The Use of Credit Scoring in Moscow and St Petersburg

					          FINANCIAL SERVICES VOLUNTEER CORPS




Recommendations on the Use of Credit Scoring for

        Micro and SME Lending in Russia


                      April 2011
                                                                                               1




         Recommendations on the Use of Credit Scoring for

                         Micro and SME Lending in Russia

                                                April 2011




                                                Prepared by:

                                     David Snyder
               Vice President and Risk Management Manager, Wells Fargo


                                        Tim O’Brien
                       Regional Director for Russia and the CIS, FSVC




Recommendations on the Use of Credit Scoring for Micro and SME Lending in Russia. April 2011
The Financial Services Volunteer Corps
                                                                                               2




        Acknowledgments

        This commentary was informed from meetings we had with senior managers of a
number of financial institutions in Moscow, St. Petersburg and Yekaterinburg between
July-October, 2010. We’d like to thank all of the managers that took time out of their
busy schedules to meet with us and answer our questions. Findings presented here
represent to the best of our knowledge a summary of what we learned from our visits.

        The authors would like to thank Alisa Lockwood of FSVC who helped develop
and manage the online survey.




Recommendations on the Use of Credit Scoring for Micro and SME Lending in Russia. April 2011
The Financial Services Volunteer Corps
                                                                                                                  3

           Introduction

           Small and medium-sized enterprises (SMEs) account for a significant share of
world economic output. Among high-income countries, SMEs account for an average of
49% of national GDP; among lower-income countries they account for 29% 1. Yet in
many developing countries, SMEs have limited access to formal credit. Access to
finance for SMEs in the Russian Federation is particularly limited. A cross-country
comparison study conducted by the MFPA Centre for Economic Research found that
among 32 countries in Europe and North America, Russia ranked 28th in measures on
the availability of retail finance 2 and according to the Ministry of Economic Development
and Trade, SMEs in Russia account for less than 20% of GDP 3.

           A 2008 study sponsored by the U.S. Russia Centre for Entrepreneurship and
conducted by the Financial Services Volunteer Corps (FSVC) identified several barriers
to financing for SMEs in Russia, including bank requirements of collateral for most loan
requests, short-loan terms resulting in high loan repayment amounts permitting only
smaller loans, and SME difficulty in demonstrating creditworthiness through formal
financial statements 4. The study also emphasized that banks prefer to lend to larger
SMEs, who are presumably of lower risk, and that underwriting of SMEs, including of
smaller ones, is primarily a judgmental, case-by-case, process, which may not be
sufficiently efficient for lending to smaller SMEs, who comprise the vast majority of
businesses.

           A 2007 report by the Russian Microfinance Centre titled Financial Services for
Russian Small Businesses: Market Situation and Perspectives highlighted several
instruments to improve the effectiveness and efficiency of delivery of financial services
to small businesses5. One key instrument is the expanded use of credit scoring, which
when used appropriately, can help safely expand access to credit for small businesses.


1
    “Financial Access 2010: The State of Financial Inclusion through the Crisis”. CGAP, World Bank Group, 2010.
2
    MFPA Centre for Economic Research.
3
 Pre-crisis data estimated SME contribution to the national GDP at 17.5% Data provided by the Ministry of
Economic Development and Trade (www.economy.gov.ru)
4
    Dayal J, Drozda S, Wishon T. “Assessment of Obstacles to SME Finance in Russia”. FSVC, 2008.
5
 “Financial Services for Russian Small Businesses: Market Situation and Perspectives”. Russian Microfinance
Centre, 2007.

Recommendations on the Use of Credit Scoring for Micro and SME Lending in Russia. April 2011
The Financial Services Volunteer Corps
                                                                                               4

        Credit scoring is now emerging in Russia as a tool for measuring and managing
micro, small and medium-sized enterprises (MSME) credit risk. Since 2008, Fair Isaac
Corporation has worked with the Russian National Bureau of Credit Histories (NBKI),
the largest credit bureau in Russia, to produce the FICO score in Russia. Since 2009,
Equifax has produced a bureau score in Russia. Experian teamed with Interfax and
Sberbank to also provide bureau scores.

