Insurance pricing is a complex exercise. To set the premium price for a policy, an insurer must evaluate various risk factors associated with the person, property or entity to be covered. And insurers shouldn’t be limited to their own internal risk assessment
alone; they should combine their own data with market data to understand the pricing strategies and rationale of their competitors.
Unfortunately, that process isn’t as simple as adding one data silo to another. It’s crucial that both data sets are complementary and achieve the same high standard of detail, granularity and completeness.
Insurers confront a landslide of data every day, including policy and risk assessment details across online portals, forms and paper. Yet there is a
drought in data that reaches the appropriate level of detail and granularity to make accurate, equitable pricing.
It’s much easier than most of us would like to think to inadvertently reach an unjustly biased decision based on incomplete or inconsistent data.
Technology, led by trustworthy innovation, presents the exciting possibility to overcome this issue – through the use of synthetic data.
What is synthetic data?
Synthetic data is generated by either by applying a sampling technique to real-world data or by creating simulation scenarios in which models and processes interact to create completely new data.
Usually, we refer to two different algorithms for generating synthetic data: GAN (Generative Adversarial Network) and SMOTE (Synthetic Minority Oversampling Technique).
GANs are advanced deep learning tools that generate new data mimicking specific distributions. They efficiently produce synthetic data that closely resembles real-world data.
SMOTE, on the other hand, is used to create new data and augment existing data sets to enhance data richness. It generates new examples specifically from the minority class to achieve balance.
So how can synthetic data support the actuarial pricing process?
Balancing price and competition
Insurance pricing and reserving is all about market share and profitability. If you increase the price of an insurance product to become more profitable, how does that affect your overall market share in a consumer segment?
The answer depends not only on an insurer’s own risk assessment, but on the pricing and product strategies of competitors.
Let’s say an insurer is profitable with a specific product and increases its price by 10%. If their competitors increase their prices by 20%, the insurer retains its market share and increases profitability.
Naturally, if an insurer knew they could increase their prices by 20% alongside the competition and still hold onto market share and increase profitability further, they would. But how can you know if that’s possible?
Optimizing the premium pricing process: the old way
Currently, most insurers calculate prices using Generalized Linear Models (GLMs). These statistical models are fundamental tools in insurance pricing and risk assessment, most often employed in the insurance industry by actuaries and data scientists to predict
the likelihood and cost of claims.
All the factors generally used with this approach are derived from an insurer’s internal pricing structure. So it ignores other market players, and the carrier is flying blind to the competition.
Optimizing premium pricing means combining an insurer’s own data with available market data, then fine-tuning the relationship between the carrier’s individual premiums and market premiums. Easier said than done.
Increase the price too much where you are making losses, and you may lose profitable business, too. But if competitors are also suffering losses and have to increase their prices to the same extent, that price increase may not cost you any business.
How do you know? And even if you know, how long does it take you to make the right decisions based on this insight?
Most insurers have a single department that collects competitive market information. They analyze the data using spreadsheets, which can take up to several months – too late to inform the best business decision. By the time a carrier makes a decision, frequencies,
loss ratio and dynamics have changed. In short: Their data has gone stale.
So in the end, predictions would only be based on limited, inaccurate data, leaving the carrier struggling to come up with correct pricing.
Enter synthetic data
The good news is that such external market data is available. Most insurers already have access to it via external web services.
However, this data still must be integrated into the carrier’s pricing system and combined with internal data, KPIs and other information to discover any trends. And how would these trends affect your own decisions?
Let’s say a competitor increases its prices and would therefore lose some clients; these clients might be willing to buy your products. But if the competitor’s price increase was driven by high losses from these customers, then selling them your services
would mean adding a lot of unprofitable customers, deteriorating your own loss ratio.
You would need to be alerted to such dynamics so as not to make the wrong decisions when these customers knock on your door. This means you need to combine your own information about market trends, competitive information about market trends, and the data
from your own pricing system. And you need to do this fast and get the results quickly.
We have heard from some insurers that this process can take up to six months. Even if those are extreme cases, anything close to that is far too long.
So how can the process be accelerated?
With synthetic data, you can augment your data sets at the level of detail required for your analysis. Sophisticated AI techniques can provide the models needed to analyze market data as it becomes available, calculate forecasts based on market trends, and
match the data with your own internal data to improve your own actuarial process.
With this near-real-time decisioning approach, you could get the alerts and triggers in time to adjust your pricing approach faster than your competitors.
Conclusion
Synthetic data can help insurers overcome incomplete or outdated information and make faster, more accurate premium pricing decisions. By combining market insights with internal data, carriers can anticipate competitor moves, optimize profitability and maintain
market share, giving them an advantage – or at least allowing them to keep up – in a competitive environment.
