.44 –1 1
Vol 3714 0.36 0.13 0 0.65
Bm 3714 0.66 0.27 0.21 1.11
Rda 3714 0.06 0.05 0 0.19
Rds 3714 0.09 0.08 0 0.31
Rde 3714 0.04 0.04 0 0.16
Rdcta 3714 0.12 0.13 0 0.45

Table 6
Dependent variable description

y 0 1 Total
0 97.98  2.02 100
1 37.8 62.2 100
Total 95.01  4.99 100
Table 7
Descriptive statistics for indirect distress costs model

Variable N Mean Std. Dev. Min Max

Sales 389 30 493.49 45 317.72 41.1 283 934.3
Industry 389 49 192.7 19 413.8 21 609.2 76 061.6
Averagepm 389 0.09 0.12 –0.62 0.69
Ni 389 2032.97 4407.57 –13 494.52 53 394

Companies’ sales and net income have 2015, even in such giants as Volkswagen,
very high standard deviation, but the es- General Electric, Tesla, Yahoo! and so on.
timation of indirect distress costs is con- This explains why the minimum average
ducted for each firm separately, so there is profit margin is negative.
no need to drop outliers. As stated above,
the firms from the sample operate in nine Evaluation of direct distress costs
industries. Sectors’ aggregate sales for the Estimation of direct financial distress costs
10  years prior to 2015 range from $21.6  tril- is based on panel data analysis. Overall pan-
lion to $76.1 trillion. It is notable that the el data has advantages over the other types
net income was negative at 42 companies in of data as it gives an opportunity to take
RMJ 16 (1): 37–62 (2018)
50 E. Y. Makeeva, M. O. Khugaeva

Table 8
Initial nine variables: pooled, fixed effect and random effect regressions results

Model
Variable
Pooled Fixed effect Random effect
Size –0.4*** 0.19 –0.49**
(–3.46) (0.42) (–2.17)
Leverage 0.77 0.06 0.1
(1.33) (0.18) (0.94)
Ldummy –0.32 2.49** 0.42
(–0.92) (2.17) (0.68)
Ocftl –0.23** –1.35* –0.34**
(–2.51) (–1.68) (–2.53)
Clca 4.14** 2.41 5.83*
(2.49) (0.44) (1.83)
ROA –37.03*** –55.23*** –58.35***
(–10.23) (–5.44) (–7.57)
Wcta 0.11 –3.78 –0.61
(0.11) (–1.13) (–0.33)
Nigrowth –0.14 –0.1 –0.6
(–0.58) (–0.26) (–0.18)
Nn 0.74** –0.19 0.55
(2.22) (–0.29) (1.04)
Constant 0.21 –0.73
(0.14) (–0.25)
Observations 3326 528 3326
Pseudo R2 0.549
LR chi2(9) 724.04 212.73
Wald chi2(9) 88.06
Prob  > chi2 0.0000 0.0000 0.0000
N o t e: Z-statistics based on standard errors are in parentheses;
*, ** and *** indicate statistical significance at the 10%, 5% and 1% levels, respectively.

into account individual differences of enti- The basic Ohlson model is estimated with
ties during various periods of time. random and fixed effect models, and then the
There are two main models that analyze one best suited for estimating the probability
the panel data: of financial distress is chosen. After that,
1) random effect regression; industry and market variables are included
2) fixed effect regression. one by one in this chosen model. However,
The crucial difference between these two before the panel data analysis, it is neces-
models is in unobserved firms’ effect (it in- sary to run a pooled regression. The pooled
cludes factors that correlate with the regres- regression perceives the sample not as panel
sors). The fixed effect model supposes that data but as a cross-section, i. e., it does not
there can be a connection, but the random take the individual characteristic of an en-
effect model assumes that unobserved ef- tity that changes over a period of time into
fects do not have such relationships with account.
other variables and that they only influence The outcomes of the three regressions
residuals. are given in Table 8.
RMJ 16 (1): 37–62 (2018)
Evaluation of Financial Distress Costs of Innovative Companies 51

