Category Archives: FICO

Do Consumers Seek More Credit After Their Score Recovers?

In a previous post, we noted that the majority of consumers who had a 7-year-old delinquency purged from their credit file saw improvements in their FICO® Scores. Now let’s look at whether these consumers’ credit-seeking behavior changed after the delinquency was purged and their score recovered. Were they more likely to apply for credit? Get approved and open new accounts?

To assess this, we looked at the proportion of the “delinquency purge” population (those that had a delinquency removed from their credit report between May 2016 and July 2016) that had a new inquiry or opened a new account in the three months following the purge window (August through October 2016).

In Figure 1, we compared those values to the same period a year earlier, to avoid capturing seasonal changes in credit habits.

Delinquencies FICO Scores 3 Do Consumers Seek More Credit After Their Score Recovers?

The data showed that there was a minor increase in the percentage of consumers that had a new inquiry compared to a year prior, and a material increase in those that opened a new account. This is particularly noticeable for new bankcards.

For the “full recovery” population (those in the “delinquency purge” population who had no remaining serious delinquencies), the changes were slightly more pronounced, as seen in Figure 2.

Delinquencies FICO Scores 4 Do Consumers Seek More Credit After Their Score Recovers?

We see a greater increase in both new inquiries and account openings for this segment than is observed in the broader “delinquency purge” population. Furthermore, we see higher rates of new account openings despite lower rates of inquiries than the “delinquency purge” population, suggesting a higher approval rate.

This supports the idea that these consumers, who have rebuilt their credit profile and FICO® Score through years of responsible behavior, are now very well-positioned to access new credit at more favorable terms than they would have been offered before (which in turn may induce them to take on new credit).

Is It the Score or the Lending Environment?

But perhaps the increase was due to other factors (such as less restrictive lending standards), and had nothing to do with the removed delinquencies. So we looked at the inquiry and new account activity for the “delinquency baseline” population (those who had a serious delinquency in 2009-2010 but did not have a delinquency removed between May 2016 and July 2016) as a benchmark for comparison.Delinquencies FICO Scores 5 Do Consumers Seek More Credit After Their Score Recovers?

In the “delinquency baseline” population, we found that there was actually a small decline in inquiries between 2015 and 2016, and relatively minor changes in new account openings. The figures look quite different from the increases in new credit activity seen in the “delinquency purge” population.

The difference in the year-over-year change in new inquiry rate between “delinquency purge” consumers and “delinquency baseline” consumers suggests an increase in the willingness of the “delinquency purge” population to seek credit, and increased confidence in their ability to get approved. And the data supports that confidence – the increase in new accounts opened is disproportionately large when compared to the increase in inquiries.

As we move further away from the Great Recession, consumers whose credit situation was adversely impacted but who have since righted the ship via responsible use of credit should continue to see marked improvements in their FICO® Score. Based on what we’ve observed from the “delinquency purge” population, their increased creditworthiness comes with expanded interest in and access to credit: a potential win-win for both lenders and consumers.

Let’s block ads! (Why?)


How Are Telecom Providers Managing Cybersecurity Risk?

Data breaches are in the news on an almost daily basis and telecom companies are not immune to attack – indeed, cyber-attacks have led to the loss of almost 50 million customer records in the past 10 years.

For telecom businesses it’s not only the frequency and number of records that are breached that is of concern, it’s also the type of data that can be lost. Telecom companies are guardians of a rich set of customer data, including financial information. They also hold a valuable set of behavioural data about their customers based on how they use their services. This rich data set is a honey pot for cybercriminals – particularly those using data to fuel identity theft.

We included telecommunications providers in the cybersecurity survey carried out on our behalf by independent research company Ovum. What did we find?

A full assessment of our findings for the Telco industry can be found in our e-book “Cybersecurity for Telecoms –Views from the C Suite.” Here are some headlines:

Telecom companies have seen an increase in attacks

53% said they’d experienced an increase in attempted data breaches. The larger telecom businesses had experienced this more frequently – 71% of those with over 10,000 employees said they’d seen attempted breaches increase in the past year. They also expect the risk to carry on increasing; 81% said they expect attacks to increase in the coming year.

Telecom companies think they are doing very well at fighting cybercrime – but are they?

The majority of telecom industry respondents thought that their organization’s cybersecurity was better than average, when compared to their competitors. In fact 39% rated themselves as top performers, recognized for their cybersecurity efforts. None thought they were below average.

This is statistically unlikely and a lack of objective measurement of cybersecurity might be to blame. 35% self-assess based on their own benchmarks and criteria and 6% don’t carry out a measurable assessment. Objective measurement could well be a starting point for telecom companies that really want to understand and improve their cybersecurity position — taking a free trial of the FICO Enterprise Security Score will help!

