See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/370602716

Developing a Scoring Credit Model Based on the Methodology of International

Credit Rating Agencies

Article in Journal of Corporate Finance Research / Корпоративные Финансы | ISSN 2073-0438 · April 2023

CITATIONS READS

0 72

3 authors, including:

Sergei Grishunin
National Research University Higher School of Economics
39 PUBLICATIONS 127 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Developing serveral online courses at Coursera Platform View project

ESG, non-financial and integrated reporting: theory and practice View project

All content following this page was uploaded by Sergei Grishunin on 08 May 2023.

The user has requested enhancement of the downloaded file.

Journal of Corporate Finance Research / New Research Vol. 17 | № 1 | 2023

DOI: https://doi.org/10.17323/j.jcfr.2073-0438.17.1.2023.5-16
JEL classification: G20, G24, G32

Developing a Scoring Credit Model

Based on the Methodology of
International Credit Rating Agencies
Alyona Astakhova
Supervision Department of Systemically Important Credit Institutions, Bank of Russia, Moscow, Russia,
[email protected], ORCID

Sergei Grishunin
Managing Director, National Rating Agency (NRA) LLC,
Researcher of Joint ESG Laboratory between NRA LLC and Moscow State University, Doctorate Research Associate,
Peter the Great St. Petersburg Polytechnic University, Moscow, Russia,
[email protected], ORCID

Gennady Pomortsev
Research Analyst, National Rating Agency (NRA) LLC, Moscow, Russia,
[email protected], ORCID

Abstract
The purpose of this work is to examine the relationship of various financial and non-financial (qualitative) factors of per-
formance of non-financial companies and their credit ratings.
We developed the scoring model which was based on the methodologies of international and Russian rating agencies.
The modelled ratings of non-financial companies for 2018–2020 were compared with actual ratings assigned by the rating
agencies and discrepancies were explained. The sample includes companies from retail, protein and agriculture, steel, oil
and gas sectors from Russia, USA, Luxembourg, England, Canada, India, Ukraine and Brazil.
The paper proved that addition of business and environmental, social and governance factors improved the quality of
scoring models in comparison to those including only financial metrics. There are strong patterns in the resulting ratings
of companies for some industries. Retail industry companies are associated with high sales indicators, while steel indus-
try companies have high interest expenses coverage ratios. Oil and gas industry companies mostly show high results in
reserves coefficients.
The study developed a credit rating forecasting tool that emulates the work of analysts of rating agencies and therefore has
a high predictive power. The developed model can be used by financial market practitioners to predict the credit ratings of
Russian companies in the face of the refusal of international rating agencies to rate Russian issuers.
Keywords: credit default prediction, credit rating modelling, credit rating system, ESG rating
For citation: Astakhova, A., Grishunin, S. , Pomortsev, G. Developing a Scoring Credit Model Based on the Methodology of
International Credit Rating Agencies. Journal of Corporate Finance Research. 2023;17(1): 5-16. https://doi.org/10.17323/j.
jcfr.2073-0438.17.1.2023.5-16

The journal is an open access journal which means that everybody can read, download, copy, distribute, print, search, or link to the full texts of these
articles in accordance with CC Licence type: Attribution 4.0 International (CC BY 4.0 http://creativecommons.org/licenses/by/4.0/).

