An interview with Łukasz Kuryłowicz

Łukasz Kuryłowicz is a doctor of economic sciences at the Warsaw School of Economics. In combining both his position at one of the largest insurance groups in Poland and his  academic research in the fields of non-life insurance, risk theory, equilibrium of insurance markets and insurance telematics he showcases the power of mutual interconnections between academia and business. In this interview we have to honour to get an insight in his research and his motivation to follow this unique path. 

You are a professor at the Warsaw School of Economics and work at one of the largest insurance groups in Poland. Could you tell us what you like the most about both of your positions and why you decided to follow an academic as well as corporate career?

I was halfway through my PhD program when I started my first job in insurance. And although at that time I was mainly devoted to banking issues, I quickly realized that insurance gave me not only greater satisfaction, but also a wider array of opportunities for future research. The ability to combine business and scientific work quickly bore fruit. Insurance companies are an inexhaustible source of new issues and real problems that often require a scientific approach to solving them. The university, in turn, is a source of knowledge and new ideas that allow for a creative and comprehensive approach to problems occurring in the insurance sector, and the results of research permanently expands available knowledge. 

The best thing about this synergy is the opportunity to test purely theoretical ideas on a living organism.

What I value most is that working in business allows me to have a real impact on the lives and well-being of our clients, providing them with the real insurance protection they need so much. At the university, in addition to teaching activities, which brings a lot of satisfaction, I also have access to "strong minds" which have often helped me open my eyes and look at certain problems from a completely different perspective. This is priceless.
 

In your research you investigated the efficiency of Usage based insurance/ Pay as you drive insurance. Could you sketch the difference to traditional insurance and how does UBI benefit both parties of the contract?

In insurance everything revolves around the correct risk selection and pricing. When deciding on insurance, each of us hopes to pay a fair price. The problem, however, is that insurers have only limited knowledge about the risk related to the insured item or asset – I mean there is information asymmetry –  and clients are not always willing to share additional information that would fill this gap. When asked directly, clients often do not provide reliable answers because they know that this may negatively affect the premium to be paid. Insurers must therefore make certain assumptions as to the possibility of filing a claim by a specific client – based on easily accessible data such as the policyholder's age, profession or place of residence – and consequently spread the cost of future damage equally among all policyholders. The more egalitarian fashion clients are treated in, the greater the chances for the emergence of another undesirable mechanism – adverse selection. As a result, low-risk clients will stop taking out insurance because they will believe that the price offered to them is too high. At the same time, an unexceptionally large number of high-risk clients will opt in because they will consider the offered premium to be particularly attractive. In the long term, this may lead to the insurer's ruin.

Usage-based insurance or “UBI” seems to be a remedy for the above problems. With this type of insurance – especially when it uses insurance telematics – an insurer no longer has to rely on some general assumptions about the risks. An insurer is able to automatically obtain objective data regarding the use of the insured asset, e.g., a car, as well as the actual behaviour of the insured, which may affect the likelihood of filing a claim. Thereby, the insurer is able to offer a fair premium that is in a correct relationship with the real risk connected with each policy.

UBI also gives customers the opportunity to offer certain additional benefits that are meant to increase customers’ safety. When it comes to motor insurance, current apps provide customers with feedback on their driving style, certain tips to improve it, or instantly inform an insurer about an accident that occurred on the road  based on a crash detection. This allows for the emergency services to be quickly dispatched to the scene, often saving the lives of the injured. There are also more down-to-earth benefits such as loyalty points obtained for safe driving which can then be exchanged, for example, for coffee at the gas station or a discount at the garage. 

There are many more benefits of UBI implementation but if I were to enumerate the main ones, I could say with all certainty that thanks to UBI, insurers are able to price risks correctly and therefore forecast financial results more accurately and improve a customer portfolio by attracting fewer high-risk clients. Moreover, the gathered data allow insurers to recreate the circumstances of the accident if one takes place. In return, clients receive not only a fair premium, tips and instructions that add to their awareness about risks and road safety but also access to additional services that would be otherwise unavailable without UBI.