All three models are significant as the and Leverage became significant at a 10%
chi-square is lower than 0.05; the hypoth- level. Fixed effect models show the causes
esis that their coefficients are equal to 0 of variation within a company, and as men-
was rejected. In the pooled regression, the tioned above, some companies were founded
coefficients behind Size, operating cash flow during the observation period, which reduces
scaled by total debt (Ocftl), current ratio the size of the sample. This is perhaps why
(Clca), return on assets (ROA), dummy for the coefficients do not demonstrate signifi-
net income (Nn) were significant at 5%, cant values. The random effect model also
Leverage at 20%. The rest of coefficients did not change, except in that Net income
were not significant even at a 30% level. growth sign varies, but its coefficient yields
This model yields more significant coeffi- an insignificant result.
cients than the other two. Thus, it is necessary to determine which
The random effect model shows almost model is better than the other two. To start
the same meaningful coefficients, except with, the fixed effect model should be com-
the dummy variable for net income, which pared with the random effect model and with
nevertheless has the same positive influence. the pooled regression. It seems appropriate
Ldummy and Wcta change their signs, but to use the Hausman test for this purpose.
remain insignificant. The null hypothesis suggests the random
While running the fixed effect model, effect model, and if the null hypothesis is
the number of observations and groups was rejected, then the fixed effect model will be
dropped by the system, obviously because of preferred as it is better suited. An estima-
unbalanced sample. The model shows two tion of the fixed effect model vs. the random
significant factors, i. e. the Ldummy and effect model is presented in Table 10. The
ROA at a 5% significance level. Besides, Hausman test did not yield any statistical
Ocftl is significant at a 10% level. result, as chi square is equal to –97.46,
Overall, it was unexpected that the other which is less than 0.
factors would yield insignificant results, but The next pair that should be compared is
in order to overcome this, we decided to the fixed effect model vs. the pooled regres-
adjust the models with dummies for years. sion. It is also evaluated with the Hausman
The results are presented in Table 9. test. The outcome is presented in Table 11.
Although all three models are significant As a consequence of the Hausman test, the
at a 1% level, the situation did not change pooled regression is more appropriate than
noticeably. The dummies for years yielded fixed effect model. The p-value is much
significant results in 2012, 2014, and 2015. higher than 0.05, so the null hypothesis,
Year of 2012 can be explained by the results that the pooled model is preferred, was not
of the global financial crisis. Most interest- rejected. Notably, both pairs of tests, the
ing are 2014 and 2015 as they show that the fixed/random and fixed/pooled, showed that
financial results for the previous two years the fixed effect model is worse. This is prob-
had a positive influence on the probability of ably due to the unbalanced sample.
financial distress. Models demonstrate that The last pair that should be examined is
the coefficients behind Size, Ocftl, ROA, the random effect model and pooled regres-
2014, and 2015 are significant. That means sion. When estimating the random effect
that the more assets a firm has, the lower is model, we run a test, whether the fraction
the probability of financial distress. Return of dependent variable’s variance due to er-
on assets and Operating cash flows scaled by ror between entities is significant or not.
total debt have the same sign and influence This fraction is called rho. So, if it is not
the dependent variable adversely. equal to 0, then the random effect model is
The pooled regression pseudo R-squared preferred over the pooled regression. In this
increased slightly, from 54.9% to 56.23%, study, it can be concluded that the random
RMJ 16 (1): 37–62 (2018)
52 E. Y. Makeeva, M. O. Khugaeva

Table 9
Initial nine variables: corrected pooled, fixed effect and random effect regression results

Model
Variable
Pooled Fixed effect Random effect
Size –0.47*** –0.43 –0.77***
(–3.94) (–0.75) (–2.94)
Leverage 0.1* 0.06 0.17
(1.89) (0.28) (1.53)
Ldummy –0.19 3.28** 0.76
(–0.52) (2.44) (1.10)
Ocftl –0.22** –1.43* –0.34***
(–2.54) (–1.70) (–2.57)
Clca 4.25** 2.7 6.62*
(2.54) (0.44) (1.93)
ROA –39.41*** –61.53*** –64.15***
(–10.23) (–5.18) (–7.2)
Wcta 0.22 –3.49 –0.13
(0.22) (–0.93) (–0.06)
Nigrowth –0.14 –0.11 0.12
(–0.57) (–0.23) (–0.33)
Nn 0.55 –0.68 0.21
(1.52) (–0.88) (0.34)
2008 –0.21 0.01 –0.52
(–0.35) (0.01) (–0.59)
2009 0.74 1.99* 1.33
(1.29) (1.84) (1.56)
2010 0.83 1.95* 1.51
(1.26) (1.72) (1.57)
2011 0.5 1.53 1.13
(0.81) (1.28) (1.20)
2012 1.09* 2.05* 1.61*
(1.91) (1.89) (1.85)
2013 0.49 1.05 0.72
(0.78) (0.9) (0.77)
2014 1.3** 2.32* 2.09**
(2.14) (1.95) (2.24)
2015 1.17** 1.78 1.82**
(1.99) (1.56) (2.02)
Constant 0.3 0.08
(0.02) (0.02)
Observations 3326 528 3326
2
Pseudo R 0.5623
LR chi2(9) 741.49 225.40
Wald chi2(9) 73.71
Prob  > chi 2
0.0000 0.0000 0.0000
N o t e: Z-statistics based on standard errors are in parentheses;
*, ** and *** indicate statistical significance at the 10%, 5% and 1% levels, respectively.

