Related Blogs


More Models, More Regulations, More at Stake

Comply + Compete webinar
To some degree, we're living in a world where we are cursed with our own success. Financial institutions have seen tremendous benefits from analytics, and as a result, they are using predictive models on an increasingly broader scale, to measure capital reserve requirements and manage complex customer decisions. But as my rap doppelganger would say: “More Models, More Problems.”

The greater complexity and number of predictive models in use makes it even more difficult to track and manage model performance, not to mention comply with regulatory requirements. Since the financial crisis, banking regulators have increased their scrutiny of how institutions use predictive analytics. These days, regulators are not only concerned with the safety and soundness of the analytics themselves, in terms of how the models are built and whether they are still validating. Regulators are also focused on the impact of the decisions—that is, who gets a particular decision and why. In the US in particular, with the foundation of the Consumer Financial Protection Bureau (CFPB), there’s a growing emphasis on making decisions that are deemed “fair,” and it’s a perspective that I suspect will become increasingly common worldwide. This fairness often translates into making decisions that are consistent, as well as easy to understand and explain.

With model management, the good news is banking institutions can kill two birds—performance and compliance—with one stone. The same model tracking, monitoring and documentation practices required for regulatory compliance also enable institutions to evaluate and refine model performance in ways that control losses and boost portfolio profitability.

I recently discussed this very subject in the webinar Comply and Compete: Model Management Best Practices. I invite you to watch the recorded session, where I share five smart practices with the dual goals of compliance and competitive advantage in mind. 


In European Fraud, Size Isn’t Everything … Or Is It?

European Fraud Map

There's been a great deal of interest in—and even some confusion with—the findings of our latest European fraud map. Some observers have commented that, whilst the card fraud trends may be interesting, the underlying messages are somewhat mixed.

The European card industry has long praised chip and PIN technology as the antidote to historically spiralling card fraud losses in areas such as counterfeit and lost and stolen—and now (finally) the US has followed suit with its own chip card issuance and acceptance plans. And yet, the overall levels of European card fraud, in absolute terms, are back on the rise. This seems like a contradiction.

Dig a little deeper, and it is evident that the robust European chip and PIN defence saw fraud attacks evolve and mutate rather than be eliminated. Just like a snow plough on a snow-laden road, whilst the path immediately ahead is cleared, the displacement merely results in a greater build-up in adjacent areas. Our fraud map commentary references fraud migration to cross-border and into remote (card-not-present) transactions. These are areas where the chip and PIN defences did not extend, and therefore, they became the chosen targets for the criminal fraternity.

One might expect fraud levels to drop as fraud migrates to areas that were historically less attractive or more complicated to target. But the longer an area of exposure is left untreated, the greater the attraction to criminals, and the greater effort and emphasis they put upon it. Combine this with the facts that, as a society, we are now far more inclined to make “remote” purchases (such as on-line or via telephone), to spend with merchants overseas (how many of us realise that e-commerce transactions on eBay settled through Paypal may be routed through Luxembourg?), and to travel far more widely than ever before (international tourism has more than doubled in less than 20 years). Needless to say, fraudsters have a rich seam of opportunity.

Industry reactions to fraud trend changes are often far slower than that of criminals, and the latter have come to realise this. Chip and PIN reached critical mass maturity in the UK in 2005, and yet there are still large parts of the developed world that have yet to embrace the technology, and will not do so until what will eventually be more than 10 years later.

With the rise in European fraud levels again, some observers have questioned whether it is statistically significant. Fraud is frequently reported in terms of hundredths of a percent (basis points), and the highest levels of fraud within the region equate to 7 basis points or 0.07%. That doesn’t sound a lot.

The thing to bear in mind is relativity, as I’ve indicated before on this blog. As those of us familiar with quality and process assurance know, statistical significance from defects are evaluated in manufacturing and service industries to a measure of Six Sigma. Basis points, whilst a fraction of a percent, are still significantly higher than levels of quality or process tolerance.

Of course, if total genuine card spend only equated to $1m, then a 7 basis points loss only equates to $700. But if total spend in the UK, for example, is £520 billion, then the UK’s loss level at just under 6 basis points equates to circa £300 million. That’s a lot of money in anyone’s book!