        The Benefits of Credit Scores

        Credit scoring can bring a number of benefits, including:

        •   An objective, standardized, consistent measurement of risk, resulting in all
            customers of similar risk receiving a consistent credit decision.

        •   Quantification of risk, which permits risk-based pricing

        •   Increased automation in decision-making; at underwriting, requests that the
            score indicates should be obvious approvals or declines can be auto-
            decisioned, which frees up underwriters to focus their expertise on more
            complex cases.

        •   In addition to underwriting, credit scores are very useful for account and
            portfolio management and for early warning systems. For example, the credit
            score of an account can be obtained on a monthly or quarterly basis, and if
            the updated score indicates a significant deterioration in customer credit
            quality, account management actions can be taken.

        Concerns with the Use of Credit Scores

        Credit scores are sophisticated tools that are subject to a number of risks. If not
managed appropriately, scores and their associated strategies can become ineffective
and do more harm than good. Some concerns with credit scores include:

        •   Bank staff inadequately trained to develop and manage score-based
            strategies and effectively monitor model performance.
        •   Management information systems that are inadequate for effective evaluation,
            monitoring and validation of the model




Recommendations on the Use of Credit Scoring for Micro and SME Lending in Russia. April 2011
The Financial Services Volunteer Corps
                                                                                               5

           •   Improper application of credit scoring models on products, subsets of
               applicants, or in geographic areas for which they were not developed without
               verifying effectiveness.
           •   Models can lose effectiveness over time for a variety of reasons, hence, the
               need for constant and thorough scorecard monitoring. 6


           In MSME lending, there are specific issues that need to be considered, including
the unit volume of accounts used for model development and applications being scored;
the size of the exposure; the heterogeneity of SME loans; quantification and validation
of cash-flow; collateral; and given the size and length of the customer relationship, the
model segments that should be developed and the cost-benefit of potential scoring
strategies.
           Introduction to this Paper

           This paper aims to expand on previous FSVC research and assess the use of
credit scoring in a sample of banks in Moscow, St. Petersburg and Yekaterinburg. We
visited with several banks and financial institutions in these cities and also with credit
bureaus such as the National Bureau of Credit Histories (NBKI) and Equifax. FSVC
also circulated an online survey. In commenting on the usage of scoring in Russia, this
paper draws on the findings from our meetings and survey responses, but does not cite
specific statistics or responding banks.


           Commentary
           Russia is emerging from a major credit crisis that resulted in negative GDP
growth in 2009 and essentially a freeze in lending to small businesses and households
in Russia from late-2008 through the end of 2009. The microfinance institution (MFI)
sector was significantly impacted, as the credit crisis triggered an outflow of savings
from MFIs in the fourth quarter of 2010 and MFI funding from commercial banks
declined. On average, the number of customers and assets in MFI loan portfolios
declined by 15% during 2009. The percentage of MFI portfolio balances that were at




6
    OCC Bulletin. “OCC 97-24”. Office of the Comptroller of the Currency, 1997.


Recommendations on the Use of Credit Scoring for Micro and SME Lending in Russia. April 2011
The Financial Services Volunteer Corps
                                                                                                               6

least 30 days past due increased from 3.3% to 7.3% between January, 2009 and
January, 2010. 7

           During the credit crisis, trust in credit scores as useful tools for risk assessment
declined, and the traditional benefits that credit scores bring became of lower priority.
Among some lenders, score-driven decision processes were halted and converted to
judgmental processes. Also, loan-decision authority levels were reduced at the branch
level and more decisions were performed centrally by senior credit officers or credit
committee. Approval rates declined and the time to loan decision increased.

           In 2010, as the Russian economy improved and government sponsored SME
support programs came online, financial institutions began to lend again to small
businesses and renewed interest in improving operational efficiencies and standardizing
loan decisions in the MSME sector. Thus credit scoring regained attention as a useful
tool for risk measurement and management.


           Definition of Micro and SME
           Many banks, even those with relatively few, large borrowers, are wondering
whether credit scoring would work well for their Micro and SME lending and whether or
not they have enough data to build and use a scoring model. The answer to these
questions depends in part on the definition of Micro and SME. The definition of SME
varies considerably around the world and even substantially across financial institutions
in Russia. In accordance to federal law 209-FZ on the “Development of SMEs”8, the
definition of SME in Russia includes individual entrepreneurs and firms with fewer than
250 employees and annual turnover of less than RUR 1B (approximately $34M) 9.
According to the RosStat, the government statistical agency, at the end of 2009 there
were more than 5.6M SMEs in Russia; the overall majority of which were individual
entrepreneurs (4M) and microenterprises (1.4M) 10.