RMJ 16 (1): 37–62 (2018)

Evaluation of Financial Distress Costs of Innovative Companies 53

Table 10
Fixed effect vs. random effect models, Hausman test

Variable Difference between random and fixed effect models

Size 0.34
Leverage –0.1
Ldummy 2.5
Ocftl –1.09
Clca –3.92
ROA 2.62
Wcta –3.37
Nigrowth 0.1
Nn –0.88
2008 0.53
2009 0.66
2010 0.45
2011 0.4
2012 0.45
2013 0.33
2014 0.23
2015 –0.4
Chi2(17)  = –97.46 < 0

Table 11
Fixed effect model vs. pooled regression, Hausman test

Variable Difference between random and fixed effect models

Size 0.04
Leverage –0.04
Ldummy 3.46
Ocftl –1.2
Clca –1.58
ROA –22.11
Wcta –3.72
Nigrowth 0.03
Nn –1.23
2008 0.22
2009 1.25
2010 1.12
2011 1.03
2012 0.96
2013 0.56
2014 1.02
2015 0.62
Chi2(17)  = 16.64
Prob  > chi2  = 0.4788

RMJ 16 (1): 37–62 (2018)

54 E. Y. Makeeva, M. O. Khugaeva

Table 12 effect model is better than pooled regres-

Random effect model with a robustness check sion, as the hypothesis of rho equal to  0 was
rejected at a 1% significance level.
Variable Coefficients Therefore, if the pooled regression is
Size –0.77*** better than the fixed model, and random
(–3.29) is better than pooled, it can be concluded
Leverage 0.17* that the random effect model is the best.
(1.72) However, it seems unforeseeable that some
Ldummy 0.76 variables have insignificant coefficients with
(1.06) a p-value higher than 0.1, for example, le-
Ocftl –0.34** verage should logically have an influence.
(–2.35) Perhaps it occurs due to the heteroscedastic-
Clca 6.62**
ity of the sample. That is why the random
(2.16) effect model is adjusted and standard errors
are clustered at the firm level to deal with
ROA –64.15***
(–6.00)
potential heteroscedasticity. The results of
the regression checked for robustness are
Wcta –0.13
presented in Table 12.
(–0.06)
The outcome of the random effect logit
Nigrowth –0.12 regression checked for robustness showed
(–0.30)
that Size, Ocftl, Clca and ROA are signifi-
Nn 0.21 cant. Leverage shows positive impact on the
(0.31) probability of distress at 10% significance
2008 –0.52 level.
(–0.81) The economic interpretation for the size
2009 1.33* of a company is the same as for the initial
(1.93) random effect model, i. e., the larger the
2010 1.51 company, the lower is the probability of
(1.62) bankruptcy. Leverage and the Ocftl ratios
2011 1.13 are based on the total debt level and show
(1.42) that the higher debt is, the more likely is
2012 1.61** insolvency. However, in Leverage debt is
(2.21) the numerator, so the impact of the debt to
2013 0.72 market capitalization ratio is positive, and
(0.85) Operating cash flows to total debt influences
the dependent variable in a negative man-
2014 2.09***
(3.08) ner. The current liability to current assets
ratio (Clca) has positive relationship with
2015 1.82**
the probability of bankruptcy. Return on
(2.49)
assets influences the likelihood of insolvency
Constant 0.08 negatively as the higher net income is, the
(0.03)
higher is the interest coverage ratio. Thus,
Observations 3326 the dummy for interest coverage yields 0
Wald chi (17)
2
73.71 value.
Prob  > chi2 0.0000 It should be mentioned that the dummy
variables for the years also yield significant
N o t e: Z-statistics based on standard errors are
in parentheses;
results in 2009, 2012, 2014, and 2015.
*, ** and *** indicate statistical significance at The next step of the research supposes
the 10%, 5% and 1% levels, respectively. adding market variables to the chosen model.
RMJ 16 (1): 37–62 (2018)
Evaluation of Financial Distress Costs of Innovative Companies 55