The moral of the European fraud map, therefore, is that whilst there is much to celebrate in terms of recent anti-fraud advances, there is no room for complacency. Until there is complete global interoperability on card fraud defences, and unless the industry begins to move faster and in concert, fraud will continue to prevail to some degree. And whether card fraud is skewing up or down, fraud loss is something that should concern us all.


Who Wants Automated Fraud Alerts? 91% of People

In light of FICO’s European fraud map, published last week, we know fraud is on the up! Now the debate is over what to do about it.

One reason for the rise in losses is a focus on the customer experience, as FICO’s Martin Warwick has explained. Banks in the UK and some other countries brought fraud levels much lower, then focused on how to make the purchase process much more streamlined. This has allowed fraud losses to rise. Martin Warwick has suggested that banks will now look at how they can bring losses back down, including through automated contacts and verification.

This seems obvious enough, but has prompted some publications (perhaps desperate for a headline) to suggest that consumers will be barraged by “annoying phone checks” to verify transactions are genuine.

We beg to differ. A recent FICO survey found that 91% of participants were happy with auto resolution, and 89% said auto-resolution had increased their confidence in using their card. Targeted and focused customer contact doesn’t have to mean encroachment, it can mean empowerment!

Martin would be the first to note that automated verification isn’t the only way card issuers will be fighting fraud. Continued investment in fraud detection capabilities, and advances in online risk assessment such as FICO’s new behavior-sorted lists, adaptive analytics and merchant profiling will give financial institutions the ability to identify more fraud.

But there will always come a time when customer engagement is necessary to confirm abnormal activity, and to protect accounts from potential abuse. Automated transaction and activity verification is a vital tool in the armory of many financial institutions to reach more customers than would be possible with just human agents.

My colleague Brian Kinch and I have recently been blogging about the necessity of customer interaction during fraud verification, how to make fraud protection a better customer experience, and how to use customer preferences to improve fraud resolution.

Automated communication and case resolution is becoming a necessity for all institutions with fraud verification needs. With the emergence of mobile devices and electronic communication channels, institutions can engage with their customers using multiple media channels, in intelligent automated dialogs. By getting to know customers, and using this knowledge in segmented contact strategies, lenders can create a better experience for them, and ultimately increase the speed and success of communications.

One FICO client has seen that customers aged 30-59 are 7% more likely to resolve fraud alerts by voice than customers aged 29 and under. Interestingly, another FICO client has seen that those aged 30 and under can be twice as likely to respond to an interactive SMS than older customers. It’s all about knowing YOUR customers.

Correctly built and implemented verification tools using an intelligent, demographic-based contact strategy, and targeting earlier activity prior to transaction declines, can ultimately lead to a better customer experience, and more importantly, customer loyalty.

For a look at fraud trends across Europe, check out our new infographic:

European Fraud Map


Is the US Facing the Dawn of a New Debt Crisis?

Man Worried About Debt
By Todd Rollin

A colleague recently forwarded me a link to a recent Urban Institute study, “Delinquent Debt in America,” where a random sample of US credit data revealed that approximately 35% of adults “have debt in collections reported in their credit files,” with an average debt of over $5,000. The study focused on non-mortgage debt, including medical debt, utility bills, membership fees and phone bills.

That 35% figure, however, really stood out. More than 1 in 3 Americans have been or are being collected on (and in some cases, they may not even know it). Keep in mind that the study excluded adult Americans without a credit file (roughly 22 million), many of whom are likely to already have financial issues. So, in reality, the number is ostensibly higher.

Somewhat less surprising were the study results segmenting debt by geographic location. Areas that exceeded the norm are still struggling to recover from the financial crisis – particularly Nevada, which had a staggering 47% with debt in collections. Regions with fewer instances of overdue debt were characterized by factors such as less spending and more job availability.

What does all this mean for the US debt management industry? Well, first, let’s consider that the 35% statistic is actually marginally lower than the 36.5% figure from a Federal Reserve analysis conducted a decade ago. So we can’t point to being in a period of “runaway” bad debt.

Still, the US economy today is quite different than it was during that 2004 study. As immigrants flock to low-wage jobs, manufacturing jobs are outsourced, and many office and manual jobs are lost to automation, wage rates are bound to remain stable or slip with respect to inflation. Add to this the fact that the US has one of the world’s highest corporate tax structures in the developed world, so capital investment is less likely to land in the US and drive employment. If inflation picks up – particularly in food and housing – the gap between income and living expenses could grow, making it more difficult to repay financial obligations. 

