Author Archives: Karan Chadha

About Karan Chadha

BI & Analytics Professional, a tech enthusiast, book and sports lover, admirer of finer things that life has to offer!

How fraud drives the motor insurance premiums skywards?



I was a keen student of economics at my Business School. I had learnt that when competition in the market increases, prices of commodities fall. Well, this law hasn’t held true in the UK motor insurance industry. It is a fiercely competitive industry, and the premiums are one of the highest in the world.

Fraud has been a major contributor to this anomaly. The cost of fraud for UK motor insurers is more than £1bn annually. As insurers pass on the costs to customers, fraud adds £150 on average to a motor insurance policy in the UK.

Motor Insurance fraud in UK is highly organized. The ‘crash-for-cash’ gangs orchestrate collision and raise claims with increasing sophistication. A collision can take the form of either a staged accident where two vehicles, both owned by fraudsters, are intentionally crashed. Or it can be an induced accident where the fraudsters deliberately slam the breaks making unsuspecting victims crash from behind. Fraudsters then file for damages with a hope of benefiting from personal injury or loss of earnings claims.

Whiplash (a term used to describe an injury to the neck) claims are particularly common as it is hard to disprove whiplash medically. Since the pay-out for whiplash can be relatively small (less than £2000), most insurers decide to pay rather than contest in the court.

This has resulted in whiplash accounting for 80 percent of the personal injury claims in the UK, whereas it does for just 3% in France. We know for sure the British don’t have weaker necks than Europeans. This disproportionate number only reflects how the motor insurance fraud has thrived more in UK compared to Europe.

Fixing the menace

Tackling fraud requires a multi-pronged strategy. First, insures should pursue suspicious claims in court more frequently. Due to high legal costs and low conviction rate most insurers prefer to pay rather than contest especially for low value claims. This needs to change, as contesting should be considered an investment towards nipping the problem in the bud. Contesting can also lead to busting of broader supply chain involving colluding doctors, lawyers, garages and planted eye-witnesses.

Second, insurers should leverage technology and invest towards developing counter-fraud systems. Telematics data from black-boxes inside cars is already helping insurers with vital evidence to repudiate fraudulent claims. Big data technologies like Hadoop can combine structured and unstructured data to identify fraudulent patterns, such as flagging that both parties involved in the accident are actually connected in real life.

Third, there needs to be a robust regulatory framework which supports prevention and detection of fraud. Measures like banning referral fees between insurers and claims management companies for injury claims was a step in the right direction. Other initiatives like offering treatments rather than cash for minor injuries, improved data sharing across insurers and aggregators, and clamping down on nuisance calls asking ‘‘whether you were involved in an accident which was not your fault’’ will play their part in fixing the widespread problem.

So the next time the vehicle ahead of you suddenly stops for no evident reason and the driver (fraudster) looks unusually calm, there is a high chance that you are the next unlucky victim of insurance fraud!

Also cross posted here



Will banking be disrupted by the peer to peer phenomenon?

Savers often complain of the low interest rates they earn. This should ideally translate into happy borrowers, who should be able to borrow cheap. But this logic, although mathematically true, doesn’t work since the intermediary (bank) takes a cut from both ends to fund its high operational costs.

Peer-to-peer (P2P) lending platforms have surfaced to exploit this inefficiency. Much like traditional banks they connect borrowers to lenders, but through online platforms. A lender can either pick a borrower via an auction or choose to spread the funds across a portfolio of borrowers.

Being digitally operated means that P2P lending platforms don’t need to maintain the branch infrastructure which is one of the largest cost components for retail banks. Advantage is passed to both lenders and borrowers. Operational leanness of their business model further reduces costs and benefits the customers.

The phenomenon is gathering momentum. In Britain, P2P loan volumes are quadrupling every year. Zopa, a British P2P lender and one of the pioneers of the concept, has lent more than £700 million in P2P loans. Lending Club, its American peer recently came up with an $870 million stock offering that valued the company at $5.4 billion. Most of the prominent P2P lenders have managed to have stalwarts of the financial services industry on their boards, extending further credibility to their businesses.