5 Higher School of Economics

Journal of Corporate Finance Research / New Research Vol. 17 | № 1 | 2023

Introduction In particular, the developed scoring model considers other

important metrics in addition to financial data: govern-
The paper examines the relationship between various fi-
ment and group support factors, environmental influence
nancial and qualitative indicators and the credit ratings of
with social and management efficiency and sovereign rat-
non-financial companies based on publicly available infor-
ing adjustment. Other novelty factors include: independ-
mation. The study assessed the creditworthiness of non-fi-
ent calculation of qualitative indicators without expert
nancial companies from the following sectors: retail; steel;
guidance; identification of patterns in the values of finan-
agriculture; oil and gas. The research is dedicated to the
cial indicators and the resulting credit rating. Moreover,
development of a scoring model based on the methodol-
there is a lack of research that includes qualitative factors
ogy of international rating agencies for predicting credit
for non-financial companies, however, its importance is
risk and probability of default of international non-finan-
underpinned by several studies. For instance, the papers
cial companies. Along with the financial position of the
by G.M. Bodnar et al. [4], B. Lehmann [5] and J. Grunert
company, the scoring model allows to take into account
et al. [6] conclude that accuracy is increased with the in-
support factors such as group and government support,
corporation of several non-financial qualitative factors to
environmental influence, and to consider social factors
analysis; however, these results are only valid for certain
and management efficiency, as well as the company’s key
countries and only for financial companies, and thus could
success factors.
not be used for companies from other countries. ESG rat-
The assessment is based on the methodologies of interna- ings could be used as an additional indicator of financial
tional rating agencies that are integrated into the devel- performance, as F. Kiesel and F. Lücke [7] mention in their
oped model. paper, and hence it would also be reasonable to use the
ESG rating when modelling credit ratings. Therefore, the
Relevance of research creation of a model which incorporates qualitative factors
seems practical in future research in related fields.
First of all, the financial crisis of 2007–2008 demonstrated
how important objective high-quality ratings are for the The practical relevance of the research is high. The present
stability of the global economy. Erroneous ratings have led study developed a model ready for use and implementa-
to the bankruptcy of a large number of firms [1]. A similar tion, presenting an interface and data output that is under-
situation can emerge if the approach to companies’ credit standable to all users. Such a tool is especially relevant, first
ratings is not thorough enough. Secondly, an independent and foremost, for assessing the creditworthiness of compa-
credit rating methodology that unifies the existing models nies whose ratings have not been published by rating agen-
and adjustments is required to assess the creditworthiness cies. In this case, its application is the quickest and most
of companies internationally, given the recent trends in the plausible way to obtain a rating. Secondly, the open-source
sphere of credit ratings and adjustments. Thirdly, large loss- code allows the model to become a universal foundation
es could be shifted to private investors, who may be driven for further improvement, implementation of third-party
by the incorrect ratings of the firms in which they invested. tools and connection to various resources.
For instance, the Yutrade broker went bankrupt in 2008,
and its clients lost all their investments [2]. Fourthly, cor- Literature Review
porate governance, and the environmental and social im-
The theoretical base of the paper comprises the studies of
pact of a company on its creditworthiness score are now
foreign and Russian researchers in the field of corporate
gaining popularity [3].
finance and risk management. The works of the following
The objects of the research are the international non-finan- Russian and foreign researchers were used: T.M. Zador-
cial companies from the retail, steel, protein and agricul- ozhnaya, A.M. Karminsky, A.A. Polozov, B.H. Bergrem
ture, and oil and gas industries for the period between 2018 and others.
and 2021. Therefore, the subject of research is the relation-
The literature review demonstrates that there is a limited
ship between various financial and qualitative indicators
number of studies on credit ratings and the development
and the credit ratings of non-financial companies.
of models for assessing the credit risks of non-financial
The goals of the research are the selection and study of companies. In most cases, the significance of independent
the scientific literature on the topic; choosing the most credit ratings and their impact on the financial system is
relevant methods and methodologies for building scoring provided. For instance, the paper by T.M. Zadorozhnaya
models; collecting sample data for non-financial compa- [8] presents the basic definitions and objectives related to
nies for 2018–2021; creating an Excel VBA-based interface credit ratings and, most importantly, the tasks that the ex-
to calculate financial and qualitative indicators; conducting istence of ratings solves, i.e., information disclosure, setting
a detailed analysis of model prediction accuracy; adding limits on credit risk, forming an objective assessment of
adjustments to improve prediction accuracy; building the the borrower by the lender, promoting the diversification
scoring model suitable for rating distribution. of funding sources, promoting the reduction of the cost of
The scientific novelty of the research is underpinned by capital and directly regulating financial markets. Moreover,
the limited number of studies on the topic of independ- credit ratings are important in the financial performance
ent credit rating modelling for non-financial companies. assessment, as revealed in M. Singal’s paper [9]. The author