7
    Russian Microfinance Trend Report, 2008-2009. Russian Microfinance Center, 2010.
8
  209-FZ law on “the Development of SMEs” classifies the sector as follows: Micro- up to 15 employees and annual
turnover of RUR 60M; Small – from 16 to 100 employees and an annual turnover of up to RUR 100M; Medium –
from 101 to 250 employees with an annual turnover of RUR 1B.
9
“ Financial Access 2010: The State of Financial Inclusion through the Crisis”.
10
  Sharov, AV; Deputy Director, Department of SME Development for the Ministry of Economic Development and
Trade. August 2010.


Recommendations on the Use of Credit Scoring for Micro and SME Lending in Russia. April 2011
The Financial Services Volunteer Corps
                                                                                               7

        Typically microloans can range from RUR 350,000 to RUR 3M (approximately
USD11,000-USD100,000 ). Medium loans can go up to RUR120M or more, depending
on the collateral. The average size for small business loans are in between Micro and
Medium. The vast majority of the unit volume is in the Micro end of the MSME
spectrum. Unit volume can range from portfolios as small as 50-60 medium-sized SME
relationships in the portfolio of a mid-sized regional bank to several hundred thousand
credit card customers in the Retail segment of a national bank.
        Recommendation: Credit scoring is most appropriate for large volume,
small, standardized transactions. Credit scoring is best suited for lenders with high-
application and/or high account volume, with standardized products, and where each
credit request or account is a small percentage of the total portfolio exposure. In Russia,
Micro is the most appropriate loan segment for credit scoring.
        A bank with several hundred thousand credit card accounts would be in a very
good position to build and use its own credit score. Banks with several thousand
microloans may not have sufficient performance history to date in order to build their
own robust scoring model, but they likely have the appropriate volume and loan size to
begin routinely testing the usage of generic credit bureau scores (at least as an
additional input to the loan decision).
        In MSME lending, as business and loan size increase, products become more
customized (less standardized), unit volume decreases, and the consequences of
making the wrong loan decision become greater. A bank with a small portfolio of
relatively large loans would not have sufficient volume to build its own credit score, and
credit bureau scores may or may not prove useful as an aid in decision-making.
        Recommendation: Credit scoring should not be used for automated
decisions for new credit facilities with loan amounts above RUR 3M, and will in
fact be most cost-effective for loan amounts up to RUR 1M.
        Decisions that are largely determined by scores or rules are best suited for small
loans, not only because these are typically of the highest volume and therefore
statistical models developed on these loans are more robust, but also because the cost
of the underwriter as a percentage of the loan amount decreases as the loan size
increases. The cost to manually underwrite a RUR 100M loan is typically not 100 times
greater than the cost to underwrite a RUR 1M loan. Therefore, as loan sizes increase,
using scores to achieve operational efficiencies becomes less important; more
important is the accurate assessment of risk and the avoidance of approving loans that


Recommendations on the Use of Credit Scoring for Micro and SME Lending in Russia. April 2011
The Financial Services Volunteer Corps
                                                                                               8