Not the same as Payday lenders

A comparison of P2P lenders with Payday lenders like Wonga is noteworthy. Fundamentally, they exist to serve different needs. Payday lenders typically provide instant funds for very short durations (ranging from days to weeks). And, loans can be extended to customers even with poor credit history. Short time duration and high default risk means that interest rates can be exorbitant (APR of 1500% is not uncommon). This has earned Payday lenders the fury of many, including the church.

P2P lenders on the other hand offer funds for longer durations and perform credit checks on the borrowers, much like traditional lenders. A two year loan can be availed at an APR of 5%, which is better than the rates charged by banks on personal loans. Lender can expect to earn 4%, which is again better than what banks offer on deposits in developed countries.

Extends beyond lending

The P2P phenomenon isn’t merely limited to lending. P2P money transfer is also catching traction. TransferWise, a British start-up is a case in point. Instead of moving cash cross-border in the traditional way, it connects users across countries whose funds can be deposited into each other’s accounts in the local currency.

Say Mr A wants to send funds from the U.S. to the U.K. and Mr B wants to send funds in the reverse direction. Instead of following the traditional money transfer method and thus incurring exchange rate cuts for both, Mr A’s funds in USD are deposited into Mr B’s U.S. account and Mr B’s funds in GBP are deposited into Mr A’s U.K. account. TransferWise acts as a matchmaker across thousands of users and money never physically move across borders.

Promise amidst challenges

Despite initial success, there are challenges ahead for P2P platforms. They are still relatively unknown with surveys showing that not more than 10% of the bank users in developed countries know about them. Another concern is that P2P lending is not regulated in most countries and there are no rules on handling the collapse of a P2P platform. Neither are the deposits with P2P platforms insured by government guarantees like is the case for bank deposits. At present, P2P platforms handle no more than 0.1% of the deposits handled by banks but with the early promise they have shown and the inherent strengths of their business model, banks would rather watch out, as the tagline of TransferWise boldly proclaims, ‘bye, bye, banks’.

Also cross-posted here

Traditional Business Intelligence vs Modern Data Sciences

I have been a BI professional for a few years now. My exposure, like most of us, have been towards the traditional side of BI where the core job revolves around transforming and loading transactional data into a Data Warehouse, and developing reports on top of it.

Over these years, a few key trends like Social, Mobile, Cloud, Collaborative, Self-Service, In-Memory, etc. have had a significant impact on the BI market. As a result, the horizons of BI have expanded, and a lot more is now in the realms of possibility. These trends have also contributed towards the changing (increasing) expectations of Business users. Fair is to say that the metamorphosis of BI Application triggered by these trends has made the world of BI so much more interesting, challenging and rewarding.

From time to time, my interest has meandered from one trend to the other and each one has looked to have the promise of crossing over from being an academic fad to a practically implementable functionality. Of late, it was the Data Sciences which captured my imagination, and though am a bit late in the game as I realize, I decided to catch up and understand what Data Sciences is and how it stands to impact (transform) Business Intelligence.

As a convenient start, I have put across this table to differentiate between traditional BI and Data Sciences. I thought this comparison will be a worthy starting point since we invariably tend to understand the unknown (like, data sciences) by connecting it with the known (like, Traditional BI).

  Traditional BI Data Science
Volume of Data Handled Typically in GBs Typically in TBs, PBs
Diversity of Data Sources Lacks Diversity; Limited to Structured data that fits in conveniently into rows and columns of a Relational Database Highly Diverse; Combines Structured and Unstructured; Unstructured includes Social Media, Audio Files, Documents, E-Mails, etc.
Role of Statistics and Programming Minimal or no role of statistics; Programming limited to PL/SQL Statistics is a major influence; extensive programming in R, SAS, etc.
Interpretation of Results BI Practitioner’s role is limited to creation of reports. Interpretation of results left to Business Users. Data Scientists interpret results and communicate them to Business Stakeholders
Use of Visualizations Charts/Graphs used extensively but no usage of advanced visualizations Advanced visualizations required to communicate results, primarily because of the complexity of data that is being analysed

Surely the above table is simplistic and incomplete. But it gives me a handle to hold while I move forward on my exploratory journey of Data Sciences.

I will delve deeper on the topic in subsequent posts. Would love to hear your thoughts.