These include stock return volatility and the Table 13

total assets book to market value ratio. The Model with market variables
results are presented in Table 13.
The new added variables show significant Variable Coefficients
results, i. e., Bm at a 5% significance level Size –0.47*
(–1.74)
and Vol at 10%. It is also noticeable that
Leverage 0.17**
other significant factors’ p-value changed
(2.02)
only slightly, i. e., it rose for Size and Clca.
Ldummy 0.9
On the contrary, it fell for Leverage and (1.26)
Ocftl. Overall, at a 10% level the market Ocftl –0.34**
variables are significant, and there are sev- (–2.44)
en significant coefficients: Size, Leverage, Clca 4.9*
Ocftl, Clca, ROA, Vol, and Bm. (1.60)
The next part of the investigation is fo- ROA –61***
cused on the industry indicators, i. e., the (–5.77)
four R&D ratios. They are separately added Wcta –0.7
to the model. The results are compiled in (–0.35)
Table 14. Nigrowth –0.14
Three of the R&D indicators demonstrat- (–0.33)
ed a positive influence on the probability Nn –0.04
of financial distress, i. e., R&D to assets, (0.06)
R&D to sales, and R&D to the number of Vol 4.1*
employees. However, only the last two yield (1.82)
significant results. Model  4 highlighted the Bm –2.54**
meaningful negative coefficient behind the (–2.06)
innovative sector index. It can be explained 2008 –0.59
by the fact that this factor includes R&D (–0.74)
expenses for the previous four years, and the 2009 0.74
(0.92)
logic is the following: if a firm invests in
2010 1.74*
innovative project(s) for such a long period
(1.92)
of time, then it (they) must be profitable.
2011 1.39*
Intuitively, this is why the sign is negative.
(1.61)
Size has a significant influence only in
2012 1.63**
Model 4 and it has negative impact. Lever­ (2.21)
age, Ocftl, ROA, and Vol also de­mon­strated 2013 0.74
significant results in each model, unlike (0.84)
Ldummy, Wcta, Nigrowth, and Nn. Clca 2014 2.35***
variable is insignificant in Models 1 and 3, (3.26)
but in Models 2 and 4 it demonstrates a 2015 1.86**
positive influence upon the probability of (2.47)
insolvency. There is a negative relationship Constant –1.8
between Bm and the dependent variable in (–0.68)
Models 1 and 4. Observations 3326
Thus, there are four models with R&D Wald chi2(17) 82.53
indicators. However, even if Model 1 will Prob  > chi2 0.0000
not be taken into account given that its
N o t e: Z-statistics based on standard errors are
coefficient is insignificant, the model that in parentheses;
operates best and predicts the probability *, ** and *** indicate statistical significance at
of bankruptcy most accurately should be the 10%, 5% and 1% levels, respectively.

RMJ 16 (1): 37–62 (2018)

56 E. Y. Makeeva, M. O. Khugaeva

Table 14
Four models with R&D indicators

Coefficients
Variable
Model 1 Model 2 Model 3 Model 4
Size –0.378 –0.208 –0.407 –0.744**
(–1.36) (–0.75) (–1.54) (–2.30)
Leverage 0.173** 0.173* 0.156* 0.177**
(2.01) (1.91) (1.70) (2.00)
Ldummy 0.923 1.12 0.975 0.781
(1.26) (1.43) (1.32) (1.13)
Ocftl –0.351** –0.38** –0.359** –0.341**
(–2.43) (–2.59) (–2.53) (–2.46)
Clca 4.68 5.744* 4.592 5.233*
(1.53) (1.79) (1.53) (1.73)
ROA –60.41*** –59*** –60.048*** –64.003***
(–5.56) (–5.39) (–5.58) (–5.51)
Wcta –0.824 0.199 –0.513 –0.554
(–0.41) (0.10) (–0.25) (–0.29)
Nigrowth –0.144 –0.191 –0.143 –0.14
(–0.35) (–0.45) (–0.35) (–34)
Nn –0.066 –0.133 –0.109 0.286
(–0.10) (–0.20) (–0.16) (0.43)
Vol 4.183* 4.777** 4.314* 4.11*
(1.84) (2.03) (1.87) (1.87)
Bm –2.34* –1.523 –2.009 –2.678**
(–1.84) (–1.14) (–1.55) (–2.15)
Rda 5.14
(0.76)
Rds 12.553**
(2.54)
Rde 11.756*
(1.78)
Rdcta –5.543**
(–2.06)
Observations 3326 3326 3326 3326
Wald chi2 76.33 56.57 72.43 81.90
Prob  > chi2 0.0000 0.0000 0.0000 0.0000
N o t e: Z-statistics based on standard errors are in parentheses;
*, ** and *** indicate statistical significance at the 10%, 5% and 1% levels, respectively.