There are gaps in the cyber-readiness of telecom companies

When we looked at specific measures that telecom companies could take to combat cybercrime, it became clear that not every telecom company had taken all the steps they could to enhance their cybersecurity posture. Perhaps most shocking was that only 54% of telecom companies had a tested data breach response plan — in Sweden it was only 38%. Telecom companies were doing better in regards to the other specific measures we asked about, but as the graph below shows there are still gaps.

Cyber survey chart 7 How Are Telecom Providers Managing Cybersecurity Risk?

Next year the General Data Protection Regulation (GDPR) comes into force in Europe. This legislation means that the repercussions of a data breach will become more severe and include a substantial increase in the fines that can be imposed. Telecom companies cannot afford to relax their cyber-readiness efforts.

Download the e-book: “Cybersecurity for Telecoms –Views from the C Suite”

Let’s block ads! (Why?)


The Future of AI: Meet the Multi-Tasking Machines

AI 25 Years FICO Machine Learning The Future of AI: Meet the Multi Tasking Machines

To commemorate the silver jubilee of FICO’s use of artificial intelligence and machine learning, we asked FICO employees a question: What does the future of AI look like? The post below is one of the thought-provoking responses, from Chahm An, a lead analytic scientist at FICO, working in San Diego.

Artificial intelligence, like human intelligence, uses observations from the past to learn a general model that can be used to make predictions about future similar occurrences. The future I see for AI is based on current work being done in this field, and grounded in what I saw 20 years ago, when I first began to study AI.

What’s Really Accelerating AI?

In 1997, and IBM’s Deep Blue had just defeated reigning world champion Garry Kasparov at the game of chess in what was seen as a landmark event of artificial intelligence surpassing a human champion at an intellectual challenge. Automated speech recognition systems were beginning to replace touchtone menus in commercial applications. The human genome project was nearing completion, and dot-coms were popping up like weeds.

Conventional wisdom on the state of AI at the time was that while some tasks that appeared to be easy for humans, such as driving a car, were extremely difficult for computers to learn, other tasks that were difficult for humans such as playing chess at a grandmaster level could easily be achieved by brute force branching computation.

These brute force optimizations were not feasible for problems with many more possible outcomes, as found in more complex games such as Go, or other tasks such as computer vision on natural language processing. Speech applications were thus limited to menu interfaces that limited choices and optical character recognition was similarly limited in scope.

The key development that most would attribute to overcoming this common challenge is the development of neural network technology that has allowed us to train models that help machines make complex decisions at a higher, more abstract level similar to the way the brain functions. However, this isn’t the complete story – after all, as we know neural networks have been in use by FICO for 25 years, so they can hardly be considered a new development.

More accurately, what has changed is that the cost to train neural networks has decreased dramatically, allowing the development of complex networks to be feasible and further accelerating the development of neural network technologies, making them more efficient and accurate. Geoffrey Hinton, a leader in the field of machine learning, has humorously noted that the training of deep belief networks had become feasible due to a 100,000-fold speed increase in training that could be partly attributed to new, more efficient algorithms, but mostly due to the fact that computers had become 1,000 times faster in the 15 years of stagnation since developing the building blocks.

Although Moore’s law is appearing to slow down, I would predict that as more focus gets put toward developing hardware specialized for machine learning, and if research of more efficient training algorithms continues on its trend, the cost to perform machine learning tasks in 2027 should be roughly 0.1% what it is today. This means that models that would take years to train with today’s technology would take less than a day, and learning tasks that currently require resource only available to supercomputers will be feasible on everyday mobile consumer devices.

How will this ability manifest in our lives? Here are three predictions.

Prediction 1: Dynamic Learners – Everywhere, All the Time

Despite the hype and fanfare surrounding deep learning, most of these advanced neural network architectures are stuck in a box. The latest deep convolutional nets can correctly identify obscure breeds of dogs better than the average human, but they’re highly optimized towards strictly defined snapshots as inputs and are limited to a pre-defined set of classifications.

With greater availability of computational resources and data, I believe that the trend will move from deep architectures with a single snapshot input and single classification output to much more complex, deep recurrent networks that take in multiple streams of varying input and offer a multitude of varying output types. Instead of static image classification, a computer vision system may work with two continuous streams of binocular video similar to what we process as humans. Not only will it identify that a dog is a beagle, but also that the dog is taking a walk, and that the lady who’s taking it on a walk looks kind of like Meryl Streep.