6 Higher School of Economics

Journal of Corporate Finance Research / New Research Vol. 17 | № 1 | 2023

concludes that the changes in credit ratings are reflected in tors affect the probability of default, which is lowered, for
stock prices and the corresponding investors’ reaction, and instance, when the CEO of a company is also its co-owner
thus affect a company’s financial performance. and increases when a company becomes a subsidiary. The
An important part of the paper is related to qualitative second finding is significant for current research when the
factors. Several research studies agree that the incorpora- results are compared with an adjustment for being a part
tion of qualitative and non-financial variables in the model of a group, which traditionally has a positive effect on the
could improve the accuracy of credit rating prediction. The credit rating. Therefore, using different variables to assess
papers by B. Lehmann [5] and J. Grunert et al. [6] investi- credit ratings for companies in different industries is theo-
gate the impact of qualitative factors on the credit rating retically reasonable.
assessment, therefore, this study accommodates for the Since not only the company itself affects its credit quality, a
non-financial qualitative factors to improve model accu- deep investigation into adjustments to its stand-alone cred-
racy. itworthiness assessment is required. Karminsky’s paper
Another important point is the distinction between devel- [10] highlights the applicability of ratings and their distri-
oped and emerging countries. The paper by A.M. Karmin- bution in today’s financial world and shows the importance
sky entitled “Corporate rating models for emerging mar- of using external support factors on a par with internal
kets” [10] presents several financial, macroeconomic, and factors, both quantitative and qualitative, in evaluating a
qualitative indicators and their effect on the credit rating of company’s financial stability in one way or another. But
a company using econometric models that use these coef- it is important to mention the adjustment for the overall
ficients in different proportions. This study also examines sovereign rating. The paper by A.M. Karminsky and A.A
the important question of how results differ for companies Polozov – “Handbook of ratings” [14] notes that a compa-
from emerging markets and what the key differences and ny’s credit rating rarely exceeds the the sovereign rating.
specifics are in assessing their credit ratings. These findings A company’s stand-alone rating is measured in a “bubble”,
have a key value in current research, and help to interpret but there are macroeconomic risks that the company does
results and make the correct conclusions for companies not control: political stability, competitive environment,
from emerging markets. Thus, companies from emerging strength of invention protection. However, there are exam-
countries are more exposed to macroeconomic factors, ples of companies that refute this rule. Such companies are
which are considered qualitative variables, or an adjust- assigned a rating higher than the sovereign rating because,
ment for the sovereign rating. due to certain circumstances, it is possible to exclude nega-
tive factors affecting credit quality from consideration (un-
In addition, most studies involve the use of different ex-
like in the calculation of the sovereign rating), or simply
ternal factors and specific indicators for each non-finan-
because other strongly positive features are present.
cial industry to assess credit risks. A.M. Karminsky’s paper
“Credit ratings and their modelling” [11] completely cov- Reconciling the results obtained by using methodolo-
ers the issues of credit quality assessment and their emer- gies with different scales is important. The paper by N.F.
gence. The study discusses the classification of ratings and Dyachkova “Comparison of rating scales of Russian and
conducts an analysis of existing methodologies and princi- foreign agencies: industrial and financial companies” [15]
ples of credit rating formation used by the most recognised reveals the importance of correct conversion of Russian
rating agencies. Moreover, B. H. Bergrem’s paper “An em- rating scales to international ones. The study examines the
pirical study of the relationship between credit ratings and relationships between rating scales used by different rating
financial ratios in the E&P industry” [12] examines the key agencies, and it is mostly valuable for current purposes,
indicators that are unusual for other methodologies, and since several companies have not been assigned a rating by
are important in the E&P (Exploration and Production) Moody’s. This paper presents a method of forming numer-
sector of the oil and gas industry. The cost of discovery ic rating scores. These scores are used in empirical mod-
and development is one of the vital keys to understanding els to study relationships between ratings and explanatory
the operating efficiency of a company, and one of the fun- factors.
damental indicators in assessing the scale of a company’s Highlighting the patterns for specific industries could be
unproven reserves. In this case, the stable replenishment complicated due to various difficulties and a dissimilarity
of reserves, their volume and geographic diversification, of the companies. However, there are research studies that
unlike company revenue, can serve as the best indicator draw almost the same conclusions about the most impor-
of long-term stability. Finally, A.I. Rybalka [13] demon- tant factors for a specific industry. The scale-related factors
strates how different specific indicators of non-financial generate many advantages for a retail company over its
companies could affect the probability of default using competitors, such as market power and price leadership.
logit regressions. The author determines the importance These advantages can lead to greater investor attractive-
of including qualitative indicators and their effect on the ness compared to smaller companies. Such a strong effect
results. The results reveal the difference when several cor- of the scale is confirmed by several studies. For instance,
porate governance coefficients are included, and are also A.B. Curtis et al. [16] argue that the revenue variable is
valuable for current research since the paper investigates the main component in the retail companies’ financial
companies’ ESG ratings and specifically, their governance performance forecast. As for the steel industry, profitabil-
components. It has been established that governance fac- ity-related variables, particularly that of financial perfor-

7 Higher School of Economics

Journal of Corporate Finance Research / New Research Vol. 17 | № 1 | 2023

mance, are considered the most significant in assigning a

credit rating, as confirmed by A. Banerjee [17]. The oil and
Methodology
The methodological base of the paper was formed by the
gas industry functions in the long-term perspective, i.e.,
work of international rating agencies. The following meth-
companies need to consider their reserves, which leads to
odologies were used: Moody’s retail industry methodology
higher values of these variables for such companies. More-
[18]; Moody’s steel industry methodology [19]; Moody’s
over, B.H. Bergrem [12] also underpins the relevance of
protein and agriculture industry methodology [20];
the scale variables for the oil and gas companies, however,
Moody’s oil and gas (E&P) industry methodology [21];
the monitoring of average daily oil and gas production as
ACRA government support methodology [22]; ACRA
better representation of industry-specific factors that in-
group belonging methodology [23]; Expert RA ESG rating
fluence financial performance is also considered impor-
methodology [24].
tant.
Moody’s published methodologies used in the construc-
The literature review demonstrates that research tasks are
tion of the model do not completely reflect the procedure
of primary importance for researchers and practitioners.
of companies’ rating formation by Moody’s. They only re-
The previous studies indicate that the results of credit risk
flect the principles of assessment of the most common im-
assessment analysis differ when qualitative indicators from
portant indicators in a specific industry, which allow the
external databases vs. other factors are assessed.
authors of the model to use other relevant tools when com-
piling the rating calculator. The presented methodologies
Data comprise a method of indicator evaluation on an 8–9 point
All financial data was obtained from official company scale, converting them from this scale to a quantitative
reports, and qualitative factors are measured based on scale according to Moody’s rating evaluation formula and
publicly available information. The financial data was converting them into a final credit rating as demonstrated
retrieved from Bloomberg and Thomson Reuters termi- in Figure 1. A certain advantage of the selected methodol-
nals. The financial variables used are revenue, EBIT, in- ogies over those of international competitors – Fitch and
terest expenses, retained cash flow, total debt, EBITDA, S&P – is the more expansive grading in the calculation of
net debt, return on tangible assets, book capitalization, qualitative indicators, with more “binomial” parameters
cash flow from operations, dividends. Specific variables (value 0 or 1) in evaluation. The data for evaluation can
for the oil and gas industry are also used: proven and de- only be found implicitly, by studying the companies’ of-
veloped reserves and average daily production. Therefore, ficial presentations to investors or similar documents, in
the database is a sample of the five largest companies in which they disclose information relevant for the study us-
each industry with previously published Moody’s ratings ing the model.
and publicly available reports and forecasts. Addition-
Figure 1. The formula for the overall stand-alone credit
al qualitative data on 5 companies is obtained to test the
rating
ESG rating model. In most cases, since the majority of the
companies in question publish IFRS statements, all their n

