Mike Ashurst:

Hello everyone, and welcome to today’s ICMIF webinar. I’m Mike Ashurst and I’m pleased to introduce today’s webinar on property pricing from a reinsurer’s perspective. Today’s webinar is from Peak Re, and without further ado, I’ll hand you over to Andy Souter, the Director of Global Markets, who will start the presentation. Over to you Andy.

Andy Souter:

Thanks very much, Mike. And everyone, good afternoon, good morning, and thank you for joining our Peak Re webinar today. Recently ICMIF and Peak Re ran a joint survey on risk and capital management. And as a risk and capital tool, reinsurance ranked significantly in both the important and very important categories, which is expected, especially for mutuals, where access to debt and equity can be more limited. Furthermore, for insurance companies including mutuals, reinsurance can be a notable cost item, while at the same time it may not always be clear how reinsurers arrive at the pricing views that drive the cost of reinsurance.

What are the components that drive the pricing? Why does pricing vary between reinsurer? And how can the reinsurance buyer influence the pricing result? For example, how does the quality and detail of data affect pricing? Our webinar today is designed to lift the lid on the building blocks of pricing from a general reinsurer perspective. And providing insights today are two colleagues from our P&C team, Chi-Hang Wong and Andrew McGuinness. We hope you find the webinar insightful today and we’d welcome questions at the end of the webinar. So I’ll pass over to Chi-Hang Wong, who will start the presentation. Thank you.

Chi-Hang Wong:

Thank you Andy. Today my topic is about property pricing from a reinsurance perspective. Before I go into details how we perform our pricing, perhaps as a start, I should break down the price into different components. In my opinion, in the price there are five different components. Number one is expected loss, because when we enter into a reinsurance agreement, reinsurer will have to pay the loss, so the price after you factor in the expected loss. The second factor is acquisition costs because we have to pay the cedent commission and also we have to pay brokerage to our brokers. And third is management expense, because we have to take into account our salary, the rents of our office, cost, et cetera. And fourth component, which may not sound too obvious, but is cost of capital. Because cost of capital normally does not appear in your financial statement so it may be ignored by some people. But then cost of capital should be considered because when our shareholder raise the fund to set up a reinsurer, they inherit the cost.

When we define cost of capital, it can be defined either as the interest rate when our shareholder raised the fund, or the expected return, which our shareholder can earn when they invest in other places. In other words it’s the opportunity cost for the capital. And because of capital is sensitive to the volatility of the underlying business. Why I’m saying this is because if you refer to your internal capital model, and very likely you’ll agree that capital is based on the volatility of your entire portfolio. That’s why I said, when we do pricing, when we allocate cost of capital to a treaty we can see there the volatility of that treaty.

And the final component is the expected profit margin. And similar to cost of capital, it is a function of capital. Because normally when shareholders state the requirement on profit margin, they normally state it as return on equity, which means it’s a return, which using your capital as a basis. That’s why profit margin is also affected by the volatility with similar rationale to the cost of capital.

And just a bit of clarification on the definition of profit margin here, in here, in my slide, the profit margin is defined as the excess return above cost of capital. Because one thing is because some people may just lump profit margin and cost of capital together, and they define the profit margin as the premium minus expected loss and expenses. But then why I would like to split it up is because, in my view, cost of capital is more related to the cost when the shareholder raised the capital. While the expected profit margin is actually more driven by the current market condition. It is because the reinsurance market is pretty cyclical. Sometimes when reinsurance capacity is abundant, we will say we are entering a soft market, because capacity is too many and the price will go down, and as a result the profit margin will be lower and vice versa, when we enter a hard market, the return on equity expectation from our shareholder will be higher because they expect the price to be higher. That’s why I split it up, cost of capital and profit margin, so that we can define different return on equity depends on the market situation.

All right, after breaking down the price into the five components, we can then define what is the technical price? A technical price is actually means, when the premium pays all these five components, they are considered as a technical price. Say for example, in our present model it gives you $75 of expected loss, $10 of acquisition cost, $5 cost of capital, $5 management expense, and 10% expected profit margin, and all in total gives you $105. But then if the reinsurance treaty only pays you $75, as a reinsurer we will consider it as severely inadequate. But then if the reinsurance treaty pays you $105, we consider this as adequately priced and $105 is actually our technical price.