Since we’ve got supercomputers full of detectors in our pockets, this continuously streaming process will also be learning on the job as well. While learning tends to be a large, computationally expensive batch job performed on a server in today’s world, this is more likely to be shifted to end devices to some extent, both to distribute costs and to make each device adapted to its unique environment. My phone might ask me if I was interested in the price of dog leashes, or whether I think Meryl Streep lady looks cute, and continuously build a profile on my preferences.

Prediction 2: Generalized Learners – Not Just One-Trick Ponies

I believe that there will be a significant trend towards intelligent systems that do well in more than a single task. The same neural network that is designed to filter out spam emails may also reuse its knowledge to detect phishing attempts, prioritize your inbox by importance, and help you draft responses to common requests. Again, this follows the trend of multiple inputs, multiple outputs, but the result will be that we have more robust systems that are capable of understanding abstract concepts the way that we do.

To achieve this robustness, we will probably see multiple types of learners interacting with each other – perhaps a self-contained Bayesian network may process text in an unsupervised fashion to provide feedback to both a recurrent neural network and a random forest classifier to form a consensus opinion, all within a reinforcement learning system. With a glut of unlabeled data to work with, we are also more likely to see more unsupervised learning of generative models that are able to understand the underlying distribution of variables of interest, rather than the discriminative models that are currently popular with supervised learning models.

The result of these generalized learners will be intelligent systems that are not just optimized to some pre-defined objective, but actually have some degree of expertise in an area of knowledge. Instead of getting a binary decision maker, we may get systems that can explain their judgement in terms that are easy to understand and adapt to more customized objectives that we can have greater confidence in.

Prediction 3: Mixed Reception, But Gradual Trust

As seen with the issues surrounding self-driving cars, there is likely to be a great deal of resistance to certain advances in AI systems. A computer that can defeat Lee Sedol at a game of Go seems innocuous enough, but are we ready to trust an artificial intelligence with deadly force? Do we trust smart devices and their manufacturers to listen in to every occurrence in our daily lives? Are we afraid that computers will become better at our jobs than we are and render us unemployed?

I believe that this apprehension will be the biggest challenge to the advancement of AI in the next decade. There will likely be legislation put in place to further increase the privacy, job security and safety of the consumers who would benefit most from future advances in AI.

However, progress appears to be inevitable, as we already count on technology for so much nowadays. Who uses a physical map to navigate anymore, for example?

Perhaps artificial intelligence systems will need to learn to become experts at PR and marketing before moving on to the next stage of adoption.

Making Progress Today

As I was writing this blog post, I realized that FICO already does most of the things that I have predicted to become mainstream for artificial intelligence. We are already great at learning continuous customer profiles over time, auto adaptation of models, and providing reasons along with scores. This speaks to the vision that put FICO ahead of the game 25 years ago, but will continue to keep us ahead in the future of AI.

See other FICO posts on artificial intelligence.

Let’s block ads! (Why?)


US Average FICO Score Hits 700: A Milestone for Consumers

FICO regularly tracks the national FICO Score distribution as an important gauge of US consumer credit behavior. When I last blogged about this topic based on data from April 2016, the key takeaway was “the beat goes on.” US consumers continued to show improvement in managing their debts, which began shortly after the bottoming out of the economy in 2009-2010.

We have pulled the latest FICO Score distribution information based on a snapshot of millions of US consumers’ credit data as of April 2017, and can report that consumer credit health and responsibility continue to be strong! For the first time since we’ve been tracking these stats, the average national FICO Score reached the 700 threshold — some 10 points above what it was just prior to the recession in October 2006.

Average FICO Score 700 1 1 US Average FICO Score Hits 700: A Milestone for Consumers

Let’s dig a little deeper by examining the score distribution over the past 10+ years. The movement in the “tails” of the FICO Score distribution — the lowest and highest score ranges — can often be most telling.

  • The percentage of consumers scoring in the lowest score ranges — populated most frequently by consumers with high debt to credit limit ratios and numerous recent and significantly past due payments — continues to drop.
  • By contrast, the percentage of consumers scoring in the super-prime range (800 and up) has steadily increased since 2010.
  • In fact, we reached another milestone with the April 2017 data: the number of US consumers scoring 800+ outnumbers those consumers scoring 600 or below.
    • Much of this upward momentum in FICO Score distribution can be attributed to the period of stable growth that the US economy has experienced since the Great Recession.
    • Another important driver has been increased consumer awareness around FICO Scores and credit education, through programs such as FICO Score Open Access, which launched in 2013 and now provides 200 million consumer accounts with free access to FICO Scores. 