calculations are conducted in US dollars. But for repre- ∑ x

i =1 i
⋅ weighti =
Xs .
sentativeness we added two companies that use national Note: xi is a grade of a subfactor i and X s is an overall
currency for calculations, X5 Retail Group (Russian ru- stand-alone numerical credit rating.
bles) and Husky Inc. (Canadian dollars). The companies Source: Moody’s and authors’ calculations.
from eight countries are examined: Russia, USA, Luxem-
bourg, England, Canada, India, Ukraine and Brazil. The The assessment is based on key indicators such as: scale,
average value 2018–2020 is calculated for each factor in company’s business profile, profitability and efficiency,
the model. This is due to the fact that ratings are assigned leverage and coverage, and the financial policies pursued
through a cycle. With this approach, seasonal fluctuations by the company. Generally, these indicators also contain
in business activity are averaged. Since our model takes sub-factors, which, when combined, will better reflect the
into account sovereign risks, the sample includes sover- value of the overall indicator itself. Each subfactor value is
eign ratings for the studied countries with forecasts. This measured as a weighted year average: 2018 – 15%, 2019 –
data was obtained from the Bloomberg system. However, 25%, 2020 – 30% and 2021 – 30%.
a selection bias problem could be present due to the small Therefore, each factor and sub-factor is assessed and then
number of observations in the dataset. Hence, the findings transposed to the numerical value according to Table 1 to
corresponding to the industry patterns are only relevant proceed to a calculation of the final rating using the weights
for similar situations. specified in Tables A1–A4 depending on the industry.
Table 1. Rating scale
Credit rating Aaa Aa A Baa Ba B Caa Ca C
Grade 1 3 6 9 12 15 18 20 21
Note: The rating grade is calculated for each sub-factor in the model.
Source: Moody’s.

8 Higher School of Economics

Journal of Corporate Finance Research / New Research Vol. 17 | № 1 | 2023

The created model will take the external influences into

consideration; the result will not be a stand-alone rating.
Results
The model will allow to examine the influence of state sup- Baseline credit assessment results
port, group support, as well as to calculate the ESG rating. The results of this model can be divided into two categories:
The selected methodologies are used to study, evaluate and company ratings compared to Moody’s ratings and gener-
take into account the influence of parent structures and the al patterns and trends identified based on the results of the
state on companies in the Russian Federation, but, since model. To begin with, it should be noted that the model
the model is designed to calculate the rating for companies does not allow a company to possess a rating higher than the
around the world, these methodologies were taken only as corresponding sovereign rating of its country. These results,
the foundation and, as a result, the points relevant to the presented by Moody’s, were designated as outliers, and these
specifics of the Russian Federation were adapted to other companies’ ratings were equated to the sovereign ratings.
countries. The estimates of the influence of state support The research results demonstrate that the resulting model is
and the of being a part of a group will be provided as cor- highly accurate, as the average deviation from the Moody’s
rections to the original stand-alone rating according to the rating without adjustments is –0.75 points. With applied
specified methodologies, while the ESG rating was created adjustments, model accuracy becomes –0.25 points, with
as an independent rating, which could be included with the an average ESG rating deviation of 0.5 points. This high
overall results of any company. accuracy value indicates that all the required coefficients
To account for external support from the state or share- were considered because, when the amount of data under
holders, the joint default analysis approach was used. This consideration increases, the amount of discrepancies de-
approach includes assessment of two dimensions of sup- creases, which also indicates that they should be consid-
port: (1) the strength of the links between the company ered when assessing credit quality.
and its shareholders; and (2) the probability of sharehold- The difference in the results may mainly indicate the exist-
ers’ support of the company (Table A5). The probability ence of discrepancies in the data between potential users
of support is assessed using the creditworthiness of the and rating agencies due to different years studied or dif-
shareholder (SICA) with the following factors: presence of ferent exchange rates of national currencies. The model
a legal relationship, presence of contingent liabilities (in- mostly underestimates the ratings, which is caused by the
cluding sureties and guarantees), strategic importance and presence of crisis years, when the main financial indicators
operational integration. Subsequently, the final adjustment are traditionally lower, in the sample.
value to stand-alone creditworthiness assessment (SCA) is
Certain patterns emerged in the database, and the “Sales”
calculated according to Table A5. The adjustment for state
coefficient for the retail industry is the most common
support requires an assessment of systemic importance
successful result out of financial coefficients in terms of
and state influence levels as qualitative factors, and subse-
value added to the financial rating, not adjusted for coef-
quent calculation of the adjustment value for the support
ficient weight for all the companies in the sample (Table
from the shareholders (Table A6).
2). It could confirm that the key characteristic of the retail
Moreover, the final credit rating is then compared with the industry is that its sales generate the main profit, because
stand-alone rating to avoid outliers and adjust for the specif- it directly dictates the company’s position in the market.
ic country’s macroeconomic risks. Therefore, the final quan- On the other hand, 4 out of 5 companies demonstrate the
titative credit rating is calculated as presented in Figure 2. least successful results in financial coefficients in terms of
Figure 2. The formula for the final credit rating value added to financial rating, not adjusted for weight of
debt-related coefficients (Table 2). It is important to note
min(sovereign rating, X s + GOV + GROUP) =
X f.. that the results obtained for X5 Retail Group, which are
calculated in American dollars and rubles, do not differ
Note: GOV and GROUP represent adjustment by state and
from each other, which may indicate the correct account-
group support, respectively.
ing for the currency in which the reports are presented.
Source: Authors’ calculations.
Table 2. Model ratings with adjustments of companies from the retail industry

