Four steps to data-driven reinsurance
Is data readily available to those who can use it
and are they using it?
A reinsurance company outcompetes its rivals by being better
at interpreting and reacting to data. The principle of data democracy is to provide the
means of analysing data to people across the organisation. At a truly data-led reinsurance
company, the relevant data is always in the hands of the people who can make use of it.
Too often, reinsurance companies spend substantial amounts of time warehousing and cleansing
data that is not then used below the level of the board room. Some data does flow further
down the chain to individual underwriters, but it is rarely enriched with data from complementary
sources. For example, if you give an underwriter data they already know, their natural response
is: “So what?”. By contrast, if you give an underwriter the same data enriched with external
market data feeds, reserving inputs and claims developments, they can start to use it to price
business and make connections between their own work and that of their colleagues.
‘Performance obsession should pervade
the whole organisation’
As Pat Saporito argues in Applied Insurance Analytics1:
‘Analytics improve business processes,
decision making and overall business performance and profitability through insights gleaned and
actions taken based on these insights’. Performance obsession should pervade the whole
organisation. For this to happen a clear rationale and buy-in are needed. As far as data
democracy is concerned, there needs to be full commitment at management level, with clear
roles and responsibilities defined for driving good data practice. However, senior members
of an organisation need to be careful not to push the data agenda simply by making more data
requests. That is not data democratisation. It is merely giving more work to the already
overloaded underwriters, actuaries and reporting teams. Often data is held by finance, IT or
a business analytics team. It is important to ensure that there is a mechanism for other
functions of the organisation to query and analyse the data too. Individuals will only be
willing to add and interact with data if they see how it benefits them. Data should empower
each employee to fulfil their function more effectively. Data democracy gives more employees
the ability to make data-driven decisions.
Is data-crunching creating a bottleneck?
When modelling or reporting on risk, actuaries and underwriters spend a significant portion
of their time sourcing, structuring and cleansing data rather than conducting analytics and
reviewing results. While it is essential for them to be familiar with the data and to check
overall accuracy of the numbers, software should be doing the heavy lifting. Actuaries and
business analysts should be using their expertise to drive the business rather than perform
The issue becomes magnified when the sourcing, structuring and cleansing of data is performed
separately by different teams within the organisation. Often, business analysts will discover
that they have duplicated the work of other teams – both in manually cleansing data and in
creating similar charts and graphs. A lot of unnecessary time and frustration is wasted in the
inevitable reconciliation – one of the most expensive words in insurance.
Centrally managed and automated cleansing dramatically reduces this overhead cost. More
importantly reliable and accepted data gives users confidence that they are making decisions
based on solid data.
Does the organisation contain information silos?
The organisation must encourage each function to have access to the outputs of other teams.
Inevitably, as different functions become familiar with each other’s data they will want to
react and use the data. Cross-functional teams can then be set up to maximise the value of
insights developed across the organisation. For example:
- underwriters can help the reserving team by showing how the portfolio mix has changed and
the impact it will have on future pay-out patterns
- reserving actuaries can show their incurred but not reported (IBNR) estimates to underwriters
so that they are clear as to whether the book of business they are writing is performing close
to their underwriting assumptions
- claims departments can highlight emerging risks and claims developments to underwriters
Automated feeds make it easier to have regular discussions around the analytical findings.
Now these discussions can occur when it is convenient, rather than as a reactive process when
an issue is finally escalated.
An inability to manipulate a pivot table or query a
database should not prevent valuable human capital
from participating in discussions.
The visualisation of data should be clear enough that a user with a relatively low level of
analytical sophistication can pick out the core messages and respond. An inability to manipulate
a pivot table or query a database should not prevent valuable human capital from participating
in discussions. It is important that data can be easily visualised to help users make sense of it.
Is performance measurement transparent?
The organisation must be set up so that the correct incentives are in place to reward
performance. In particular, underwriters should be incentivised to write good business.
Underwriters often have pricing adequacy targets but it is also essential that these key
performance indicators (KPIs) are reviewed against actual results. Reliable and rich data
on business written is the key to accurate judgment of underwriting quality. When underwriters
and actuaries feel empowered to avoid unprofitable business and focus on profitable business,
the overall business benefits.
Financial budgeting, planning and forecasting should be performed as routine, rather than
on an ad hoc or intermittent basis. The process should take account of market realities with
challenging but fair adjustments to accommodate the results (e.g. have motor premiums increased
due to new cars sold or due to changes in rates?). A static 12-month forecast does not encourage an
underwriting unit to look for new underwriting tactics, as the relatively short time window
constrains their ability to review the relevant information and resolve any problems arising.
If a unit’s performance is determined by factors outside of their control (e.g. a big deal was
won halfway through the year, or government sanctions have prevented a unit from writing a portion
of their portfolio), there is less incentive to improve performance as the KPIs are no longer
relevant. Implementing dynamic targets and perpetual monitoring ensures that staff members are
able to fulfil their objectives whilst obtaining value from the process.
Applied Insurance Analytics