Average FICO Score 700 2a US Average FICO Score Hits 700: A Milestone for Consumers

There continues to be a steady decrease in the percentage of consumers with recent serious delinquencies (defined as 90+ days past due) — from 19.4% of the population in October 2013 to 16.5% in April of this year. This includes a 3.5% year-over-year decrease in the recent serious delinquency rate between April 2016 and 2017 (17.1% vs 16.5%).

Since payment history comprises roughly 35% of the overall FICO Score calculation, this sustained reduction in recent delinquency is clearly a key driver of the ongoing upswing in the FICO Score distribution.

Average FICO Score 700 2 US Average FICO Score Hits 700: A Milestone for Consumers

The upward trend in FICO Score is also present across different age groups.  We found that over the last year, 29% of “Generation X” consumers (those aged 38-52) experienced an increase in FICO Score of 20 or more points, whereas 21% of the Gen X population saw their FICO Score decrease by 20 or more points.  This finding is supported by our recent consumer research that shows improved financial behavior by Gen Xers, even though many carry larger debts than other generations.

The most pronounced decrease in serious delinquencies is in the real estate segment. The latest data shows real estate delinquency continuing a marked decline, while auto loan delinquency rates have steadily inched upwards over the past four years.

Bankcard delinquency rates are also of note: in April 2017 the delinquency rate increased for the second straight period, after having been virtually level for the 2 years prior. Stay tuned for future blog posts, as we dig into the question of whether we are at an inflection point for bankcard delinquencies.

Average FICO Score 700 3 US Average FICO Score Hits 700: A Milestone for Consumers

Figure 5 provides some additional granularity into consumer-level trends in key credit behaviors that are factored into the FICO Score. As already mentioned, current US consumer repayment behavior on the aggregate is strong — this can be seen in the continued decrease in the percentage of US consumers who have indications on their credit file of recent delinquency or other associated negative items such as collection agency accounts.

 Average FICO Score 700 4 US Average FICO Score Hits 700: A Milestone for Consumers

In addition to showing improving repayment behavior on the aggregate, the April 2017 data also provides some evidence that consumers are starting to rein in their credit-seeking behavior. Credit-seeking behavior is factored into the FICO Score calculation, with greater incidence of credit-seeking consistently found to be an indicator of elevated repayment risk.

  • The percentage of the population with one or more “hard” inquiries (those posted to the file as a result of a consumer-initiated search for credit) hit a three-year low of 43.2%.
  • The percentage of consumers with a new credit account opened in the past year dropped to 46.5% as of April 2017, down from a high of 47.6% a year prior.

Average FICO Score 700 5 US Average FICO Score Hits 700: A Milestone for Consumers


Where Do We Go From Here?

The general trend in US consumer credit behavior is positive. We are seeing more consumers scoring at higher levels than at any time in the past decade, driven largely by a sustained reduction in US consumers’ overall delinquency rate.

That said, the uptick in the delinquency rate for the auto and bankcard segments merits further investigation. We will be tracking these measures closely over the coming quarters. Is the increased bankcard delinquency rate a sign of a tipping point in the recovery in consumer credit that we have observed over the past 7 years? A reflection of lenders’ willingness to underwrite more risky borrowers in an effort to grow volumes? Or just a momentary blip in the extended credit recovery that our country has experienced since the Great Recession?

The fact that the national average FICO Score has reached 700 speaks to how US consumers are, on average, managing their credit behaviors. But when considering where delinquency rates might head in the future, it is important to keep in mind the numerous factors that are exogenous to the credit report (e.g., macroeconomic indicators such as unemployment levels or interest rates), which play a crucial role in consumers’ ability to pay their credit obligations.

Even when US credit quality is healthy, prudent lending practices dictate that all of the “four Cs” of credit be considered: credit behaviors, capacity, collateral and conditions (including economic factors). Following these practices can lead to sustainable levels of credit being granted at fair terms — a win-win for lenders and consumers alike.

Let’s block ads! (Why?)


Analytics Achievers Wanted for FICO Decisions Awards

FICO Decisions Awards 2017 Analytics Achievers Wanted for FICO Decisions Awards

A company that solved an “unsolvable” wind farm optimization problem. An automaker that uses analytics to keep more delinquent customers in their cars. An airline that uses optimization throughout its business to improve customer service and efficiency.

These are just some of the past winners of the FICO Decisions Awards – and now it’s time to find the next set of innovators. We’re looking for companies that are achieving outstanding success using FICO analytics and decision management solutions.

Awards will be presented in six categories:

  • Artificial Intelligence and Machine Learning
  • Customer Onboarding and Management
  • Debt Management
  • Decision Management Innovation
  • Fraud Control
  • Regulatory Compliance.