The net result is a widening gap between population and employment, driving more people to government assistance programs. This growing segment will be more vulnerable to debt issues, as they may have insufficient savings to draw upon to cover medical bills and other unforeseen financial problems. Subprime and payday lending can be expected to increase, as consumers try to make ends meet.

Ultimately, a growing number of consumers are becoming financially stressed, and the banking industry has taken note. In our latest survey of U.S. and Canadian risk managers, 43% expected delinquencies to rise on all consumer loans. That sentiment is at its highest level since our Q4 2011 survey. Combine this financial pressure on consumers with a newly transformed and demanding regulatory environment, and debt collection operations will soon have their hands full.


European Card Fraud Map Shows Record High Losses in 2013

European Fraud Map

Credit card transactions are generally the best-protected credit channel, with wave upon wave of new technology and methodology used to strengthen protection. So it may come as a shock to learn that card losses in Europe are going up, not down.

That is made clear by FICO’s latest map of card fraud in Europe, which shows card fraud losses in 2013 for 19 European countries reached €1.55 billion, slightly more than the previous peak in 2008.

While the UK had the highest losses, mainly fuelled by card not present (CNP) fraud, France and Greece both had higher ratios of fraud losses to card sales, at 7 basis points (.07 percent). And fraud grew fastest in Russia, jumping nearly 28%.

My colleague Martin Warwick provided the commentary for the map, which is based on data from Euromonitor International. As Martin notes, the rise in losses may serve as a wake-up call, and prompt a new wave of anti-fraud investments. FICO has already seen a greater interest in interactive, automated customer communication services that contact customers in real time when a transaction triggers a fraud alert.

The map provides a fascinating look at trends in countries across Europe. Check it out at


Customer Centricity Success Story: Tackling Fraud

I’ve been blogging about customer centricity success stories, and today, I’ll share the fourth FICO case study in my series, a client which I’ll call Bank D. This global leader in credit cards wanted to demonstrate its commitment to protecting customer information and assets from fraud, and its use of innovative technology for that purpose. It understood that customer centricity is not about becoming so focused on engaging customers that we overwhelm them with attention. Rather it’s about making very careful decisions about why, when and how to make contact.

By adding intelligent communications management to analytics-driven fraud detection, Bank D accomplished this goal, while also dramatically improving fraud detection performance and efficiency. The new automated process, which generates batches of alerts every 15 minutes, replaced a slow, totally manual and very costly outbound dialing process.

The diagram below shows how this approach tightly links fraud risk analysis and customer contact actions—with data driving both. Once the transaction is scored for fraud risk, the communications manager pulls together other data from multiple sources in real time to assess whether or not to contact the customer.


Bank D’s business rules drive that decision. If it’s a “go,” a subsequent decision is then made (based on urgency and the customer’s behavior history and preferences) about how to make contact. In many cases, this contact happens in a self-serve mode: Customers can auto-resolve the situation by simply tapping a button or speaking a word to verify that it’s really them making the transaction.

According to a Bank D survey, customers are impressed. Not only were 76% of respondents highly satisfied with auto-resolution fraud checks, but 89% said they had increased confidence in using their cards again.

It’s quite likely that many of these more confident customers will increase utilization, thus positively impacting Bank D’s revenue and profit. In addition, the system has reduced declines by 32% and point-of-sale referrals by 80%, with respective declines in complaint rates of 27% and 11%. These improvements are also likely to raise utilization, fee income and profit.

The stunning thing about these results, to my mind, is that Bank D achieved them while resolving 250% more fraud cases with zero staff increase. Even more evidence that becoming customer-centric is good for bank financial performance.

To read other customer centricity success stories, I invite you to download our new Insights white paper: “Customer Centricity: Four Bank Success Stories” (No 78; login required). Or read my recent blog posts with other customer-centric case studies (Bank A, Bank B and Bank C).


Better Communication with Overdue Customers Is an Award-Winning Strategy for BNP Paribas Bank Polska and Daimler Financial Services


Collections, a trending topic in our blog, is an invaluable tool for providing an outstanding customer experience. The collections unit is one of a firm’s important contact channels.

