Retail banks and big data

To learn more about the intersection between big data and risk management at banks, the Economist Intelligence Unit EIU surveyed 208 risk management and compliance executives at retail b

Trang 1

Retail banks and big data:

Big data as the key to better risk

management

Trang 2

The business of banking depends on evaluating risks and then acting on those insights In theory, more information should yield better risk assessments, which is why big data and its associated tools couldn’t have arrived at a better time

The ability to harness larger and more diverse data pools in support of business decision-makers holds the promise of both reducing losses by managing risks and increasing revenue by highlighting business opportunities Successfully managing risks today requires that bankers identify, access and analyse trusted data and share their results across the bank

The growth of risk

As recent headlines bear out, risks increase with complexity, and complexity has grown across every dimension of the banking industry Banking has become more concentrated, which means that a handful of giant institutions must coordinate a diverse array of products, processes, technologies, organisational structures and legal contracts

Financial innovation leads to new instruments and specialties Markets are more interconnected and

information traverses those connections more rapidly As a result, when things go wrong, volatility can switch markets from tranquil to turbulent almost instantly, leading to “volatility clustering” that can result in liquidity crises like those seen in the 2007-2009 financial crisis or the

2001 bursting of the dotcom bubble

Clearly, the landscape of banking risks is vast

“We have identified 13 different types of systemic risk: cyber-risk; high-frequency trading risk; counterparty risk; collateral risk; liquidity risk; and the list goes on,” says Mike Leibrock,

vice-president of systemic risk at the Depository Trust Clearing Corporation (DTCC), which provides clearing and settlement services to large banks

“And there is an entire category of interconnectedness risks, which arise from linkages among a handful of key banks for clearance and settlement activities.”

As regulators—and therefore also the institutions that they regulate—focus as never before on identifying, measuring, and managing emerging risks across the financial system, data management practices are also changing

Big data as the key to better risk management

Trang 3

The promise and potential of big data

Banks are experts in manipulating the rows and columns of numbers captured from past transactions and stored in vast data warehouses

On a daily or even intra-day basis, banks package these facts in the form of reports for credit or finance officers to review for trends and outliers

Big data is different It is vast in scope, varied in form and instantaneous in velocity, encompassing data from mobile devices, social media applications and website visits as well as information from third-party providers of credit, spending, auto and legal data It promises to reveal hidden consumer behaviors that may not be immediately apparent even to those with highly sophisticated knowledge and experience Big data potentially allows banks

to measure and manage risk at an individual customer level, as well as at a product or portfolio level, and to be much more precise in credit approvals and pricing decisions

To learn more about the intersection between big data and risk management at banks, the Economist Intelligence Unit (EIU) surveyed 208 risk management and compliance executives at retail banks (29%), commercial banks (43%) and investment banks (28%) in 55 countries on six continents The results demonstrate that growing numbers of bankers are embracing the analysis and sharing of big data, but that they still face

challenges in applying the results to delivering superior risk management performance—especially around liquidity and credit risk

The survey asked executives to rate their own institution’s performance in controlling and mitigating risk Those that rated their institution above average were also more likely to use:

l basic big data tools to integrate, manipulate and access structured and unstructured data (35% for the above-average risk managers versus 7% of those rated average or below)

l more advanced big data tools such as predictive analytics and visualisation (33% versus 8%)

In other words, banks that perform better are more likely to use a variety of different

methodologies, including both basic and advanced analytics, to understand and manage their risks Moreover, they’re more likely to bring large amounts of data to bear on risk management problems

Investments in big data to support risk management

In addition to the four regions, survey respondents came from three types of institutions: 43% from commercial banks and the rest divided evenly between retail and investment banks All three are more concerned about liquidity and credit risk than other types of risk At the same time, the

importance they assign to different types of risk varies by industry and region

l Retail banks are more concerned about credit risk (53% versus 43% among commercial and investment banks)

l The commercial banks tend to be slightly more concerned about market risk (28% versus 23% among investment and retail banks)

Regional breakdown of survey respondents

% of all respondents North America Asia-Pacific Western Europe Latin America Middle East Africa Eastern Europe

Source: Economist Intelligence Unit survey, July, 2014.