Company Agency rating With adjustments Best coefficient Worst coefficient

X5 Retail Group (USD) Ba1 Baa3 Sales EBIT / Interest Expense
X5 Retail Group (RUB) Ba1 Baa3 Sales EBIT / Interest Expense
Costco Aa3 Aa3 Sales RCF / Net Debt
Walmart Aa2 A2 Sales RCF / Net Debt
Starbucks Baa1 Baa1 Sales RCF / Net Debt
Party City Holdco Inc. Caa1 B3 Sales Debt / EBITDA
Note: Best and worst coefficients are the most and least successful results among financial coefficients in terms of value
added to financial rating, not adjusted for coefficient weight.
Source: Authors’ calculations.

9 Higher School of Economics

Journal of Corporate Finance Research / New Research Vol. 17 | № 1 | 2023

In addition, there is a clear pattern in the steel indus- funds from operating activities. This indicator is impor-
try, with the “EBIT/Interest Expense” coefficient being tant as it directly reflects the amount of cash that the
the most successful for 4 out of 5 companies (Table 3). company generates from its income. Also, 2 companies
It can indicate a company’s positive net profits and low have an equally unsuccessful sales indicator (Table 3),
interest expenses on short-term and long-term debts. which may be caused by the lack of demand for goods
While the worst indicators differ, 2 companies have in the years under consideration or hint to at a weak
almost the same deficiency in the CFO/Debt indicator position in the markets where the company carries out
(Table 3), which may indicate a small amount of free its activities.

Table 3. Model ratings with adjustments for companies in the steel industry
Company Agency rating With adjustments Best coefficient Worst coefficient
MMK Baa2 Baa3 EBIT / Interest Expense Sales

NLMK Baa2 Baa3 EBIT / Interest Expense CFO / Debt

Severstal Baa2 Baa3 EBIT / Interest Expense Sales

EVRAZ Ba1 Baa3 EBIT / Interest Expense CFO / Debt

ArcelorMittal Ba1 Baa3 Sales Debt / BookCap

Note: Best and worst coefficients are the most and least successful results in financial coefficients in terms of value added
to financial rating, not adjusted for coefficient weight.
Source: Authors’ calculations.

The oil and gas industry, especially the exploration and cash flow. This would be a negative signal for investors, as
production sector, is directly related to the reserves and this indicator is used to determine the company’s ability to
volumes of daily production. Therefore, the most success- repay its debts from cash generated from operations, i.e.,
ful results in this industry are revealed by the Debt / PD sales, after dividend payments. Notably, the only compa-
reserves indicator, and this is the case for each company ny with a different least successful indicator is Russneft,
(Table 4). Hence, it is possible to state that in this industry whose possible bankruptcy has been discussed in the
reserve indicators are important for companies and even news. It has the least successful results in the Average Daily
despite the crisis years, the management board monitors Production coefficient that indicates poor sales estimates,
and maintains this indicator at the proper level. The least which would negatively affect all financial results and, im-
successful indicator results for 4 out of 5 companies are portantly, the company’s lack of willingness to compete in
reflected in RCF/Debt coefficient (Table 4), which may the market and shows little impact on the development of
be due to the companies’ high debt ratio or low retained the industry.

Table 4. Model ratings with adjustments of companies from the oil and gas (E&P sector) industry
Company Agency rating With adjustments Best coefficient Worst coefficient
Oil India Baa3 Baa3 Debt / Reserves RCF / Debt

Husky A2 A3 Debt / Reserves RCF / Debt

Russneft Caa2 B2 Debt / Reserves Avg Daily Prod

EOG resources A3 Baa3 Debt / Reserves RCF / Debt

Murphy Oil Corp Ba3 Ba2 Debt / Reserves RCF / Debt

Note: Best and worst coefficients are the most and least successful results in financial coefficients in terms of value added
to financial rating, not adjusted for coefficient weight.
Source: Authors’ calculations.
It is difficult to identify clear patterns for the most or least markets. Only two companies have the same most success-
successful indicators in the protein and agriculture indus- ful metric, which is CFO/Debt (Table 5), and while it is
try, which may be due to the sample of companies in the equally positive, it does not have a very high rating. Be-
database: they are not similar to each other and may rank cause of high competition and low market power, the debt
differently in sales and systemic importance within their load is the key distinguishing factor between solvent and