For proof that the collections practice is becoming a customer service area, just look at the winners of the 2014 FICO Decision Management Award for Debt Management. Both companies — BNP Paribas Bank Polska, which is a part of the international BNP Paribas Group, and Daimler Financial Services, the financing arm of automotive giant Daimler AG — are using customer-centric communication services to improve consumer collections.

BNP Paribas Bank Polska saw an opportunity in its Polish business unit to improve collections by automating contact with customers, segmenting customers more precisely, and identifying the most effective collection actions to take with each customer. The bank uses FICO® Risk Intervention Manager to contact overdue customers with interactive SMS and automated voice messages in order to expedite resolution of overdue debt.

BNP Paribas Bank Polska increased its collections effectiveness on delinquent accounts from 82 percent in 2011 to 86 percent in 2013, while reducing headcount assigned to collections by 39 percent, thus keeping the customers, the relationships and the profitability.

Daimler Financial Services also used FICO Risk Intervention Manager to meet its goals in the Italian market, increasing collections and cutting costs without eroding customer satisfaction. Daimler Financial Services used the FICO solution to understand when and how customers would be most receptive to collections actions, reducing the overall risk of the portfolio.

The results these companies achieved impressed our panel of judges. As Brian McDonough, research manager in IDC's Business Analytics Solutions research service, said: “BNP Paribas Bank Polska and Daimler Financial Services have proven that tailoring the contact strategy to the customer can pay off in a big way, for multiple industries.”

Winners of the FICO Decision Management Awards will be featured in presentations at FICO World 2014 in San Diego, November 11-14.

This blog also includes discussions of the winners in the Customer Originations, Customer Growth & Retention and Fraud Control categories.


Customer Centricity Success Story: Reducing Compliance Risk

Recently on the blog, I shared stories of two FICO clients (Bank A and Bank B) that are successfully overcoming challenges to advance customer centricity goals. Today I’ll share the story of a third client, which I'll refer to as Bank C.

Bank C, like many others in its markets, was fined by regulators for misconduct in selling fee-based extra services to new accountholders. Moreover, the light shined on this problem made it clear that over-zealous sales activities weren’t the institution’s only vulnerability to fines and reputational damage from regulatory noncompliance.

Bank C executives began asking questions such as: Are automated originations processes confirming that customers understand and accept product and credit terms? Are collections agents making the required disclaimers at the proper time in the conversation, and avoiding using inappropriate language and threats? The answers convinced them the bank needed to put in place policies, systems and processes to provide better visibility into and controls over compliance risk exposure.

As part of this solution, Bank C deployed analytic models to score new sales for compliance risk. The objective was to identify potential high-risk cases for review and follow up with the customer. To improve efficiency, the bank got started by working with FICO to implement an automated scoring service across a number of different product lines, including credit, savings and insurance.

But in such a dynamic market, where regulations, bank products and customer behavior are all changing, how could the bank ensure that the deployed models continued to accurately identify the risk level of these sales? Factors indicative of high risk today might not be significant a few months from now, and vice versa.

The solution was to also implement centralized, automated model management, illustrated below. Regularly scheduled model validation processes will fully document analytic performance, generate alerts when it drops below specified thresholds and even capture actions taken in response to validation findings. An accompanying model development environment will allow the compliance risk models to be quickly refreshed and deployed back into the scoring service. This framework also has the potential to support other conduct risk assessments, such as those used for collections calls.


To read other customer centricity success stories, I invite you to download our new Insights white paper: “Customer Centricity: Four Bank Success Stories” (No 78; login required).


Research: New Analytics That Boost Fraud Detection

On this blog, I regularly share information about the latest advances in fraud analytics. But our clients are not looking for innovation purely for innovation’s sake. An advance is only worth its salt if it measurably improves fraud detection.

Recently, my team tested the performance of two new fraud analytics that I’ve blogged about before—specifically Behavior Sorted Lists and Adaptive Analytics. In this post, I'll quickly recap each innovation and share those latest performance results.

Behavior Sorted List technology identifies cardholder "favorites"—or recurrences—over the transaction streams. These might include favorite ATMs that are close to work or home, favorite gas stations along a daily commute, and preferred stores for internet shopping. Behavior Sorted Lists can distinguish between frequently repeated transactions that indicate normal spending (what we data scientists call “in-pattern” transaction activity) and infrequent activity that is far more likely to be fraudulent (“out-of-pattern” activity). This ability enables faster fraud detection with lower false positive rates—that is, fewer declines on legitimate transactions.