Table 15
ROC curve estimates

Model Model 1 Model 2 Model 3 Model 4

Area under LROC 0.9399 0.9365 0.9370 0.9452

chosen. In this case, a receiver operating and the area under this curve is cal­culated.
characteristics (ROC) curve is constructed The results are presented in Table 15.
RMJ 16 (1): 37–62 (2018)
Evaluation of Financial Distress Costs of Innovative Companies 57

Thus, it can be concluded that Model 4 Table 16

is the best one. It predicts insolvency with Insolvency prediction with Model 4
a probability of 94.52%. The final equation
(Model 4) has the following form (16) Variable Coefficients
Size –0.74**
yi, t = 0.28 − 0.74 ⋅ Sizei, t + 0.18 ⋅ Leveragei, t + (–2.30)
+ 0.78 ⋅ Li, t − 0.34 ⋅ Ocftdi, t + 5.23 ⋅ Clcai, t − Leverage 0.18**
(2.00)
− 64 ⋅ ROAi, t − 0.55 ⋅ Wctai, t − Ldummy 0.78
(16) (1.13)
− 0.14 ⋅ Nigrowthi, t + 0.29 ⋅ Nni, t −
Ocftl –0.34**
− 2.68 ⋅ Bmi, t + 4.11 ⋅ Voli, t − (–2.46)
− 5.54 ⋅ Rdctai, t + ε i, t . Clca 5.23*
(1.73)
The outcome is presented in Table 16. ROA –64***
After that, the probability of distress is (–5.51)
calculated in Stata 13.0 as a post estima- Wcta –0.55
tion. Next, in order to obtain the absolute (–0.29)
value of direct distress costs, this value Nigrowth –0.14
is multiplied by sales. Relative values are (–0.34)
also evaluated as the direct distress costs Nn 0.29
scaled by total assets and by value of a com- (0.43)
pany. The overall results are presented in Vol 4.11*
Table 17. (1.87)
Bm –2.68**
Evaluation of indirect costs (–2.15)
An ordinary least squares regression of a Rdcta –5.54**
company’s sales on the aggregate sales of (–2.06)
the industry for the ten prior years was Constant 0.28
conducted. The outcome of the model yields Observations 3326
a significant result for the independent vari- Wald chi2 81.90
able at a 1% level. Thus, the following equa- Prob  > chi2
0.0000
tion is provided:
N o t e: Z-statistics based on standard errors are
in parentheses;
Sales2015 = 4867.269 + 0.0005264 × *, ** and *** indicate statistical significance at
(17)
× Aggregate industry sales2005−2014 + ε. the 10%, 5% and 1% levels, respectively.

In order to find estimated sales for each Furthermore, the mean profit margin for
company, the amount of aggregate industry the ten years from 2006 to 2015 is calculat-
sales for 2005–2014 is substituted in every ed, and this figure is multiplied by estimated
of 389 cases. sales of the firms. Such manipulations give
Table 17
Direct distress costs

Variable Mean value

Direct distress costs $313.88 mln
Direct distress costs/Total assets 2.6%
Direct distress costs/Value of company 4.18%

RMJ 16 (1): 37–62 (2018)

58 E. Y. Makeeva, M. O. Khugaeva

Table 18
Indirect distress costs

Variable Mean value

Indirect distress costs $339.84 mln
Indirect distress costs/Total assets 5.35%
Indirect distress costs/Value of company 7.6%

Table 19
Total distress costs

Variable Mean value

Total distress costs $658.72 mln
Total distress costs/Total assets 7.95%
Total distress costs/Value of company 11.78%