Telematics based insurance can have a huge impact on the privacy of the policyholders. Could you give us examples of how to anticipate the danger of intervening significantly into the privacy of the policyholders?

I consider this to be some kind of myth. I agree that UBI is about collecting a huge set of data. In the case of motor insurance, this will include, for example, data on car speed, rapid accelerations, harsh braking or use of a mobile phone by a driver. However, the greatest concern is GPS-based location data. Since it is a Big Data set, the analysis and use of the data collected require the use of advanced algorithms. Dealing with them, if not in an automatic fashion, is basically expensive, very difficult and de facto pointless.

To put it more clearly, the insurer is not interested in whether a given policyholder visited an ABC store on a specific day and at a specific time. From the insurer’s point of view, it is more important how much time the client spends on local roads and on highways. Does it happen mainly at night or during the day. Does he/she drive more in the city or outside the city, etc.

Insurers in principle do not collect more data than what we provide, on a daily basis and for free, to Google or Facebook. 

The difference is that in the case of UBI, it is done with the informed consent of the policyholder and in exchange for a real benefit in the form of a fair premium and additional services. Moreover, by law this information is subject to special protection and may only be used for a specific purpose  – one agreed on with the policyholder.

Would you say that safe drivers are more likely to accept an invasion into the privacy for a discount in the insurance contract and thus benefit more from an UBI? If yes/no why (why not)?

The conducted research  – not only by me, but also by, for example, Frederik Seger and Kathrin Figl –  seems to confirm this thesis. Even intuitively we might expect that people who are aware that they have nothing to hide will be more willing to agree to share their private data. Similarly, in the case of high-risk policyholders, it would be irrational to disclose data because it would involve a higher insurance cost for them. At least in theory, we could expect that the so-called self-selection works, and the very fact of choosing between UBI and traditional insurance can suggest the level of risk associated with the client.

In practice, everything depends on many factors, e.g., the model adopted by the insurer that offers UBI, the level of competition, the level of development of the insurance market or the characteristics of a given society. We can imagine that there is a certain group of high-risk drivers who, due to their insurance history and consequently very high insurance costs, will strive to find cheaper cover, regardless of the costs involved. Such drivers will allow for their privacy to be invaded even for the small discount they might receive. There are also safe drivers who are very protective of their privacy and even a huge discount will not convince them to sign a UBI contract.

In principle, however, we can state that safe drivers need fewer incentives to opt for UBI than high-risk ones do.

Your investigation on acceptance of UBI mainly focused on the Polish insurance market. Would you expect another outcome of a study in a different country for example the Netherlands and if yes, why? How would you describe the relationship between the willingness of sharing driving information with the overall technological development of a country ?

The degree to which UBI is accepted undoubtedly varies significantly from country to country. As I already mentioned, this is influenced by many different factors. In Italy UBI is widely accepted by all groups of drivers  –  this is mainly the result of appropriate legal legislation, which greatly stimulates the development of this type of insurance. If we think about the countries of Eastern Europe, historical experiences are a huge obstacle. A significant part of the society still remembers the ubiquitous surveillance during the socialist period and takes the issue of disclosing their private data very seriously. 

Intuitively, technological development should potentially be one of the factors that will have the greatest impact. However, such a statement requires further research and analysis. Let's look at Estonia, which is called a digital country or digital economy. Despite the widespread acceptance of technological solutions, according to some preliminary estimates, the degree of UBI acceptance does not differ much from the one that is observed in neighbouring countries. Therefore, it can be suspected that a technological development may have a significant impact, but only when the appropriate environment (legal, economic or even social) is provided. Going forward, even the most digital society will not accept UBI if it does not see a specific value proposition in such a solution.

At the university, we have recently started additional research into European countries. The research that could provide answers to the above and additional questions. But so far we have not managed to collect enough data to be able to conduct an analysis with the appropriate level of statistical significance. It is still too early to draw correct conclusions.

Could you give us a glimpse of your current research? Which current technological developments will have a major impact on the insurance market in your opinion?

What interests me now is not a specific technological solution, but the forces that are often the creators of such solutions. I am talking about Insurtechs that have entered the insurance market changing its landscape by offering a completely new value proposition for both policyholders and insurers. So far, their presence has significantly influenced insurers' ability to develop new products, their sales channels and the reception of insurance by policyholders. 