The graphics below compare performance of FICO® Falcon® Fraud Manager 6 with and without Behavior Sorted Lists, using the most recent International Credit Models (ICM) 12. We observed substantial improvements using Behavior Sorted Lists, looking at both account detection rate and real-time value detection rate. I’ve highlighted improvements at a few account false positive ratios.

Zoldi 1

Zoldi 2

Digging deeper, we analyzed performance on a number of fraud types, including cross-border and card-not-present (CNP) fraud transactions. The following two plots show that the ICM 12 model with Behavior Sorted Lists outperforms the base model. It significantly improves transaction detection rates over a range of non-fraud transaction review rates.

Zoldi 3

Zoldi 4

We also evaluated the performance lift of Adaptive Analytics. Adaptive models work in conjunction with neural network fraud models, continually adapting the neural nets based on ever-changing fraud patterns that emerge over time. This not only improves model performance, but also extends the useful lifetime of static neural network models.

This ability to reduce model degradation is especially helpful in emerging international markets where we observe higher market dynamics. That’s why we tested our latest International Credit Models on not only in-time data (that is, the development data), but also on out-of-time data (outside the development data range). This evaluates whether the model will maintain robustness when deployed in a more dynamic production environment.

Performance results are shown in the graphics below. “AA” stands for Adaptive Analytics. The green curve represents the base ICM 12 performance—without Adaptive Analytics—on the in-time data that the model was built on. The red curve shows performance of the same model evaluated on out-of-time data. As you would expect, we see some performance degradation compared to the green curve. The blue curve shows the performance of the adaptive model on the same out-of-time data.

Zoldi 5

Zoldi 6

The adaptive model clearly outperforms the base model on the out-of-time data. Compared to the in-time performance of the base model, the adaptive model’s out-of-time performance is close for account detection rate, and approximately the same level for real-time value detection rate. It’s strong evidence that using Adaptive Analytics is highly effective in preventing model degradation.

Clients of our International Credit Models can take advantage of both Both Behavior Sorted Lists and Adaptive Analytics with the release of v12 later this year. ICM 12 has been trained on a more comprehensive pool of international credit consortium data, and it is capable of identifying a wide spectrum of credit fraud patterns across continents. These improvements will allow our clients to keep up with the latest fraud schemes in their respective regions.


Nationwide Building Society Makes More Credit Available Using FICO-Powered Strategic Risk Infrastructure


One day your bank is at the top of the online pricing table for loans, with the most competitive pricing. The next day, your rival is on top. What do you do?

If you’re the UK’s Nationwide Building Society, the world’s largest building society with over 14 million members, you can whip your way right back to the top of the table. That’s one of the advantages of the Strategic Risk Infrastructure Nationwide built using advanced decision management technology from FICO. The SRI has enabled Nationwide to increase lending by at least £25 million a year.

The project’s results have made Nationwide this year’s winner of the FICO Decision Management Award for Customer Originations. 

The key aim for Nationwide was to make more credit available. Using FICO® Blaze Advisor, Nationwide have replaced all the disparate systems previously used in its mortgage and personal loans operations, enabling the society to treat each customer more consistently.

One important way the SRI helps Nationwide extend more credit is by enabling the approval of qualified applicants who would otherwise be declined for affordability reasons. If a customer applies for a loan that has regular payments they may not be able to afford, FICO Blaze Advisor can review the affordability of the loan spread over a longer term, and if appropriate present that option to the applicant.

This is a good example where asking the right questions is key to growing the number of loans. However, it is also important not to ask too many questions. The SRI has streamlined the sales process, reducing the number of questions asked during a loan application process by focusing on those with the greatest relationship to an applicant’s creditworthiness. This has reduced the consumer abandonment rate during application by approximately 17%.

Another benefit for consumers is the speed with which Nationwide can implement new competitive deals using the SRI.

“Changes in pricing strategy that used to take 10 days to deploy can be made the same day now,” said Mark Tuton, senior risk manager at Nationwide. “We have used this capability to remain at the top of ‘best buy’ credit tables published by third parties. When a competitor changed its prices in an attempt to put itself at the top of the table, we reviewed our strategy and made a change the same day to remain at the top.“

Read more about this award winner in our news release. Plus, check out the awards for Garanti (Fraud) and Westpac (Customer Growth & Retention).