Winners will receive recognition at FICO World, which will be held April 16-19, 2018, in Miami. Winning implementations will be featured in conference activities, and two representatives of each winning company will receive complimentary conference passes.

Nominations are due September 1, and winners will be announced in early October.

As in past years, we’ve assembled a panel of independent judges with deep industry expertise. This year’s judges include:

  • Giorgi Alibegashvili – Strategic Project Manager, TBC Bank (2016 award winner)
  • Julie Conroy, Research Director, Aite Group
  • Doug Gray, Director, Enterprise Data & Analytics Technology, Southwest Airlines (2016 award winner)
  • Joy Macknight – Transaction Banking and Technology Editor, The Banker
  • Daniel Mayo, Chief Analyst, Financial Services Technology, Ovum

The FICO Decisions Awards can really illuminate your success. After Toyota Financial Services won in 2015, they were written up in several publications, including this article in CIO World, and then were named to the InformationWeek Elite 100.

For more information and to enter a nomination, visit

Let’s block ads! (Why?)


The Story Behind Europe’s €1.8 Billion Card Fraud Problem

FICO has just released an interactive infographic on European card fraud trends since 2006, showing that card fraud in the 19 countries studied has hit a new high of €1.8 billion. In the UK, card fraud also a hit a new high in 2016, £618 million, though the rise was less than the rise from 2014 to 2015.

To understand what these numbers mean — and what issuers need to do now — it’s important to know how got here. That’s why our map tracks the data back to 2006.

Why is 2006 so important? That was the year of ‘love my PIN day on Valentine’s day’ in the UK, as the liability shift between magnetic stripe and chip & PIN technology had been put in place during 2005. So chip & PIN was officially rolled out across Europe and executives expected to see the benefits of that investment in technology at point of sale.

What we actually saw was a migration in fraud attacks away from face-to-face transactions to card-not-present (CNP) and cross-border fraud, where magnetic stripe technology was still accepted and carried no liability shift. In the US, this liability shift didn’t take place in 2015, 10 years after Europe, and criminals used this as an opportunity to skim cards and use counterfeit ones in the US.

UK fraud would continue to rise to a peak of £610 million in 2008, when banks introduced a tough-on-fraud stance to get this back under control. Fraud detection dials were turned up using real-time analytic detection, principally FICO® Falcon® Fraud Manager and its AI-based analytics. Over the next year, £170 million of fraud was prevented and this trend would continue to the bottom of the trend trough in 2011.

European Card Fraud UK FICO The Story Behind Europe’s €1.8 Billion Card Fraud Problem

CNP Fraud Takes Over

CNP fraud had reached £328 million in 2008 and represented 53% of the total fraud losses in the UK. To put this CNP trend into context, 10 years earlier in 1998 it was £13.6 million and just 10% of the fraud losses.  By 2011 banks using real-time detection and advanced analytics banks had reduced this to £220 million, but there was an underlying trend in genuine spending that could not be ignored.

Digital e-commerce spending was £45 billion in 2008 but by 2016 it had reached £248 billion. That trend was the perfect place for criminals to focus their attention. By 2013, UK CNP losses had just breached the previous high of 2008 at £331 million by £3 million, which means that some clever ‘real-time’ detection and prevention analytics capability had kept the lid on the fraud attacks.

Our latest map shows that CNP fraud reached £432 million in 2016 in the UK, and digital e-commerce is estimated to be £309 million of that figure. Digital now represents 50% of the total fraud losses at £618 million and confirms the challenge that UK banks need to get to grips with.

AI Breakthroughs for Fighting Fraud

Many European countries, including the UK, are seeing CNP fraud at 70% plus of their losses. FICO and these banks are focused on using new advances in machine learning to cut into these losses, but allowing consumers to spend where and when they want. No small challenge but it’s doable!

Introducing more behaviour analytics such as FICO’s patented behavior-sorted lists and customer behavior archetypes has enabled FICO to separate more subtle changes in genuine behaviours from fraud attacks trying to fly under banks’ analytic radar.

There is no silver bullet, which is why we’re working on multiple fronts, introducing:

  • Better CNP analytic models that focus on digital e-commerce
  • Analytics that rate a transaction’s risk in part based on identifying the device and IP address as the consumer’s norm
  • Mobile analytics that check a consumer’s mobile behaviors
  • Archetype analytics that tap into the rich source of mobile context such as advanced geolocation, feeding that information into real-time fraud detection on Falcon

You can learn more about how AI and machine learning technologies are advancing fraud detection in our white paper, Using Advanced Analytics to Stop Fraud. Also, check out the blog posts by Scott Zoldi on the FICO Blog and Twitter @ScottZoldi.