10 Higher School of Economics

Journal of Corporate Finance Research / New Research Vol. 17 | № 1 | 2023

insolvent companies in this industry. The least successful suggesting that the metric that measures a company’s total
performers differ even more, although the two companies outstanding debt as a percentage of total company capitali-
have similarly lagged Debt/Book Capitalization (Table 5), zation is lagging and requires work in the future.
Table 5. Model ratings with adjustments of companies from the protein and agriculture industry

Company Agency rating With adjustments Best coefficient Worst coefficient

Cherkizovo Group B1 Ba3 Debt / EBITDA EBIT / Interest
Expense

Archer-Daniels-Midland A2 Baa2 Sales CFO / Debt

Company

MHP SE B2 B1 CFO / Debt Debt / BookCap

Minerva S.A. Ba3 B2 CFO / Debt Debt / BookCap

Ingredion Inc Baa1 Baa1 EBIT / Interest Sales

Expense
Note: Best and worst coefficients are the most and least successful results in financial coefficients in terms of value added
to financial rating, not adjusted for coefficient weight.
Source: Authors’ calculations.

The model’s ability to predict and indicate weaknesses in The ratings obtained by the model and the rating agencies
companies, which can be adjusted by substituting different for PIK, AK BARS, GLAVSTROY, GTLK and TRINFICO
values, as well as to point out the line of effort, together apparently coincide (Table 6), since the Expert RA meth-
with the resulting patterns in the relationship between the odology was taken as the foundation when forming the
credit rating and the financial indicator values can help to ESG rating in the model, which is very similar to the NRA
identify a company’s strengths and predict its level of credit methodology for the majority of coefficients. However, in
risk. the case of X5, the obtained result is different: the rating
calculated using the model is higher than the agency rating,
ESG Rating results which may be due to a different approach to evaluation and
different views on environmental, social and corporate gov-
The ESG rating is built into the model as an independent
ernance issues. The MSCI methodology is a guide to rating
tool for calculating the rating of possible environmental
indicators on a broader scale. Each indicator is assessed on
and social damages, as well as corporate governance risks
a scale of 0 to 10, adding more detail to the actions, while
in the company. When calculating the main credit rating of
increasing the subjectivity of the assessment, as the user is
a company, the potential user of the model can introduce
given an opportunity in advance to assess the company’s
corrections and proceed to the calculation of the ESG rat-
actions on a positive-negative spectrum, even though all
ing with the average variance between actual and modelled
the necessary data is publicly available, and it is easy to find
rating of about 0.5 notches.
relevant answers to all questions in each of the three areas.

Table 6. Model ESG ratings results

Company Agency rating Model rating

PIK GROUP ESG-2 (Expert RA) ESG-2

PJSC AK BARS BANK ESG-3 (Expert RA) ESG-3

GLAVSTROY ESG-4 (Expert RA) ESG-4

GTLK ESG-3 (Expert RA) ESG-3

TRINFICO B1 (NRA) ESG-3 (B1 NRA’s scale)

X5 Retail Group BB (MSCI) ESG-3 (A-BBB MSCI’s scale)

Source: Authors’ calculations.