subsequently go to default. Because scores are best suited for small loans, for the
remainder of this paper we will discuss credit scores for MSME lending in the context of
loan amounts up to RUR 3M.
        One of the key discoveries in the United States in the pioneering days of the use
of credit scoring for small business lending in the mid-1990s was that for smaller
businesses, the credit risk of the owner of the business was strongly correlated with the
creditworthiness of the small business, itself. Thus, for the vast majority of MSME
volume in Russia, which is made up of micro and small loans, there is great potential for
using the newly developing credit bureau scores for decision-making.
        Although the FICO score has been available in Russia since 2008, to date bank
usage of credit bureau scores remains nascent. For example, when we talked with the
National Bureau of Credit Histories (NBKI) in the early Fall of 2010, we learned that by
that time, over 750 of Russia’s approximately 1,100 credit institutions were already
reporting to the NBKI, and that about 650 institutions were already routinely purchasing
credit bureau reports for use in decision-making. However, only about 20 banks (mostly
foreign owned) were actually purchasing credit scores from the bureau.
        In talking with the NBKI, Equifax and several banks, we learned that there are a
variety of reasons why the purchase of credit bureau scores is not more prevalent for
usage in decisioning of MSME loans. First, among the larger banks with more
experience in scoring, there is a concern about the effectiveness of the credit bureau
scores and a belief (at times actually confirmed through back-testing) that the bureau
scores do not rank-order risk in their customer populations, and a conviction that their
own internal scores are more predictive. For example, we heard that since the credit
bureaus do not distinguish between salaried versus individual entrepreneurs, the scores
are not specific enough to apply to their customer segments.
        Another reason that banks have not yet started using credit bureau scores is due
to their own systems constraints; they may not yet have the technology in place to
automate the ordering, retrieval and usage of the score on a real-time basis. A third
reason why banks are not yet using bureau scores seems to be because of a lack of
knowledge of their availability, and/or how to use them in their loan decisioning.
        Several larger, scoring-experienced banks, however, are using credit bureau
scores, and often in conjunction (a matrix-approach) with their internal scores and
decisioning rules. This approach is commonly used among larger banks in the United
States and has a lot of potential for MSME lending in Russia. A matrix approach (where


Recommendations on the Use of Credit Scoring for Micro and SME Lending in Russia. April 2011
The Financial Services Volunteer Corps
                                                                                               9

the bank uses two scores, one internally developed with data from its own customer
base, and the other, a generic bureau score developed on a broad spectrum of the
regional or national population) provides in some ways the best of both worlds, and
strategies can be implemented that take both scores into consideration.
        Recommendation: Financial institutions should purchase the credit bureau
scores, which in 2010 cost about five rubles (about 17 U.S. cents) per requested
score, and engage in a back-testing effort to see how well the scores rank-order
risk in their customer segments.
        For most banks, which do not have the volume or internal capacity to build and
manage their own models, the generic scores are their only scoring option, so it is
worthwhile to determine the usefulness of these scores in their customer population.
Even for banks with the volume and capacity to develop custom-scores, the generic
scores may very well prove to be a useful additional data input.


        Data inputs for a MSME credit scoring model
        The 2007 FSVC study titled Assessment of Obstacles to SME Finance in Russia
emphasized the importance of building infrastructures to capture data on loan
originations and performance. This is a prerequisite to the successful use of credit
scoring. An essential requirement for credit scores is the presence of standardized,
consistent, accurate and verifiable data that is available at the time of scoring.
        Data is at the center of a credit scoring model, and there are certain inputs that
are the cornerstones of a robust, effective model. One of the most important data
inputs to a model is how the customer has performed on existing or previous loans with
the financial institution (FI). How customers have performed on previous loans with the
FI is very predictive of how they will perform on the next one. Much of the loan
application volume of many of the FIs we visited is to existing customers requesting a
new loan or the renewal of an existing one.
        Recommendation: FIs planning on using credit scoring should segment
their customers based on the amount of performance history they have about the
customer. The more direct knowledge FIs have about the customer’s previous
performance, the more able they will be to safely apply scoring models and
strategies. Institutions embarking on using credit scoring can do so in stages,
with the first model and application applied to existing customers of their bank.



Recommendations on the Use of Credit Scoring for Micro and SME Lending in Russia. April 2011
The Financial Services Volunteer Corps
                                                                                               10

