To better understand the concept of analytics, it helps to have specific examples. This section focuses on ways in which Holborn’s customized analytical tools, developed onsite by our cat modelers and actuaries, help clients better determine their risk, improve their risk profile, maintain strong BCAR ratings and successfully manage growth, among other challenges.
Every company, including yours, faces a unique set of challenges, as well as ones commonly faced by many of your competitors. For those common issues, Holborn has developed a comprehensive suite of analytical tools that can help you.
When our existing analytics don’t address your issue, Holborn will develop new tools that are specific to your needs. The key to our ability to respond is that our analytics are not black boxes, but an open architecture built by our senior actuaries and cat modelers that work with you and your team to design and build the required analyses.
Recent examples of our adaptable tools include:
- A Midwest company was struggling with how to price their policies – their rating was too dependent on the actual experiences of their clients. For those that were unlucky to have catastrophe losses, premiums were skyrocketing and for those that were fortunate not to have losses, their premiums were insufficient for the risk. To help this company, Holborn built a series of tools that analyzed each client’s experience and then blended that experience with exposure-based loss estimates to price each policy in a manner that better reflected the catastrophe and other large loss risk.
- A company had experienced a jump in large, individual fire losses and wanted to know whether that was a blip in their experience, or was it the “new normal”? Holborn analyzed their experience in the context of what was happening in the economy and found that the frequency of these large losses was likely tied to the price of home heating oil – the spike in oil prices had led homeowners to use alternative heating. These alternatives were resulting in more fires (e.g., from improperly installed or unreported woodstoves or flues that weren’t cleaned). We concluded these losses were part of the new normal until the impact of higher oil prices “burned off,” e.g., oil prices returned to more normal levels, or people had adjusted to the higher level.
Many features of our exposure management tools grew out of needs expressed by our clients, who wanted to “see” where their exposures were.
Holborn has worked with a number of its clients to meet the requirements of the NAIC’s Own Risk and Solvency Assessment (“ORSA”), and also has been working with clients to improve their Enterprise Risk Management (“ERM”). We can use the extensive experience we’ve gained by working on these issues for others to help you.
For example, we have helped a number of clients develop an inventory of risks, by either starting with our generic list of risks or acting as a facilitator for their senior management to independently develop their own list. Once this list is finished, it’s used to create a risk profile that serves as a cornerstone of their ORSA response. The risk profile gives a description of the risk: its probability and severity; cause(s); who “owns” it; how it’s monitored and measured, and how often; mitigation plans; and action plan if it occurs.
Dynamic financial analysis, using our Economic Capital Model (“ECM”), evaluates the risks outlined in the risk profile through a Monte Carlo simulation. Depending on Holborn’s role in the reinsurance placement, much of the inputs are developed as part of the annual renewal process. We have separate ECMs for P&C and life companies (as well as one that combines the two). While the model handles a wide array of risks, each company has risks that don’t fit neatly into an ECM. We call these “singular points of risk.” For example, one of our clients rewards their top agents with a trip, accompanied by many of the company’s top management. The singular point of risk is a plane crash in which all of these agents and managers die. This single event generates multiple sources of loss, e.g., WC for the managers, loss of business controlled by the agents, expense, time and effort to recruit and train new agents and personnel, as well as a significant life insurance loss (all the agents buy life insurance from the company’s life subsidiary). Not only does this event not fit neatly into either P&C or the life ECMs, the joint impact has to be modeled.
A key component of ORSA is how you will measure and manage the risks. We’ve worked with many clients to develop the appropriate risk metrics that reflect their unique risk appetite and risk tolerances, which are then incorporated into our ECM. We have also reviewed the ORSA documentation for our clients to ensure that the report accurately describes their ERM processes be modeled.
How will a change in my reinsurance program and/or investment mix impact my company?
For example, what’s the impact on my BCAR and rating, as well as my capital needs?
There are many things a company can do that materially change its risk profile. Changing the reinsurance program or the investment philosophy is among the most common and each offers very different risk/reward trade-offs. They also interact with each. For example, if the company retains more risk, the need for liquidity can rise dramatically and impact the optimal investment mix. In addition, a change in inflation, all else being equal, can impact both the amount of loss retained and the returns the company can earn on its investments.
As with any change, there is likely to be an impact on the company’s rating. To evaluate these changes, you need a model that incorporates underwriting and investment risk.