one the opportunity to appraise evaluated Conclusion

profit.
The last step of evaluation of indirect We set the goal of evaluating financial dis-
distress costs is take the difference between tress costs of innovative companies at the
evaluated net income and real net income. beginning of this paper. This was achieved
Thus, the indirect financial costs’ mean value by completing the aforementioned tasks.
is equal to $339.84  million in 2015. Average To sum up, in Table 20 we compiled in-
relative values are also calculated, the in- formation about forecasts and outcomes.
direct distress costs scaled by total assets The conducted research gives one an op-
and the value of company amount 5.35% portunity to conclude that the higher the
and 7.6%, respectively (Table 18). debt, the higher are the interest expenses
It is interesting that 34% of the firms that negatively influence the interest coverage
have positive distress costs, i. e., unexpected ratio. As the dependent variable is a dummy
profits. The sample includes gigantic hold- variable y equal to 1 if interest coverage is
ings, which are developing much more rap- less than 1 and 0 if otherwise, then higher
idly than their parent industries, so their debt leads to y  = 1. The influence of current
financial indicators tend predominate those liabilities to current assets and cash flow from
of other holdings. operating activity to total debt has the same
Thus, both the direct and indirect costs logic as the influence of leverage, so the sign
of financial distress are evaluated, and to of this variable should be positive.
complete the investigation, the total distress Variables that include net income have an
costs are estimated as the sum of direct and impact on EBITDA, which is the numerator
indirect distress costs. The results are given of the interest coverage ratio. The dummy
in Table 19. for net income, if it was negative over the
The absolute mean value of total distress last two years is equal to 1 and 0 if otherwise.
costs is approximately $660 million. The It should have a positive impact as if it is
relative figures show that these costs are equal to 1, then interest coverage is lower
equal to 7.95% of total assets, or 11.78% than 0, and the dependent dummy is equal
of firms’ value. The ratios are quite high to 1. Therefore, the return on assets and
and should be taken into account when mak- changes in net income also has a positive
ing financial decisions such as the choice of effect on the interest coverage ratio, but a
target capital structure. negative impact on the dependent dummy.
RMJ 16 (1): 37–62 (2018)
Evaluation of Financial Distress Costs of Innovative Companies 59

Table 20
Results of hypotheses testing

Hypothesis (variable) Expected sign Estimated sign

H1 (Leverage) + +
H2 (Clca) + +
H3 (Nn) + Insignificant
H4 (Size) – –
H5 (Wcta) – Insignificant
H6 (ROA) – –
H7 (Ocftl) – –
H8 (Nigrowth) – Insignificant
H9 (Bm) + –
H10 (Vol) + +
H11 (Rda) + +
H11 (Rds) + +
H11 (Rde) + +
H11 (Rdcta) + –

The size of company is presented by a kinds of effects. R&D scaled by total as-
natural logarithm of total assets. Logically, sets, sales, and the number of employees
the larger the company is, the lower is the depend on the financial results of one year.
likelihood of bankruptcy. Working capi- The influence of these variables is positive.
tal scaled by total assets is expected to be However, research and development capital
negative, as the numerator of the ratio is scaled by total assets demonstrated an ad-
estimated as current assets minus current verse impact on the probability of distress,
liabilities. Thus, the higher the working while the forecast was positive. The logic
capital, the lower are the interest expenses, here is that R&D capital includes the sum
and the higher is the interest coverage ratio, of investments in innovative projects over
so the dummy is not equal to 1. the last four years. Over such a long period
According to [Zhang, 2015] such market time, these projects begin to make a profit,
characteristic of a firm as book to market so firm’s EBITDA rises and boosts the inter-
value and stock return volatility has a posi- est coverage ratio. It should be mentioned
tive impact upon the probability of distress. that the rest of the R&D indexes demon-
This can be explained the following way: if strate a positive relationship, but they do
debt rises, then book to market value be- not include prior periods’ expenses upon
comes lower. At the same time, interest ex- such projects.
penses go up and the interest coverage ratio A strong positive relationship was found
fall, and if it is less than 1, the dependent with the financial results of the previous two
dummy variable is equal to 1, which explains years. Intuitively, it is clear that if a firm
the negative relationship. was not in distress over the last two years,
R&D expenses have significant impact it will not be in a state of distress this year,
upon the probability of bankruptcy, and and vice versa. Besides, the results of post
consequently, upon financial distress. Un­ crisis periods demonstrate a positive impact
expectedly, R&D indexes have different upon the probability of financial distress.
RMJ 16 (1): 37–62 (2018)
60 E. Y. Makeeva, M. O. Khugaeva

Although most companies from the sample 1. It is necessary to determine the influ-
are from the US, they operate in many other ence of other sets of determinants on
countries, where crisis inflicted economic the probability of financial distress.
damage not in 2007, but at a later date. 2. Unexpected financial distress costs also
Some firms showed unexpected profits, should be checked for effects arising from
i. e., their indirect distress costs became other factors.
“benefits”. However, considering the fact 3. Another index that characterizes the in-
that the sample includes massive holdings novative firms should be developed.
that invest a great deal of money and as- 4. The present sample includes large holdings.
sets in innovations and act as leaders on We recommend that small and medium
the market, such a result was predictable. firms should be analyzed in the future.
Overall, the total financial distress costs 5. Furthermore, we suggest comparing fi-
in absolute and relative terms show that nancial distress costs and tax benefits
they cannot be ignored when making finan- in order to find some trade-off values
cial decisions, especially at those companies between these variables, and therefore
investing in research and development proj- determine the target value of companies.
ects. The present paper proposes two models: The current paper could be useful for com-
those for evaluation of direct and indirect panies that must make a decision concerning
costs in the innovative companies. In future investments in research and development
research, these models can be expanded with projects. It is also helpful for the govern-
new variables in order to obtain more ac- ment, especially the tax authorities, which
curate results. might regulate innovative companies through
The following recommendations are sug- tax breaks. Finally, researchers can use this
gested for future research: paper as a basis for further investigation.