         Another cornerstone of a MSME credit scoring model comes from customer
behavior on their deposit and current accounts with the bank. Among many of the FIs
with whom we met, most of their new credit customers did not yet have deposits with
their institution. There were a variety of explanations for this, among them that many
MSME customers do not trust banks with their deposits, or that customers prefer to
diversify their financial relationships across multiple institutions, or for a variety of
reasons choose not to keep their excess funds in a bank, or simply that the bank had
not prioritized targeting the MSME sector for deposit growth. Whatever the reason may
be, it is important to recognize that credit scoring models for MSME lending are much
stronger when developed with data from the institution’s own depositors and are
typically most successful in their application on this customer segment.
         Recommendation: FIs seeking to use credit scoring for decisioning MSME
loans should proceed very cautiously in removing requirements related to the
minimum “length of time” as a customer.
         The riskiest customer segment for scoring is the one with which the bank has no
prior experience. Some banks require SME borrowers to have had a deposit account
with their institution for at least six months before becoming eligible to apply for a loan.
This is a prudent risk management practice. Six recent months of deposit transaction
behavior history will considerably strengthen the risk assessment of a prospective
borrower and a credit scoring model, for which deposit transaction history is typically
one of the more important variables.
         As performance on existing or previous loans with the institution and deposit
history are behavioral measurements obtained from internal databases, the third
cornerstone of a strong credit scoring model comes from external sources: the credit
bureau report. A credit scoring model and its associated strategy will typically
incorporate data from the credit bureau report, and also additionally may include credit
bureau scores, as well. Therefore, we recommend that credit bureau data be used as
an input to MSME scoring models and also for additional decision-rules outside of the
score.
         It was mentioned earlier that Micro is the most appropriate loan segment for the
use of credit scoring. Micro in this context refers to the size of the loan, not necessarily
to MFIs, specifically. MFIs are potentially good candidates for the use of credit scoring,
but in order to do so, they will be most successful if they can participate in and use



Recommendations on the Use of Credit Scoring for Micro and SME Lending in Russia. April 2011
The Financial Services Volunteer Corps
                                                                                               11

credit bureaus. Currently very few of the over 2,000 11 MFIs in Russia are reporting to
credit bureaus or using the bureau reports. 12 A 2010 report titled Growth and
Vulnerabilities in Microfinance, which analyzed repayment crises in four countries on
four different continents, identified common causal factors to the crises, including 1)
concentrated market competition and multiple borrowing; 2) overstretched MFI systems
and controls, and 3) erosion of MFI lending discipline. 13 Participating in credit bureaus
and using reports during the account origination and management process can help
MFIs identify applicants with multiple loans, over-indebtedness, and repayment
problems. We suggest that reporting and usage of credit bureau reports is a
prerequisite to successful use of credit scoring.
           The fourth cornerstone of a robust credit scoring model for MSME lending is
quantification of cash-flow available to service debt and a debt-service coverage ratio or
debt/income ratio. Cash-flow can be determined through tax declaration documents or
through analysis of audited or self-prepared financial statements.
           Almost all institutions we visited reported that for new customers they confirmed
the existence and viability of the borrowing business through onsite field visits. The
premise visit is a very important component of the overall risk assessment, as the FIs
use the visit to also conduct a reality check on stated business revenues. Premises
visits can be expensive for the smallest loans. For micro loans up to RUR 300,000 to
new credit customers, FIs may want to implement alternatives to the premise visit which
require some previous history with the customer, such as a requirement that the
customer has had deposits with the bank for at least three to six months.
           There are other important data inputs to MSME scoring models, including
industry and time in business, and collateral type and quality, to name a few. In order to
build a scoring model with these key inputs, it is necessary to have databases on loan
originations and performance, collateral type and quality and ideally of deposit accounts
that can be linked together with unique customer identifiers. Among the FIs we
interviewed, we found a variety of situations, including 1) the FI had a rich data
warehouse and was already mining the data; 2) the FI was collecting the data but had
not yet done any analysis; 3) the FI did not have a loan originations database, so

11
     Russian Microfinance Trend Report, 2008-2009. Russian Microfinance Center, 2010.
12
     Personal communication from Mikhail Mamuta, Russian Microfinance Centre, 2011.
13
     Growth and Vulnerabilities in Microfinance. Focus Note. No 61. February 2010., CGAP


Recommendations on the Use of Credit Scoring for Micro and SME Lending in Russia. April 2011
The Financial Services Volunteer Corps
                                                                                                                12

therefore analysis of the credit profile of applications, stratified by application status
(approved/declined), was not possible.