Insurtechs are largely responsible for changing the image of the insurance sector which is no longer perceived as a deeply traditional and ossified industry.

 Insurance is starting to become more widely available, more understandable and, above all, perceived as modern solutions that keep up with the changing world. Therefore, I think that Insurtechs, through their activities, will surprise us more than once with some new solutions that will permanently change the insurance paradigm.

Which statistical and econometric methods do you use in your work?

It all depends on the goal I want to achieve. If I am to find the tools that are most frequently used in my work, they will be methods of descriptive statistics and statistical inference, analysis of stochastic processes and the dependence of variables. Recently, I have been also using statistical learning more and more often.

Artificial intelligence, big data and machine learning are some of the greatest innovations and challenges of our time. Have they found grounding in the field of insurance and risk management, and in what ways? and what about your own work?

The above phenomena were noticed by insurers several years ago. In the case of larger, more financially strong insurance companies, they have managed to take root and are used in many areas. These are mainly pricing, underwriting, claims handling and financial risk management. Smaller insurers are also taking the first steps, but the problem they face is collecting a sufficiently large amount of data that will allow the development of AI or ML. One must have in mind that Big Data is not just a large database. Big Data sets must "live", be constantly fed with data from many different sources, sources which may, not necessarily, be obvious at first glance. Access to this additional data – that is often very expensive – is the main obstacle to the further development of insurers.

My area mainly deals with product development, pricing and risk assessment in motor insurance. AI and ML are developed every day to ensure the most precise pricing and underwriting. Therefore, we are constantly looking for new sources of data that could improve the effectiveness of risk assessment models (one such source is data collected under UBI).

How does climate change impact the world of insurance? What are some of the challenges that the insurance industry has to face to ensure its functioning in the future.

This problem is incredibly complex. As a rule, insurers should operate within the so-called insurable risk. This must be characterized by a certain statistical regularity and the losses resulting from the risk should not be catastrophic. Unfortunately, the losses resulting from climate change can in most cases be called disastrous. Their scope is so large that without appropriate risk pooling between insurers, not a single insurance company would be able to bear them. Therefore, insurers try to avoid taking responsibility for these losses, and thus, according to the estimates, only between one fourth and one third of the total losses caused by extreme weather and climate-related events are insured. In most cases, their costs are borne by the entire society via state budget funds. In the coming years, an increased emphasis will be placed on finding a way to reduce this protection gap. New solutions are being developed to ensure appropriate risk pooling, e.g., CAT bonds, which allows for the risk to be transferred to many investors. The reinsurance system should also play an appropriate role. Only ensuring an appropriate ecosystem for insurers will allow them to develop products that can provide adequate protection. 

Insurers cannot also remain passive and should now use the tools available to them to increase awareness of these risks among policyholders and motivate them to use appropriate methods to protect themselves against the consequences of climate-related claims. This is the so-called insurance prevention, which more than ever should be applied jointly by all market participants.

What are the most important skills for students seeking to work in the field of insurance?

Working in insurance provides development opportunities in many different business areas, which in turn require various competencies, both soft and hard ones. For example, a product development requires the ability to navigate through legal issues and to discover customer needs, and to notice any changes in the environment as well. Not to mention some strong business analysis competencies. In the areas of pricing, risk management and actuarial offices, solid knowledge of statistics and probability is necessary. You also need to have financial knowledge. And the main tools that are more and more invaluable are R and Python. You also need to have an open mind and to be able to notice the interdependence of certain phenomena in order to understand some anomalies appearing in the analyzed data.

Students who are more comfortable with pure economic issues or who specialize in finance will also find their place in insurance – the insurance sector is the largest institutional investor in the European Union, investing over 10,000 billion euro in financial markets.

Furthermore, other skills will be expected from people who want to work in claims settlement or in the area of insurance frauds. This often requires specialized not only technical knowledge, but also some detective skills.

It is this diversity of tasks and functions that makes work in insurance demanding but, above all, satisfying.