25 25 24 10

8 7 2

Trang 4

l Investment banks, meanwhile, tend to be more concerned about operational (29% versus 19%) and compliance risk (20% versus 14%)

From a regional perspective, Asia-Pacific and the emerging markets have heightened concern about exposure to market risk, while Europe has elevated concerns about both liquidity and credit risk

Across all regions and industry sectors, the vast majority of banks already support risk management

by investing in big data or expect to do so soon

Four out of five (81%) routinely provide comprehensive reports on the bank’s risk profile to senior executives, and another 15% plan to do so

in the next three years Almost all are committed to

pushing risk management information up to top decision-makers

“The types of things that are most commonly requested are the Volcker metric-related variables that show our liquidity, balances, risk ratios and exposures,” says Wells Fargo Chief Data Officer Charles Thomas

But the question remains: Do they have access

to the right big data tools to do so and to be truly effective?

Just over four out of ten (42%) respondents currently have the ability to integrate, manipulate and query big data when creating risk profiles Almost half (47%) have plans to invest in these

Africa, Latin America, Middle East Asia-Pacific

North America Europe

In which risk areas will your organisation face the greatest challenges in the next three years?

% of all respondents Liquidity risk

Credit risk

Foreign investment risk

Market risk

Operational risk

Compliance risk

44 45 54 58 40

41 40 58 33

33 37 26

33 31 21

17 23 25 29 11

15 8

15 25

Which of the following risk management techniques/tools

is your organization currently using and which do you expect

it to be using in three years?

% of all respondents Comprehensive information on the organisation’s overall risk profile routinely provided to the Board and

in 3 years No plans for usingin 3 years

81

3

15

Trang 5

tools over the next three years

The proportions are slightly lower for advanced big data tools, such as predictive analytics and data visualisation: 41% use them now and 44%

expect to obtain them during the next three years

Still, an overwhelming majority of banks—retail, commercial, investment, from every continent—is committed to and leveraging the power of big data

Tackling the two greatest risks:

credit and liquidity

The bankers surveyed believe that, over the next three years, liquidity and credit risk will pose the greatest challenges for their institutions They also say these two areas of risk reflect the greatest potential for big data and its associated tools to make an impact on improving risk management

Why the intense focus on credit and liquidity risk? Banks are in the business of selling liquidity

Pursuing profits necessarily thins capitalization and leaves little margin for error “The common feature of the financial services industry—banks, brokerage firms, hedge funds—is that there is a small amount of equity and a large amount of financing supporting the firm’s assets,” says Robert

Chersi, former CFO of Fidelity Investments and now

a professor of finance at Pace University in New York “Financial services firms rely on other people

to finance them, and most of that financing is short-term It can disappear in an instant.”

Credit and liquidity risk represent two faces of the same phenomenon Banks borrow via short-term instruments in order to finance the longer-term instruments that they sell to their customers That leverage can be lost quickly as funding is withdrawn At best the bank is left without products to sell; at worst, as occurred with Lehman and Barings, the liquidity whipsaw can destroy the bank—but this kind of disaster scenario is typically preceded by expectations that the bank’s creditors will default

The problem with forecasting liquidity crises to date is that liquidity risk has been difficult to model “It’s a risk that only materialises in extreme situations and is very much a binary risk,” says Michael O’Connell, a managing director at Aon Risk Solutions But according to survey respondents, the use of big data offers the promise of linking seemingly unconnected external events in real time—events that could presumably precede a

What is your organisation using now and what do you expect in three years?

% of all respondents

Comprehensive data on the organisation’s risk profile routinely provided to the board and senior executives Basic big data tools to integrate, manipulate and access structured and unstructured data Advanced big data tools such as predictive analytics and data visualization

Now In 3 years

81 15

In which risk areas will your organisation face the greatest challenges in the next three years?

% of all respondents Liquidity risk Credit risk Foreign investment risk Market risk Operational risk Compliance risk Source: Economist Intelligence Unit survey, July, 2014.

50 45

32 25

22 16

Trang 6

liquidity crisis such as rising credit spreads or a flight to quality Running a close second was the ability to predict the amount and cost of capital required in stressful market situations

Fraud applications

A large sample can reveal rare events that don’t show up in small data sets When events occur infrequently—credit card fraud, for instance, occurs in perhaps five out of every 1,000 transactions—millions of transactions become necessary to generate a usefully large sample of fraudulent ones

It’s not hard to predict events that occur near the center of a probability distribution, but it can

be quite hard to predict events that occur far out

on the edges Only when you have collected a large sample of outliers can you think about how to predict them

Survey participants were well aware of this application of big data They said that the single most useful big-data opportunity in preventing

credit fraud was the near-instantaneous contacting

of customers to verify suspicious transactions, with 45% citing it as worthwhile

Next came using predictive models to distinguish between legitimate and fraudulent transactions The third biggest opportunity highlighted by respondents involved tracking spending behavior across 100% of transactions to detect fraud by playing the game of “Which of these is not like the others?” The key phrase in this question is “100% of transactions”: Data storage is

so cheap compared to previous generations, says

Mr Thomas, that when it comes to saving data, the question becomes “why not?”