        Score Usage
        A common question we encountered among FIs using or planning to use credit
scores is whether the score alone can be used to make the approve/decline decision or
whether it should be used in conjunction with other business rules or as an aid to the
underwriter, who then makes the final decision judgmentally.
        Recommendation: In our opinion, a credit score cannot be used alone to
make a decision. Credit scores should always work within the constraints of the
FIs credit policy, and approval (and decline) decisions should make common
credit sense, regardless of the score decision.
        One reason why scores should always work within the constraints of policy is that
a new score is typically built with performance data of loans that have been approved by
underwriters. Typically the vast majority of approved requests have met all credit policy
conditions (except for a small percentage that are selectively approved via policy
exception), and therefore the credit score is a conditional one, specific to the applicant
population that meets the policy criteria. We cannot assume that loans in the same
score range will perform similarly, regardless of whether the borrower met all policy
criteria. One of the contributing factors to the subprime crisis in the United States was
due to the fact that lenders assumed that the score explained all of the credit risk, even
when the other underwriting requirements had changed substantially. 14 Therefore, the
score should be used within the bounds of the credit policy, and score-approved
applications that do not meet credit policy rules should be routed to an underwriter for
additional review.
        Even though scores should not operate on their own, this does not mean that an
underwriter should review all applications, regardless of the score decision. One of the
primary benefits of credit scoring for micro loans is the operational efficiency gains
achieved by auto-decisioning obvious approvals and declines and permitting
underwriters to focus on the more complex decisions. Therefore, for loan requests up to



14
  Rona-Tas A, Hib Stephanie. “Consumer and Corporate Credit Ratings and the Subprime Crisis in the U.S., with
Some Lessons for Germany”, prepared to the SCHUFA, Weisbaden, Germany, September, 2008.
http://weber.ucsd.edu/~aronatas/The%20Subprime%20Crisis%202008%2010%2004.pdf


Recommendations on the Use of Credit Scoring for Micro and SME Lending in Russia. April 2011
The Financial Services Volunteer Corps
                                                                                               13

RUR 3M, FIs implementing scoring can develop auto-approval and auto-decline
strategies that are built with a combination of score decision and policy rules.


        Risk-based Pricing
        Our discussions with FIs revealed that some are using risk-based pricing and
others are not. The reasons for not using risk-based pricing include 1) they do not
compete on price (all customers are assigned the same price), 2) they would like to
assign risk-based pricing, but the origination system does not permit pricing stratified by
risk-level, 3) they do not use risk-based pricing, but are planning on implementing it in
the future. The more analytically advanced FIs are using risk-based pricing for MSME
loans, and to our knowledge one of the major banks in Russia is already doing so, as
well. Given the current usage and the increasing use of credit bureau reports and
scores, FIs that do not measure borrower risk and do not assign risk-based pricing are
vulnerable to deteriorating portfolio quality due to adverse selection and adverse
retention. If two institutions have the same overall interest rate, but Bank A prices all
borrowers the same, and Bank B assigns risk-based pricing, then Bank B will charge
low-risk borrowers a lower rate than Bank A, who assigns the same rate to all
borrowers. This will result in the low-risk borrowers going to Bank B, and all the high-risk
borrowers (who actually a receive a lower rate at Bank A) going to Bank A. This
phenomenon is known as adverse selection. A similar problem can occur with existing
loans. The absence of risk-based pricing can lead to adverse retention, where low-risk
borrowers gravitate towards the bank that has lower interest rates, and the high-risk
borrowers remain where they receive more favorable rates.
        Because risk-based pricing is so important for combating adverse selection and
adverse retention, FIs will need to price according to risk in order to remain competitive.
This compels FIs to obtain quantitative risk measurements of their applicants and
borrowers. For the FIs that do not have sufficient volume or infrastructure to build and
maintain their own scoring models, the imperative is to begin obtaining credit bureau
scores on their borrowers and then validating that the scores rank-order risk in their
customer population.