Let’s block ads! (Why?)


Cybersecurity Insurance – 3 Reasons Businesses Aren’t Buying

We recently commissioned a study from independent research company Ovum on how organizations are tackling cybersecurity and what they plan to do next. Losses because of a data breach or other cyberattack can be severe, particularly when factors such as customer and shareholder confidence are taken into account. We therefore expected that cyber risk insurance would be an increasingly important way in which organizations are mitigating their risk.

The results were far from uniform:

  • The UK was the most insured country we surveyed, with 69% of respondents holding some kind of insurance, and the USA was the least insured – only 51% of US respondents had any kind of cyber risk insurance.
  • Across the industries surveyed, financial services firms were the most likely to be insured (71%), and healthcare the least likely (26 percent).
  • Even when businesses have invested in cyber risk insurance, it’s unlikely to cover them for all likely risks.

Cyber survey chart 5 Cybersecurity Insurance – 3 Reasons Businesses Aren’t Buying

We dug a little deeper into the attitudes of our respondents to try to uncover why under insurance might occur. Three explanation emerged – each is playing a part:

    1. They have limited investment in cybersecurity. 60% of those interviewed have seen an increase in attacks in the past year and 62% expect the overall level of threat from cyber-attacks and data breaches to increase in the coming year. Many respondents are also facing more consequences should they lose customer data, with legislation such as General Data Protection Regulation (GDPR) massively increases the fines that can be imposed. Even so, less than half (48%) expect spending on cybersecurity to increase in the coming year. While it is encouraging to see 23% are looking to invest in cyber-risk insurance, the pressure on finances may mean that they actually can’t afford to do this – or they can only take out insurance to cover the most obvious threats.
    2. They think it won’t happen to them. We asked respondents how cyber-ready they thought their business was compared to their competitors. 60% think they are above average or top performers, while only 6% think they are below average – this is statistically unlikely. With an unrealistic view on how well they are doing, it’s probable that they don’t appreciate their true risk and therefore don’t see the need for comprehensive insurance cover. It seems that many don’t have the ability to make objective judgements about their cybersecurity risk. This becomes evident when we look at how they benchmark their cybersecurity status; 38% use their own benchmarks and criteria and 6% don’t carry out measurable assessments.
    3. They are unclear on how premiums are set. Businesses that invest in cybersecurity want to understand what they are paying for and the value it delivers. For cyber risk insurance, this means not only understanding what the policy covers but also having confidence that the premiums charged accurately reflect risk. Only 23% believe that pricing from insurance companies is clear and transparent. 23% believe the insurance assessment for their business isn’t accurate, 19% say their premiums are based just on industry averages and 5% don’t understand how their business is assessed for cyber risk insurance.

Risk Measure Is Key to Cybersecurity Insurance

Ultimately, the part cyber risk insurance can play is dependent on a measurement of risk that both the insurer and insured can agree on. In this way businesses, are less likely to over-estimate their cyber-readiness and can build a trusted relationship with insurers based on a common understanding of the cover they need.

We have developed the FICO Enterprise Security Score to help businesses objectively assess their own cybersecurity status, as well as that of third parties. FICO Enterprise Security Score accesses billions of external data points at internet scale, and compares the subject’s cybersecurity posture to the pre-breach status of known attacks. Applying our analytics to this data gives an empirically derived score, so that:

  • Businesses have an objective measure of their cybersecurity status.
  • Insurers can score organizations to determine risk and set fair and competitive premiums.
  • Insurers can understand the risk across their customer portfolio.

The transparency offered by a score like this can help businesses make a more well-informed decision about whether to take out cyber risk insurance — and make sure they’re getting the best deal.

You can see more results of our survey with Ovum on our cybersecurity survey page, and learn about new principles for cyber risk ratings.

Let’s block ads! (Why?)


AI Spotlight: FICO’s Machine Learning Facilitates AML

AI AML AI Spotlight: FICO’s Machine Learning Facilitates AML

This is a guest post from Nikola Marcich with the Policy team at the Software & Information Industry Association (SIIA), the principal trade association for the software and digital content industry.

Walking into Bernie Madoff’s home in 2005, you would not have found piles of money under a mattress, behind a sofa or in his garage. At the time, Madoff had been running an elaborate Ponzi scheme through the wealth management arm of his business that reached $ 65 million by the time of his arrest in 2008, deliberately hiding the money intricately within the financial system.