Appendix
VIF data
Variable VIF 1/VIF Variable VIF 1/VIF
Wcta 4.90 0.203924 Size 1.94 0.516148
Clca 4.20 0.238017 ROA 1.91 0.523995
Rda 4.11 0.243115 Bm 1.69 0.591086
Rde 2.62 0.381736 Nn 1.43 0.699973
Rdcta 2.61 0.383143 Vol 1.27 0.789283
Rds 2.48 0.403303 Nigrowth 1.26 0.796759
Leverage 2.39 0.417625 Ocftl 1.04 0.961825
Tdta 2.37 0.422535 Mean VIF 2.41

References

Altman E. I. 1968. Financial ratios, discrimi- Altman E. I. 1984. A further empirical in-
nant analysis and the prediction of corpo- vestigation of the bankruptcy cost ques-
rate bankruptcy. Journal of Finance 23 (4): tion. Journal of Finance 39 (4): 1067–
589–609. 1089.

RMJ 16 (1): 37–62 (2018)

Evaluation of Financial Distress Costs of Innovative Companies 61

Andrade  G., Kaplan  S. N. 1998. How costly is Kwansa F. A., Cho  M. H. 1995. Bankruptcy cost
financial (not economic) distress? Evidence and capital structure: The significance of
from highly leveraged transactions that indirect cost. International Journal of Hos­
became distressed. Journal of Finance 53 pitality Management 14 (3/4): 339–350.
(5): 1443–1493. Laitinen E. K., Suvas A. 2016. Financial dis-
Bhabra G. S., Yao Y. 2011. Is bankruptcy cost- tress prediction in an international context:
ly? Recent evidence on the magnitude and Moderating effects of Hofstede’s original
determinants of indirect bankruptcy costs. cultural dimensions. Journal of Behavioral
Journal of Applied Finance & Banking 1 and Experimental Finance 9: 98–118.
(2): 39–68. LoPucki L. M., Doherty J. W. 2004. The de-
Bris A., Welch I., Zhu N. 2006. The costs of terminants of professional fees in large
bankruptcy: Chapter 7 liquidation versus bankruptcy reorganization cases. Journal
Chapter 11 reorganization. Journal of Fi­ of Empirical Legal Studies 1 (1): 111–141.
nance 61 (3): 1253–1303. Makeeva E., Bakurova A. 2012. Forecasting
Bulot N., Salamudin N., Abdoh W. M. 2014. bankruptcy oil and gas companies using
The size and determinants of indirect fi- neural networks. Journal of Corporate Fi­
nancial distress costs. GSTF Journal on nance Research (3): 22–30. (In Russian)
Business Review (GBR) 3 (4): 8–17. Mansi S., Maxwell W. F., Zhang A. 2012. Bank-
Bulot N., Salamudin N., Aziz R. 2017. The ruptcy prediction models and the cost of
size of indirect financial distress costs: debt. Journal of Fixed Income 21 (4): 25–42.
Which variable is reliably important? Jour­ Ohlson J. A. 1980. Financial ratios and the
nal Intelek 12 (1): 12–20. probabilistic prediction of bankruptcy. Jour­
Charalambakis E., Garrett I. 2016. On the nal of Accounting Research 18 (1): 109–131.
prediction of financial distress in developed Opler T. C., Titman S. 1994. Financial dis-
and emerging markets: Does the choice of tress and corporate performance. Journal
accounting and market information mat- of Finance 49 (3): 1015–1040.
ter? A comparison of UK and Indian Firms. Pham T., Chow D. 1989. Some estimated of
Review of Quantitative Finance and Ac­ direct and indirect bankruptcy costs in Aus-
counting 47 (1): 1–28. tralia: September 1978 – May 1983. Austra­
Chen J., Marshall B., Zhang J., Ganesh S. lian Journal of Management 14 (1): 75–95.
2006. Financial distress prediction in Chi- Thorburn K. S. 2000. Bankruptcy auctions:
na. Review of Pacific Basin Financial Mar­ Costs, debt recovery, and firm survival. Jour­
kets and Policies 9 (2): 317–336. nal of Financial Economics 58 (3): 337–368.
Fedorova  E. A., Gilenko  E. V., Dovzhenko  S. E. Warner J. B. 1977. Bankruptcy costs: Some
2013. Models for bankruptcy forecasting: evidence. Journal of Finance 32 (2): 337–
Case study of Russian enterprises. Studies 347.
on Russian Economic Development 24 (2): Weiss L. A. 1990. Bankruptcy resolution: Direct
159–164. costs and violation of priority of claims. Jour­
Ho C.-Y., McCarthy P., Yang Y., Ye X. 2013. nal of Financial Economics 27 (2): 285–314.
Bankruptcy in the pulp and paper indus- Wijantini W. 2007. The indirect costs of fi-
try: Market’s reaction and prediction. Em­ nancial distress in Indonesia. Gadjah Ma­
pirical Economics 45 (3): 1205–1232. da International Journal of Business 9 (2):
Hortaçsu A., Matvos G., Syverson C., Ven- 157–186.
kataraman  S. 2013. Indirect costs of finan- Zhang  W. 2015. R&D investment and distress
cial distress in durable goods industries: risk. Journal of Empirical Finance 32 (C):
The case of auto manufacturers. Review of 94–114.
Financial Studies 26 (5): 1248–1290. Zhdanov V. Y., Afanasyeva O. A. 2011. Mod-
Jones  S., Hensher  D. A. 2004. Predicting firm els of bankruptcy risk for companies of
financial distress: A mixed logit model. Ac­ aircraft industry. Journal of Corporate Fi­
counting Review 79 (4): 1011–1038. nance Research (4): 77–89. (In Russian)