11 Higher School of Economics

Journal of Corporate Finance Research / New Research Vol. 17 | № 1 | 2023

Conclusion, Contribution 2. Glushenkova M. Intermediaries of ruin. Kommersant

Den’gi. Jun. 01, 2009. URL: https://www.kommersant.
and Implication ru/doc/1179079 (accessed on 12.11.2021). (In Russ.).
An analysis of deviations from Moody’s estimates was car- 3. Factors behind the growing popularity of ESG invest-
ried out; the obtained difference in the results can mainly ing. Nasdaq. Apr. 24, 2021. URL: https://www.nasdaq.
indicate the presence of discrepancies in the data between com/articles/factors-behind-the-growing-populari-
potential users and rating agencies due to different years ty-of-esg-investing-2021-04-24
being analyzed or different exchange rates of national cur-
rencies. The model mostly underrates the results due to the 4. Bodnar G.M., Consolandi C., Gabbi G., Jaiswal‐Dale
presence of crisis years, when key financial indicators are A. Risk management for Italian non‐financial firms:
traditionally lower, in the sample. Currency and interest rate exposure. European Fi-
nancial Management. 2013;19(5):887-910. https://doi.
Differences in the results may also positively suggest an
org/10.1111/j.1468-036X.2012.00659.x
unbiased approach to assessing credit quality, that is, one
without strong subjectivity. The approach to the assessment 5. Lehmann B. Is it worth the while? The relevance
of qualitative indicators is different; accordingly, long-term of qualitative information in credit rating. SSRN
ratings obtained by the rating agencies are not necessarily Electronic Journal. 2003. https://doi.org/10.2139/
the only correct ones. The entire process of their formation ssrn.410186
is fully described in this paper, accordingly, the unbiased
6. Grunert J., Norden L., Weber M. The role of non-fi-
nature of the results obtained is an undeniable advantage.
nancial factors in internal credit ratings. Journal of
From the analysis of the indicators, the most and least suc- Banking & Finance. 2005:29(2):509-531. https://doi.
cessful performances in each of the industries are identi- org/10.1016/j.jbankfin.2004.05.017
fied and the reasons behind these patterns are demonstrat-
ed. The model’s ability to predict and indicate weaknesses 7. Kiesel F., Lücke F. ESG in credit ratings and the im-
in companies’ performance, which can be adjusted by sub- pact on financial markets. Financial Markets, Institu-
stituting different values, as well as to point out line of ef- tions & Instruments. 2019;28(3):263-290. https://doi.
fort, together with the resulting patterns in the relationship org/10.1111/fmii.12114
between the credit rating and the values of financial indi- 8. Zadorozhnaya T.M. The credit ratings of emitters de-
cators can help to identify the strengths of a company and fined by rating agencies. Rossiiskoe predprinimatel’stvo
predict its level of credit risk. = Russian Journal of Entrepreneurship. 2012;(14):85-
Other positive aspects of the resulting model are its versa- 89. (In Russ.).
tility, both in application and in its high adaptability to var-
9. Singal M. Firm credit rating as a measure of organiza-
ious new tasks. It can be modified in an uncomplicated way
tional and financial performance. Journal of Busi-
to study the credit quality of companies from other indus-
ness & Financial Affairs. 2013;2(3):1-2. https://doi.
tries, a region’s credit rating or the formation of sovereign
org/10.4172/2167-0234.1000e135
ratings, depending on the interests of a potential customer.
The research has also provided detailed analyses of the in- 10. Karminsky А.М. Corporate rating models for emerg-
formation power (importance) of financial and nonfinan- ing markets. Korporativnye finansy = Journal of Cor-
cial factors within each credit rating scoring model. porate Finance Research. 2011;5(3):19-29. (In Russ.).
https://doi.org/10.17323/j.jcfr.2073-0438.5.3.2011.19-
The obtained tool can be updated, supplemented, and im-
29
proved, and the example of the ESG rating shows how easy
it is to build a variety of new solutions to tasks that will 11. Karminsky A.M. Credit ratings and their modeling.
affect the final level of credit risk of the company. As this Moscow: LitRes; 2020. 305 p. (In Russ.).
scoring model is a universal tool with a user-friendly inter-
12. Bergrem B.H. The determinants of creditworthi-
face and a ready database that can be updated for further
ness: An empirical study of the relationship between
development and expansion of various specific tasks, it can
credit ratings and financial ratios in the E&P indus-
account for all possible necessary factors for the solution of
try. Master thesis in financial economics. Bergen:
risk assessment-related tasks. Therefore, such a model has
Norwegian School of Economics; 2014. 94 p. URL:
great potential for development and practical application.
https://openaccess.nhh.no/nhh-xmlui/bitstream/han-
dle/11250/218454/Masterthesis.pdf?sequence=1&is-
References Allowed=y

1. Zaidi D. The indisputable role of credit ratings 13. Rybalka А.I. Modeling the probability of default
agencies in the 2008 collapse, and why nothing has in the construction sector: Factors of corporate
changed. Truthout. Mar. 19, 2016. URL: https://truth- governance. Korporativnye finansy = Journal of
out.org/articles/the-indisputable-role-of-credit-rat- Corporate Finance Research. 2017;11(3):79-99.
ings-agencies-in-the-2008-collapse-and-why-noth- (In Russ.). https://doi.org/10.17323/j.jcfr.2073-
ing-has-changed/ (accessed on 10.07.2021). 0438.11.3.2017.79-99

12 Higher School of Economics

Journal of Corporate Finance Research / New Research Vol. 17 | № 1 | 2023

14. Karminsky A.M., Polozov A.A. Handbook of ratings:

Approaches to ratings in the economy, sports, and
society. Cham: Springer-Verlag; 2016. 356 p. https://
doi.org/10.1007/978-3-319-39261-5
15. Dyachkova N.F. The comparison of rating scales of
Russian and foreign agencies: An empirical study
for industrial and financial companies. Korpora-
tivnye finansy = Journal of Corporate Finance Re-
search. 2018;12(2):153-170. (In Russ.). https://doi.
org/10.17323/j.jcfr.2073-0438.12.2.2018.153-170
16. Curtis A.B., Lundholm R.J., McVay S.E. Forecast-
ing sales: A model and some evidence from the
retail industry. Contemporary Accounting Research.
2014;31(2):581-608. https://doi.org/10.1111/1911-
3846.12040
17. Banerjee A., De A. Determinants of corporate
financial performance relating to capital struc-
ture decisions in Indian iron and steel industry:
An empirical study. Paradigm: A Management
Research Journal. 2014:18(1):35-50. https://doi.
org/10.1177/0971890714540365
18. Retail industry methodology. Moody’s. 2018. URL:
https://www.moodys.com/researchdocumentcon-
tentpage.aspx?docid=PBC_1120379 (accessed on
10.07.2021).
19. Steel industry methodology. Moody’s. 2017. URL:
https://www.moodys.com/researchdocumentcon-
tentpage.aspx?docid=PBC_1074524 (accessed on
10.07.2021).
20. Protein and agriculture industry methodology.
Moody’s. 2019. URL: https://www.moodys.com/
researchdocumentcontentpage.aspx?docid=P-
BC_1113389 (accessed on 10.07.2021).
21. Oil and gas industry methodology. Moody’s. 2017.
URL: https://www.moodys.com/researchdocument-
contentpage.aspx?docid=PBC_1056808 (accessed on
10.07.2021).
22. Government support methodology. ACRA. 2020.
URL: https://www.acra-ratings.ru/storage/content/
attachments/7756/20201002_ACRA_GRI_ru.pdf
(accessed on 10.07.2021). (In Russ.).
23. Group belonging methodology. ACRA. 2019. URL:
https://www.acra-ratings.ru/storage/content/attach-
ments/7604/20200724_Group_Support_Methodolo-
gy_ru.pdf (accessed on 10.07.2021). (In Russ.).
24. ESG rating methodology. Expert RA. 2021.
URL: https://raexpert.ru/docbank/5f1/4a8/
b0f/3b42bf669a9bbfd305249a6.pdf (accessed on
10.07.2021). (In Russ.).