Serving as Madoff’s primary bank for over two decades, JP Morgan was one of the culprits of Madoff’s fraudulent actions and money-laundering tactics. In their innocent incompetence to identify clear red flags about Madoff’s returns and file a Suspicious Activity Report (SAR), JP Morgan’s was fined $ 1.7 billion in 2014. JP Morgan’s fine highlights the broader problem that many global banks had been facing, which was ignoring the warning signings of fraud and money laundering. Increasingly in today’s age, terrorist organizations and dangerous criminals finance their operations by laundering money in global financial institutions, presenting a huge public policy problem for regulators and policymakers.

In our artificial intelligence (AI) spotlight this week, we highlight FICO’s AML Threat Score tool, which uses AI to help financial compliance analysts detect money laundering or terrorist financing activities. This tool demonstrates AI’s transformative benefits in anti-money laundering (AML) and fraud detection. In doing so, FICO’s machine learning tool also facilitates stronger criminal justice enforcement and enhances national security by identifying the financing activities of terrorist groups and dangerous criminals. Moreover, the tool not only required human input and knowledge for its development, but also requires human interpretation to determine whether the problem identified by the tool presents a case for money laundering and need for intervention.

Innovating AML tools has increasingly become a priority for banks and financial institutions. Regulations to detect and report suspicious activity through SARs have become more strictly enforced. Additionally, with the rise of enormous piles of data, it is very difficult for analysts to sift through the abundance of information. As more cases become flagged for suspicious activity, so too do the number of false positives within the outputted data.

Another problem is that, in most instances, money laundering cases deal with multiple interactions or accounts while traditional AML tools flag individual cases, making it incredibly cumbersome for compliance analysts to connect individual interactions or accounts to broader money-laundering threats. As a result, financial institutions have sought to create more productive mechanisms to help compliance employees sift through enormous piles of data and more efficiently report suspicious activity to regulators.

Specifically, financial institutions have turned to tools like FICO’s AML Threat Score, which incorporates machine learning to generate its AML tool. Machine learning is an application of AI that gives machines access to data so that the machine, or tool in this case, can learn for itself. As FICO’s Scott Zoldi highlights in a blog about AML and machine learning, “[FICO’s] AML Threat Score prioritizes investigation queues for SARs, leveraging behavioral analytics capabilities from Falcon Fraud Manager. It uses transaction profiling technology, customer behavior sorted lists (BList), and self-calibrating models that adapt to changing dynamics in the banking environment.”

With a threat score ranging from 1-999, compliance analysts are able to identify customers whose transactions have a high likelihood of suspicious activity quicker and more accurately. Then, analysts can send in SAR reports to regulators to ensure that they don’t run into the same problems JP Morgan faced in 2014. By sending in more accurate and timely reports, financial institutions don’t just avoid fines, but also help law enforcement identify criminals or terrorist-financing linked with specific bank accounts or transactions.

Just as we highlighted in our previous spotlights on AI, the AML Threat Score does not displace human work, but rather functions as a result of human input and requires human expertise to interpret whether or not the problem flagged requires an SAR or more investigation. Without human expertise to verify and decipher real money laundering threats, the tool would generate more false positives, feed these back into the system and use this faulty data to regenerate more false positives; this creates an inefficient and unsustainable false-positive feedback loop. In this sense, machine learning tools like FICO’s AML Threat Score not only contribute to a greater social well-being by facilitating more accurate and efficient AML tools, but also help to supplement human work and expertise.

In SIIA’s Issue Brief on Artificial Intelligence and The Future of Work, we emphasize how AI is a natural outgrowth of the developments in computer technology like changes in data size, memory and processing speeds. Thus, these developments will lead to innovations not just in niche markets like social media monitoring, but they will also have the ability to drive innovation in education, health care, transportation, speech recognition, and many other markets. Additionally, though some low-skilled jobs characterized by manual tasks may be replaced by AI, more job opportunities in patient care, construction, and high-skilled technical work are all also natural outgrowths of innovations in AI. The transformative benefits brought by AI in many aspects of daily life will continue to become more apparent and universal as we move towards a future defined by technological innovation in AI.

This post originally appeared on the SIIA blog. For more information, see our white paper on Advancing AML Compliance with Artificial Intelligence.

Let’s block ads! (Why?)


“We Can Change Our Strategies in 2 Days” – African Bank

African Bank “We Can Change Our Strategies in 2 Days” – African Bank

African Bank, a large retail bank in South Africa, recently went through a large-scale restructuring in order to bring more efficiency, transparency and collaboration to the way it made decisions. Working with FICO, the bank applied a standards-based decision management methodology to fully modernize its decision system. Now that the new solution has been implemented, we spoke with Dawid Van Zyl, Program Executive of Credit Decisioning at African Bank, to learn more.