Credit applications

Just as big data combined with predictive analytics can help in predicting fraud, it also has

applications in predicting loan defaults Survey respondents pointed this out, saying that the primary big-data opportunity in the lending area is monitoring borrowers for events that may increase

Equity as a percent of total capitalisation for selected industries

% Internet software and services Computer software Household products Computer services Integrated oil and gas Healthcare services Farms and agriculture

Retail Transportation Telecom services Utilities Banks Brokerages All financial services

Source: S&P/Capital IQ.

96 92 88 85 77 77 69 69 60 60 51 25

24 16

Which of the following areas presents the biggest opportunities for Big Data to improve performance in meeting liquidity requirements?

Ability to interpret seemingly unconnected external events in real time

Improved accuracy of capital cost scenarios Automated production of compliance data and reports

46 44 38

Trang 7

the chances of default (cited by 45%) The executives didn’t just highlight the opportunity—

they also went further in saying that big data had helped them achieve useful results When asked about success in applying big data to risk management activities, the top two results were related to credit

The problem is that data inevitably goes stale:

“When you apply for a mortgage, you provide the bank with current information on your assets and your employment,” says Ozgur Kan, who heads the Berkeley Research Group’s Credit Risk Analytics practice “After that, they do not collect more information about you, and don’t really know whether there was a change in your

circumstances.”

That leads to scenarios tailor-made for behavioural models powered by data from a mix of sources: payment behavior, interactions with the bank via the website or call centers, anything available from the three big credit bureaus, and potentially social media activities and other public sources of information While there has always been a large volume of data on default and recovery rates for borrowers and loan structures, that data can now be supplemented by behavioural information, from both internal and external

sources, and often updated on a timelier basis In-house data typically covers what was purchased, the amount, the date, time and location, and aspects such as recent changes of address or authorisations for others to use cards Data from external data sources (such as credit scores, location data, or online behavior patterns) can not only increase accuracy, but also cut down

on false positives (which can reduce revenue, as they cause the bank to deny valid transactions)

Of course, the costs of outside data acquisition can be high, and internal data offers an advantage that outside data can never replicate: It typically revolves around customer touchpoints—emails, website usage, call centers—and offers a deep view into the organisation’s interaction with customers that third-party vendors cannot replicate Says Mr Thomas “We can do text mining on phone calls and merge that with transactional, demographic and product data, and the result is a robust data set that enables us to have a much better

understanding of who our customers are, what their patterns are, and what sorts of triggers we might need to identify.”

Limiting the discussion to avoiding losses—the traditional focus of risk managers—fails to give big data its due “It’s not just for risk, and it’s not just

Which of the following risk management activities has Big Data been most successful?

% of all respondents using Big Data tools Preventing credit card fraud Guarding against loan defaults Meeting liquidity requirements Supporting compliance and reporting Anticipating market trends

31 26

24 9

8

Which of the following areas presents the biggest opportunities for Big Data to improve performance in preventing credit fraud?

% of all respondents Rapidly contacting customers to verify suspicious transactions based on real-time analysis Using predictive models to distinguish between legitimate and fraudulent transactions Tracking spending behaviour across 100% of transactions

45 41 32

Trang 8

for marketing and sales; it’s really for both,” says

Mr Thomas

The advantage of centralised approaches

Bank executives were also asked about the current role of their organisations’ analytical teams in managing risk exposure The most common approach, cited by 38% of respondents—almost half of the respondents in Europe and North America and between one-quarter and one-third in Asia-Pacific, Africa, Latin America and the