In the top table you can see how profit margin and cost of capital is affected by volatility. In here we keep everything constant, we have constant expected loss, we have constant acquisition cost, constant management expense, but the only moving part is the volatility of the treaty. Let’s say in a low volatility treaty, the volatility is only $80, it’s either defined as the TVaR, the VaR, or a standard deviation. And then for the volatility we can then refer to our internal capital model, it may say that per dollar of volatility you have to put 1.2 dollars as your required capital. And then the cost of capital is 6.5%, let’s say, is the interest rate for our shareholder. So the cost of capital is the amount of volatility multiplied by the cost of capital volatility ratio and the cost of capital ratio, which gives you the cost of capital for the treaty.

And similar rationale applies to target profit margin. And you can see here, for a treaty with high volatility, the reinsurer will demand high cost of capital, and target profit margin, as a result the technical price will be higher for higher volatility treaties despite the fact that expected loss and the cost are exactly the same.

And then in the bottom table, again, we keep everything constant, but the moving part will be which market cycle we are in. And here we have three scenarios, one is soft market, average, and hard. And in each scenario our shareholders require a different return on equity. And then we apply the same formula to calculate the profit margin. In this case the cost of capital is not affected because the cost of capital is still a constant, it’s not affected by the market cycle. But then the profit margin is higher when we’re in a hard market and lower in a soft market. That’s why our technical price is actually not a constant, but then it’s moved along the market cycle.

When we place a reinsurance contract, it often comes into a situation that even for the same treaty all our reinsurers are pricing based on same data using the same model, but eventually it gives out a very different price. And in this slide, I tried to list out several factors which cause this phenomenon. I think factor number one is how diversified the reinsurer is. Because the degree of diversification is affecting how the reinsurer allocate their capital to different segment of their portfolio.

In this example we can see there are two different reinsurers. One is a global reinsurer, who did both, let’s say Japan property, and segment two will be U.S. property. And the other reinsurer is just a local reinsurer in Japan, say for example, and only do Japan property. And then the first step is to determine what is the total capital required for these two reinsurers to run their business? And before entering into that, we have to define a tail risk measure in order for them to define the capital. In this example I used the tail feather at risk. What is a tail feather at risk, is basically saying, for a given percent, how in this example I use 80% TVaR, and then we have 10 different trials, so the 80% TVaR means the two worst scenarios and we take the average and then we get the 80% TVaR. If we look at this column for the global reinsurer, the worst two scenarios… in this table I’m showing the underwriting profit and numbers in bracket representing underwriting losses. So the two worst scenarios are the 5.6 underwriting loss and the 6.5 underwriting loss, and we take the average of these two, we get the 80% TVaR of 6.1 million. And then we do the same for the local reinsurer.

And you may notice that actually the total underwriting results for these two reinsurers are exactly the same. That’s why on a overall portfolio basis, the 80% TVaR are the same, the tail risks are the same for these two reinsurers, therefore the total required capital for these two reinsurers should be the same because, from an overall perspective, the risk level are the same. In order to run this business, both reinsurer will require 10 million of capital. This is our first step to determine the overall capital requirement.

And then the second step will be to allocate this 10 million to different segment in their business. And for this local reinsurer, because they only write one segment, so there is no diversification at all, and it has to allocate the entire 10 million to this Japan property business. But then for this global reinsurer we can allocate this 10 million to different segments, and the allocation method is based on a marginal TVaR. There are many ways to allocate capital but the example I’m providing here is a very simple one. Just want to illustrate to you guys how diversification can affect the capital allocation. That’s it.