        Model Monitoring
        The scorecard monitoring process involves periodic analysis of trends in the
distribution of scores and the underlying model variables (also known as characteristics)


Recommendations on the Use of Credit Scoring for Micro and SME Lending in Russia. April 2011
The Financial Services Volunteer Corps
                                                                                               14

in the applicant population, and trends in the model’s ability to rank order and predict
risk on recent account cohorts. Front-end analyses compare application score
distributions over time and seek to determine the reasons for shifts in the score
distribution. Back-end analyses determine actual performance by score-range and seek
to determine whether the score can still rank-order and accurately predict default
rates. 15

           Both front-end and back-end analyses are critical components of credit scoring
model management. Front-end analyses are important because the assumption is that
the risk-profile of the current population being scored is similar to the population on
which the model was developed. If the application scores distribution changes
significantly, this either means that the risk profile of the applicant population has
changed, or that the data being sent to the scoring model has somehow changed,
possibly due to system error. The first case is of concern, because the model may not
be effective on an applicant population that is significantly different from the population
on which the model was built. The second case is of obvious concern, because the
model does not know it is receiving erroneous data; it is up to the risk manager to
uncover that and make changes accordingly. If the front-end analyses are not
conducted, then there will be no way of being alerted to the situations described, which
can have adverse consequences for risk management.

           Back-end analyses are essential, because they are used to confirm that the
model still rank-orders risk. If a model does not rank-order risk, it is useless.

           Among the FIs that we visited that are currently using credit scoring, regarding
scorecard monitoring we found a variety of practices, including 1) the FI had not done
any front-end or back-end analysis, 2) the FI had conducted a back-end validation
analysis to confirm that the score rank-orders risk, but they had not done any front-end
analysis, 3) the FI had done both back-end analysis and some front-end analysis
(comparison of score distributions over time), but they had not analyzed the reasons
driving changes in score distributions.

           Recommendation: FIs using or planning on using credit scoring should
integrate a routine, periodic (at least quarterly) front-end and back-end scorecard


15
     “OCC 97-24”.


Recommendations on the Use of Credit Scoring for Micro and SME Lending in Russia. April 2011
The Financial Services Volunteer Corps
                                                                                               15

monitoring process. In addition, the FIs should have on staff experienced
portfolio risk analysts and managers who monitor the score, account originations
and portfolio performance. Trends in volume, profile and performance that vary
from expectations should be investigated to determine the underlying reasons for
the trend and to guide changes to strategy or policy as warranted.

        Conclusion
        The FSVC study titled Assessment of Obstacles to SME Finance in Russia
emphasized that FIs should make the commitment to build and maintain risk
management analytical departments within their institution. For successful use of credit
scoring, this is probably the most important factor. Management of a successful lending
operation that uses credit scoring is an ongoing process of model and strategy
development, implementation, evaluation and adjustment. An essential requirement is
the presence of portfolio skilled risk managers who know how to manage the scoring
process and portfolio and to solve the inevitable problems that arise. Credit scores are
simply tools that measure risk. These tools must be managed with caution by skilled
and experienced practitioners.




Recommendations on the Use of Credit Scoring for Micro and SME Lending in Russia. April 2011
The Financial Services Volunteer Corps
                                                                                               16

About the Authors

David Snyder is a Vice President and Risk Management Manager with the Business
Direct Division/ Customer Information and Risk Management (CIRM) of Wells Fargo
Bank in San Francisco, CA. In this capacity, Mr. Snyder manages team of five senior
analysts responsible for developing data-driven strategies to profitably grow a small to
medium sized enterprise (SME) loan portfolio of $30 billion in balances outstanding.
Internal projects included the testing of operational cost effectiveness of Wells Fargo’s
scorecard for SMEs and auto-approve loans, review and analysis of the scorecard
programs, the simplification of loan underwriting process for SME clients, and other data
driven analyses and projects. Mr. Snyder also held the position of Assistant Vice
President for SME Portfolio Management. Prior to his work with Wells Fargo, Mr.
Snyder worked for the University of California, San Francisco and California Department
of Health Services as a Senior Epidemiologist. In this capacity, Mr. Snyder conducted
analytical and quantitative studies on tuberculosis which led directly to research papers
and international presentations, including several for the World Health Organization.


In addition to his professional work, Mr. Snyder has undertaking several consulting
projects for the Financial Services Volunteer Corps and other organizations. These
include a consultation on Credit Scoring and Risk Management for SME lending with
Punjab National Bank, New Delhi, India; a consultation and training on Credit Scoring
and Portfolio Risk Management for SME Lending at the National Institute of Bank
Management, Pune, India; and, a study and publication titled: The Potential for Credit
Scoring for SME lending in Kenya, Financial Sector Deepening Trust, Nairobi, Kenya
(June/July 2008).