These days, almost all companies rely, to a lesser or greater degree, on loss estimates from two or three commercially available catastrophe models. As you’d expect, Holborn runs these models for many of our clients. However, for many of these clients, the models don’t do a very good job of estimating the loss of specific events. This isn’t surprising because the models weren’t designed to predict the loss from any one event, but to provide a sense of the overall catastrophe risk to which a company is exposed. If that’s the case, Holborn has developed a number of analyses you can use to assess how well the models “fit” your exposures and experience.
One of the most important aspects of the modeling is the data on which the estimates are based. Holborn’s team of data professionals will review the exposure data you use to assess its overall completeness and quality, including building characteristics and policy terms and whether any of them are out of the norm. We also review the assumptions regarding insurance to value, business interruption, secondary modifiers and the like. We provide you with a Data Quality Assessment Report (“DQAR”) that outlines any concerns we have about the data so we can work together to get the data to be as good as it can be. This also provides valuable feedback to you to improve data in your systems.
Once we have the data ready to import into the models, we run the models (AIR and RMS) and provide you with many of the standard analyses, including average annual losses, exceedence of probability curves, TVAR, etc.
We then go a step further and try to break the standard loss output into more meaningful pieces by decomposing the EP curves into their separate frequency and severity distributions.
In another example, a client felt that the catastrophe models differed significantly in their view of the risk of hurricane in New England and the impact on his portfolio. To help them better understand the models, we broke down the hurricane losses by category (i.e., 1 – 5). We found that the views of frequency of hurricanes of a given category were fairly similar. It was the severity of loss by category that was driving the difference. The company’s management could now evaluate the models based on how a given size of storm would impact their book. We also looked at the losses if a historical event (e.g., the 1938 “Long Island Express” or the tornado that hit Joplin, Missouri) hit their book of business to give management a frame of reference for the modeled losses.
In some cases, your unique exposures may distort the results of the models. For example, a client specialized in insuring schools in the northeast. When one of the models changed, the indicated losses increased significantly. Digging in the losses, we isolated the “problem” as being their school book – schools were particularly hard hit by the new version of the model. We looked at their own experience in other states and worked with the company to bring the model results more in line with their experience.
Given the competitive pressures they face, companies can’t afford to buy an inefficient reinsurance program. Any evaluation of a reinsurance program has to look at both the cost and benefit.
The first step in knowing if you can eat more of your own cooking is to know the risks in doing so. Holborn has developed industry leading analytics that will help you understand the volatility of your losses. Our tools blend the output of exposure-based models with your own experience to estimate the mean and distribution of the frequency and severity of large losses by line of business, territory or other meaningful breakdown of your business. By looking at individual ground-up losses, we can identify trends in your experience that may influence your decision about retaining risk.
Once we have a common understanding of your risk, we input the frequency and severity distributions into our Quote Evaluator Program (“QEP”) to simulate thousands of years of loss activity. The QEP program can also reflect “gray swans,” those highly unusual losses (either severity or frequency) that drive the tail risk and are the main reasons for reinsurance. We can then run varying reinsurance programs against these losses to determine how each impacts your net loss position. The QEP provides statistics to evaluate each reinsurance program against your success metrics (e.g., net volatility, benchmark pricing).
While models may help you understand your risk, some events are so infrequent that models really don’t capture the risk to your firm. For these risks, we look at specific events (e.g., an F5 tornado hitting your largest concentration of insurance values) and exposure aggregations to determine how much reinsurance the company should buy.
Changing your reinsurance program impacts your company well beyond the dollars and cents of your premiums and losses. For example, any change will influence how A. M. Best (and other rating agencies) views your company. Our analytics provide you with the information to present to A. M. Best to demonstrate the risk-reward analysis you went through to make your decision.
Growth can create strain on a company in many different areas, including capital and ratings. Reinsurance is one form of capital (or risk management/mitigation). Our Economic Capital Model will analyze how your growth will affect your risk profile and whether your reinsurance program will support your growth or not. (See above “With all the talk around ERM and ORSA, how can I stay head of the curve?” for more information).
Our ModelFit analysis can help you understand how your growth will impact your catastrophe risk, both size of loss (when you grow inside your current footprint) and frequency of loss (when you expand your footprint, thereby creating a larger target for Mother Nature). Part of that analysis may be to create a portfolio that represents the kind and number of additional risks you intend to write, and then run them through the cat models to see how they impact your PML (both gross and net).
Our clients also ask for our advice when they want to enter into new lines of business or states. In either case, Holborn can advise you about the risks in those new lines or states. For example, one of our clients wanted to expand the states it wrote in. For the states it was considering, we provided loss experience of the companies writing in those states, a review of the regulatory and claim environment, as well as the catastrophe and terrorism potential.