RMJ 16 (1): 37–62 (2018)

62 E. Y. Makeeva, M. O. Khugaeva

The study was implemented in the framework of the Basic Research Program at the Na­tion­al
ResearchUniversity Higher School of Economics (HSE) in 2017 (“The empirical ana­lysis of cor-
porate financial decisions of innovative companies under conditions of the global transformation
to an innovative type of economy”).
For citation: Makeeva  E. Y., Khugaeva  M. O. 2018. Evaluation of financial distress costs of innova-
tive companies. Russian Management Journal 16 (1): 37–62. https://doi.org/10.21638/11701/
spbu18.2018.102

Initial Submission: November 25, 2017

Final Version Accepted: March 20, 2018

Оценка издержек финансовой неустойчивости в инновационных компаниях

Е. Ю. Макеева
Доцент, Национальный исследовательский университет «Высшая школа экономики»,
Москва, Россия
E-mail: [email protected]
М. О. Хугаева
Аспирант, Финансовый университет при Правительстве РФ, Москва, Россия
E-mail: [email protected]

В течение последних десятилетий инвестиции в НИОКР существенно возросли. Многие иссле-

дователи подтверждают положительную связь между расходами на НИОКР и стоимостью ком-
пании. Однако инвестиции в инновации характеризуются высокой степенью неопределенности.
Кроме того, инновационные проекты начинают приносить доходы, как правило, в долгосрочном
периоде. Эти факторы значительно увеличивают вероятность финансовой неустойчивости ком-
пании. Поэтому целью данной работы является оценка издержек финансовой неустойчивости в
фирмах, которые инвестируют в НИОКР. В работе проводится анализ финансово-экономических
показателей 389 инновационных компаний c 2006 по 2015 г. Анализ осно­ван на оценке прямых
и косвенных издержек финансовой неустойчивости фирм. Прямые издержки оцениваются с по-
мощью использования панельной логистической регрессии, а косвенные — как упущенная при-
быль или непредвиденный доход в сравнении с отраслевыми показателями. В результате анализа
подтверждается связь между расходами на НИОКР и финансовой неустойчивостью компаний.
Анализ показывает, что издержки финансовой неустойчивости инновационных фирм составляют
7,95% от величины общих активов, или в среднем 11,78% стоимости компании.
Ключевые слова: издержки финансовой неустойчивости, банкротство, инновационные ком-
пании, панельная логистическая регрессия.
JEL: G32, G33.

В данной работе использованы результаты проекта «Эмпирический анализ корпоративных

финансовых решений инновационных компаний в условиях глобальной трансформации экономики
в инновационный тип» Программы фундаментальных исследований НИУ ВШЭ в 2017 г.

https://doi.org/10.21638/11701/spbu18.2018.102

Статья поступила в редакцию

25 ноября 2017 г.
Принята к публикации
20 марта 2018 г.
RMJ 16 (1): 37–62 (2018)