Q: What challenges was African Bank facing with its existing credit decision process?

Dawid: Our credit decision lifecycle was fragmented over different applications and teams. This was causing inefficiencies and missed opportunities for our executives to react to the market. We knew we needed to do a complete overhaul of the existing process in order to be effective and profitable.

Q: How did you and the FICO team create a solution for this challenge?

Dawid: The first step was an extensive internal review of bank operations. We had the entire bank look at all existing logic going into our decision-making process, which was fragmented over different applications and teams. We challenged and rationalized existing frameworks until we came up with an optimal plan.

Q: How did technology help transform your decision management process?

Dawid: We were already using FICO Blaze Advisor for rules management. Through our review, we found that Blaze Advisor could make strategy changes within two days, but the ecosystem around it took as long as two months. We looked at what was happening and created a proposal to completely overhaul the way we make decisions. We decided to extend Blaze Advisor capabilities at African Bank and use it as our hub for all decision making.

Q: How did FICO help you create a solution?

Dawid: FICO introduced a new standards-based decision-making methodology to us called Decision Implementation Accelerator. We refer to it as DIA3. The FICO consultants worked with us to help us understand how to approach certain decision scenarios, as well as get strategies configured and defined in Blaze Advisor. The methodology was integral to the success of this project. As we became more familiar with the approach, each iteration improved and this was key to the reduced total time taken.

Q: You mention a reduction in time to implementation. What specific benefits have you realized?

Dawid: After this overhaul, we now have a very complex but elegant solution in place that enables collaboration and complete transparency. We’ve been able to develop and implement new strategies 30% faster than expected, and we have reduced costs by 25%. Now we have a decision system that can make changes in two days, not two months. That makes a big difference to our efficiency and profitability.

We’d like to thank Dawid for taking the time to speak with us. If you’d like to learn more about how African Bank is using FICO solutions to manage credit decisions, read the case study.

Let’s block ads! (Why?)


Cybersecurity – Is a Lack of Good Benchmarks Misleading Execs?

A recent survey carried out on our behalf by research company Ovum found that 48% of organizations think that in a year’s time, an assessment of their cybersecurity will show improvement. Only 3% think that their cybersecurity position will have worsened.

Most companies also think they are doing well compared to their competitors; a mere 6% think their cybersecurity is below average and 54% think they’re above average or a top performer. This is statistically unlikely.

It seems that many are taking an optimistic view of their cybersecurity. Could a lack of objective measurement be to blame?

Here’s what our survey participants said:

Cyber survey chart 4 Cybersecurity – Is a Lack of Good Benchmarks Misleading Execs?
It is encouraging to see that 94% carry out some form of assessment. However, consideration must be given as to how objective the methods used are.

38% of respondents say they self-assess based on their own criteria, this is the largest and possibly the most worrying category. When organizations decide for themselves what criteria to use, it is easy to lose objectivity.

28% use a third party service to assess. These external experts can provide valuable insight and advice; they are likely to be more objective than self-assessment.

However, not all third-party assessments are based on the same criteria. There is little standardization in how a result is produced, and so quality may be variable. Assessments by a third party are based on interpretation of what is needed by an assessor — this human factor introduces error and reduces objectivity.

28% use a software solution to assess their cybersecurity status. On the face of it this is the most objective way to measure an organization’s cybersecurity posture.

But all software solutions are the not the same. Some are based on a human interpretation of what good cybersecurity looks like, examining different factors and then adding or subtracting points to derive a score. True objectivity and an accurate reflection of risk can only come from an empirically derived score, such as the FICO Enterprise Security Score.

My colleague Doug Clare has written a blog looking at the six principles for cyber-risk scores. This is vital reading for any organization that is considering investing in a software solution to measure their risk.

 Cyber Risk Measurement Matters

It is increasingly important to benchmark cybersecurity performance. Accurate and easy-to-understand assessments of cybersecurity posture helps communicate to the business the priorities and importance of making the right decisions on what steps to take; it also helps show when improvement has happened.

It is also vital to demonstrate cybersecurity status to third parties. Many organizations look at the cybersecurity of their supply chain during vendor selection, while insurance underwriters need to accurately assess risk to determine premiums. Using a stable, objective and accepted method to assess cybersecurity posture can provide both a cost saving and a competitive advantage.

To learn more about our cybersecurity research, Ovum have produced a whitepaper: ‘What the C-Suite Needs to Know About Cyber-Readiness’ and there are also four e-books looking at the results for the countries we surveyed. They can all be downloaded from our cybersecurity survey page.

Let’s block ads! (Why?)