13 Higher School of Economics

Journal of Corporate Finance Research / New Research Vol. 17 | № 1 | 2023

Appendix 1
Table A1. Retail industry

Rating factors Weight, % xi Weighti , %

Scale 10.00 Revenue 10.00

30/00 Product stability 10.00

Business profile
Execution and Competitive Position 20.00

45.00 EBIT / Interest Expense 15.00

Leverage and coverage Retained Cash Flow / Net Debt 15.00

Debt / EBITDA 15,00

Financial Policy 15.00 Financial Policy 15.00%

Total 100.00 Total 100.00

Source: Moody’s and author’s calculations.

Appendix 2
Table A2. Oil and Gas industry

Rating factors Weight, % xi Weighti , %

Scale 20.00 Average Daily Production(Mboe/d) 10.00

Proved Developed Reserves(MMboe) 10.00

Business profile 10.00 Business profile 10.00

Profitability and efficiency 25.00 Leveraged Full-Cycle Ratio (EBIT Margin) 25.00

EBITDA / Interest Expense 7.50

Debt / Average Daily Production 7.50

Leverage and coverage 30.00
Debt / PD Reserves boe 7.50

RCF / Debt 7.50

Financial Policy 15.00 Financial Policy 15.00

Total 100.00 Total 100.00

Source: Moody’s and author’s calculations.

14 Higher School of Economics

Journal of Corporate Finance Research / New Research Vol. 17 | № 1 | 2023

Appendix 3
Table A3. Steel industry

Rating factors Weight, % xi Weighti , %

Scale 20.00 Revenue 20.00

Business profile 20.00 Business profile 20.00

Profitability and efficiency 15.00 EBIT Margin 10.00

Return on Tangible Assets 5.00

Leverage and coverage 35.00 EBIT / Interest Expense 7.50

Debt / Book Capitalization 5.00

Debt / EBITDA 15.00

(CFO-Dividends) / Debt 7.50

Financial Policy 10.00 Financial Policy 10.00

Total 100.00 Total 100.00

Source: Moody’s and author’s calculations.

Appendix 4
Table A4. Protein and agriculture industry

Rating factors Weight, % xi Weighti , %

Scale 10.00 Revenue 10.00

Business profile 35.00 Geographic diversification 5.00

Segment Diversification 5.00

Market share 5.00

Product Portfolio Profile 10.00

Income stability 10/00

Leverage and coverage 40.00 Debt / EBITDA 10.00

CFO / Debt 10.00

Debt / Book Capitalization 10.00

EBIT / Interest Expense 10.00

Financial policy 15.00 Financial policy 15.00

Total 100.00 Total 100.00

Source: Moody’s and author’s calculations.

15 Higher School of Economics

 Journal of Corporate Finance Research / New Research Vol. 17 | № 1 | 2023

Appendix 5
Table A5. Adjustment for support from the state or other shareholders

Degree of relationship
Very strong Strong Moderate Weak Very weak
Not higher than Not higher than
Supporting institution category

Strong Not higher SCA + 4, SCA + 3, Not higher than

SCA
than SICA* but not higher but not higher SCA + 2
than SICA* – 1 than SICA* – 2
Moderately Not higher Not higher than Not higher than
SCA SCA
strong than SICA* SCA + 2 SCA + 1

Neutral SCA SCA SCA SCA SCA

Moderately Not higher than
SICA* SCA SCA SCA
weak SICA* + 1

Not higher than Not higher than

Weak SICA* SCA SCA
SICA* + 1 SICA* + 2
* SICA or supporting institution’s credit rating, if any.
Source: ACRA.

Appendix 6
Table A6. Adjustment for state and shareholder support

Systemic importance level

Very high High Medium Low

Very strong Not exceeding Not exceeding

Parity Parity – [from 1 to 5 notches]
Level of state influence

SCA + 3 SCA + 1

Strong Parity – [ from Not exceeding Not exceeding

Not exceeding SCA + 3
1 to 3 notches] SCA + 2 SCA + 1

Moderate Not exceeding Not exceeding

Not exceeding SCA + 2 SCA
SCA + 3 SCA + 1

Weak Not exceeding

Not exceeding SCA + 1 SCA SCA
SCA + 1

Source: ACRA.

Contribution of the authors: the authors contributed equally to this article.

The authors declare no conflicts of interests.
The article was submitted 06.09.2022; approved after reviewing 08.10.2022; accepted for publication 14.11.2022.

16 Higher School of Economics

View publication stats