Mr. Snyder holds an MBA in Finance from San Francisco State University, a Masters of
Public Health in Epidemiology from the University of Hawaii’s School of Public Health,
and a Bachelors diploma in Spanish from the University of Hawaii at Manoa. He has
also successfully completed several certification courses on Risk Management and
Credit Scorecard Development.


Tim O’Brien is the Moscow-based Regional Director for FSVC. In this position, he
oversees the implementation of FSVC’s technical advisory programs in such areas as
regional bank training and consultations on such topics as risk management, small and

Recommendations on the Use of Credit Scoring for Micro and SME Lending in Russia. April 2011
The Financial Services Volunteer Corps
                                                                                               17

medium-sized enterprise (SME) lending, anti-money laundering, and financial literacy.
Prior to joining FSVC, he served both as Manager of Business Development and
Project Manager for the Pragma Corporation, an international development consultancy.
In this role he helped devise the organization’s strategic plan and implemented technical
assistance activities in the area of economic development. The technical assistance
projects in his portfolio included assisting the development of SMEs, improving access
to credit for underserved populations, supporting public- and private-sector accounting
reform, and strengthening corporate governance.


Mr. O’Brien holds an MBA degree from the University of North Carolina Kenan-Flagler
Business School and a B.S. degree in Economics from the University of North Carolina
at Chapel Hill. He speaks Russian and has a strong knowledge of Ukrainian.




Recommendations on the Use of Credit Scoring for Micro and SME Lending in Russia. April 2011
The Financial Services Volunteer Corps
                                                                                               18

About FSVC

The Financial Services Volunteer Corps (FSVC) is a not-for-profit, private-public
partnership whose mission is to help build sound banking and financial systems in
transition and developing countries. Sound financial infrastructure, together with the
rule of law, is necessary to mobilize domestic savings, attract foreign investment,
deepen international trade linkages, and create conditions that promote lasting
economic opportunity.


FSVC was founded in 1990 by the late Cyrus Vance, former U.S. Secretary of State,
and John Whitehead, former Co-Chairman of Goldman Sachs and former U.S.
Deputy Secretary of State, at the request of President George H. W. Bush. FSVC’s
Board is currently chaired by Mr. Whitehead. Paul Volcker, former Chairman of the
Board of Governors of the U.S. Federal Reserve System, is the Honorary Chairman
of the Board. John Walker, Chairman of Richina Pacific Limited and a leading
international lawyer, serves as the Vice Chairman of the Board. FSVC’s President
and CEO is J. Andrew Spindler.


FSVC’s core work concentrates on strengthening commercial banking systems,
developing central bank capabilities, and building capital markets. Major additional
areas of work include the legal framework for the financial system, payments system
development, pension reform, and the combating of money laundering and financial
corruption.


FSVC structures practical, results-oriented technical assistance and training
missions staffed by financial sector practitioners who serve as unpaid volunteers.
Over the past 18 years, approximately 8,000 experts from the financial, legal, and
regulatory communities have taken part in more than 2,200 FSVC missions,
reaching nearly 34,000 counterparts in 50 developing and transition countries.

By recruiting currently employed professionals at the peak of their careers to serve
as volunteers, FSVC is able to provide technical assistance that is objective,
independent, and state-of-the-art. In addition, recipients of FSVC’s assistance
develop valuable professional relationships with volunteers and establish



Recommendations on the Use of Credit Scoring for Micro and SME Lending in Russia. April 2011
The Financial Services Volunteer Corps
                                                                                               19

institutional linkages with sponsoring institutions. These institutional linkages
provide significant additional benefit to counterparts over the long term.


FSVC-arranged assistance is also highly cost-effective: FSVC has delivered over
$175 million in technical assistance by leveraging the pro bono service of expert
volunteers, more than triple the amount of U.S. Government grants received from
the U.S. Agency for International Development (USAID) and the U.S. State
Department.


Further information about FSVC is available at its website: www.fsvc.org.




Recommendations on the Use of Credit Scoring for Micro and SME Lending in Russia. April 2011
The Financial Services Volunteer Corps

				
DOCUMENT INFO
Shared By:
Tags:
Stats:
views:2
posted:1/27/2012
language:
pages:20