Mideast—is separate analytics teams with the analytical and subject-matter expertise needed to focus on specific areas of risk management On the other hand, when the results are taken together, the survey found that centralised enterprise-level approaches have been adopted by nearly half of

respondents

The most common centralised approach, cited

by 29% of respondents, involves creating analytics teams that respond to requests for service from risk and compliance users throughout the organisation Nearly one in five (19%) point to an even broader approach: multidisciplinary analytics centres of excellence that develop specialised skills and deploy standards and best practices across the organisation The multidisciplinary-centre-of-excellence approach is most prevalent among the Asia-Pacific respondents (29%) and least popular

in North America (15%) and Europe (11%)

The survey suggests centralised analytics groups are the most effective way of organising analytics expertise Respondents who rate their firms as well above average in assessing and mitigating risk are more likely to say that their bank uses one of the

Which of the following areas presents the biggest opportunities for Big Data to improve performance in guarding against loan defaults?

% of all respondents Monitoring borrower behaviour to anticipate and respond to default risk Enabling simulation of loan risk-pricing models Creating transparency by increasing risk visibility across the organisation Executing on-demand bank-wide stress testing Using predictive analytics

to assess borrower risks Creating an integrated 360-degree

view of the customer Using new data sources to enhance traditional credit scores Minimised response times between analysis and action

45 42 39 38 38 25

19 18

Which of the following statements best describes the current role of analytics teams

in managing your organisation’s risk exposure?

We have separate analytics teams that combine analytical and subject matter expertise focussed on specific areas of risk management

We have a centralised analytics team that responds to requests for service from risk and compliance users throughout the organisation

We have a multidisciplinary analytics centre of excellence that develops specialised skills, standards and best practices for the organisation Each functional area or business line employs its own risk analysts

We do not have a structured approach to the use of risk analytics

38 29

19 9

3

48

Trang 9

two centralised approaches (more so among commercial and investment banks than among retail banks) They’re also slightly more likely to use the service-center approach and a great deal more likely to adopt an analytics centre of excellence approach

Among top performers, 27% also have analytics centres of excellence that cut across disciplines and functions, compared with 17% of lower-performing firms Conversely, these high performers are less likely to use separate analytics teams that focus on a single area of risk

management, which is the most common approach across all banks

Key takeaways

Risks grow as markets become more tightly linked, banks become more concentrated, and banking organisations become more complex Regulators are demanding more metrics, more transparency, and better documentation of data Although banking has always been built on data, today’s data

is bigger, faster and more varied, requiring new and different tools Moreover, big data also holds more promise for mitigating risk and recognising opportunities, especially when novel and diverse data sources are integrated into traditional risk

management, underwriting and sales frameworks Banks see liquidity and credit risk as presenting the biggest challenges They also see those two types of risk as offering the biggest potential for improvement Many survey participants hope that big data can help the bank anticipate liquidity crises However, the more common applications revolve around predictive modeling for fraud prevention and closer monitoring of borrowers to predict loan defaults

Almost all banks are investing in big data to improve their risk management, but the banks that

do a better job at managing risk are moving more aggressively As big data and risk expertise grows more specialised, the best-performing banks— especially commercial and investment banks—are moving towards more centralised units that can develop expert skills, common standards and best practices that support and enhance their

organisations

Finally, the same big data infrastructure used to mitigate risks can also be used to pursue new sources of revenue “Whether it’s guarding against fraud or selling something new, being able to pull data from 80 different businesses enables us to get ahead of problems before they’re problems,” says Wells Fargo’s Mr Thomas

In the summer of 2014, the EIU conducted a global survey of 208 banking executives, with sponsorship from SAP, seeking insights into how banks are using big data to improve risk management and compliance performance

More than half of survey respondents were C-level or equivalent executives, and the remainder held SVP/VP/Director positions in risk management (63%) or regulatory compliance (38%) All respondents work for retail, commercial

or investment banks North America, Europe and

Asia-Pacific each account for about one-quarter

of the survey sample, with the remainder coming from Latin America (10%), Middle East (8%) and Africa (7%)

All of the respondents’ organisations have annual revenues of more than US$500m By size, they are roughly equally divided into three groups: the largest banks ($5bn or more in annual revenue), small banks ($500m to $1bn in annual revenue) and those falling between the two groups

About the survey?

Trang 10

Whilst every effort has been taken to verify the accuracy of this information, neither The Economist Intelligence Unit Ltd nor the sponsor of this report can accept any responsibility or liability for reliance by any person on this white paper or any of the information, opinions or conclusions set out in the white paper.

Định dạng
Số trang	11
Dung lượng	384,43 KB