And in here I allocate the capital, I look at the marginal TVaR for each segment of the portfolio. And then how we define marginal TVaR. Say for example, when we look at this segment one, the marginal TVaR is defined as when we add segment one into this portfolio, how much TVaR will the portfolio move? Let’s say without segment one, there is only segment two in their portfolio, that’s why the TVaR for the entire portfolio is 5.7 million. But then when we add segment one to the portfolio, the portfolio TVaR will become 6.1 million. That’s why we can define the marginal TVaR as this 6.1 million, and this 5.7 million, which gives you around 300K. And then we do the same for segment two, and then we discover that the marginal TVaR for segment two is actually higher and is 2.1 million. And then we allocate this 10 million of capital to these two segments, we can allocate based on the relative magnitude of the TVaR, and then which gives you 15/85 split for the capital allocation. And then around 1.5 million will then be allocated to segment one, and 0.5 million allocated to segment two.

And then when we visit the capital to TVaR ratio again for the segment, we simply define the allocated capital by the TVaR for that segment, and then we can notice that the capital to TVaR ratio for this global reinsurer is much lower than local reinsurer. That’s why we say that a reinsurer with higher degree of diversification can enjoy the benefit of running individual segment with less capital relative to their premium size.

The second step is to allocate the total capital to individual segments. And the end result we would like to achieve is to allocate capital to individual treaties, that’s why we can determine the capital required for each treaty when we do our pricing. So the third step is to just cascade this capital allocated to the segment to individual treaties within the segment. And this allocation is actually very simple. We can simply multiply the capital to TVaR ratio for this segment to the individual TVaR for the treaty to get the allocated capital. Why am I saying that is because let’s assume that… let’s further break down this segment one into individual treaties here. And then unless we assume that, within the segment, all the treaties are highly correlated because they represent the same geographic region and same line of business. That’s why we cannot simply assume that they are 100% correlated. And you can see the underwriting result for these two treaties are moving up and down at the same time.

And then we repeat the same marginal TVaR exercise again, and you will discover that the capital to TVaR ratio for individual treaty is actually the same as the segment overall. That’s why I say that we can simply multiply this capital to TVaR ratio to individual treaties TVaR to get our cost of capital for individual treaty.

This is the first factor that affect our pricing, which is the degree of diversification. Usually with everything costed a global reinsurer, or a multiliner, should charge lower than a monoliner, or a local reinsurer, because they can use less capital to run their business. This is the first factor, and the second factor is the credit rating for the reinsurers because to get a high credit rating you have to hold more capital relative to your premium size. That’s why as a natural consequence a reinsurer with a high credit rating will have high cost of capital, that’s why they will charge you high premium. And the third factor here is the operating model. A reinsurer with a lean team can enjoy lower management expenses, that’s why they can price themselves more competitively in the market.

In the next few slides I’m going to talk about for both proportional treaty and non-proportional treaty, what data we will be utilizing for the pricing and what are the methodologies for the pricing. And let’s first go through the proportional treaty. For proportional treaty, the data required, I think you guys are very familiar with it already, because this normally the data you present to reinsurers in your summation deck. And the data required are the premium and loss triangulation, and your underwriting rate change history, thirdly is the large and cat loss statistics, and last but not least the cat exposure info, how we are going to utilize it I’ll get to later.

And for the pricing for proportional treaty, we’ll split the entire loss ratio into three different components. One is the attritional loss ratio, which means the small losses, and large loss ratio, and then cat loss ratio. Why we are doing this because we have to adopt different techniques to arrive at the answer for these three different categories of loss ratios. Firstly, on attritional loss ratio, we will first utilizing the premium and loss triangulation to develop the ultimate loss ratio. It is why we need triangulation but not just the latest [inaudible 00:23:42] is that for the recent underwriting years especially for treaties, usually the both claims and premiums will still undergo development because there may be some latent reporting premium, there may be some claims still outstanding pending the settlement, et cetera, et cetera. That’s why we have to utilize the data in more mature underwriting years to project the ultimate loss ratio for the more recent underwriting years. That’s why normally we would require a triangulation is that just the latest of the incurred position.

After developing the ultimate loss ratio we have to do some trending because your portfolio is not constant, perhaps your underlying premium rate is changing, perhaps you are increasing the rate over years. That’s why your premium in the old underwriting years are too low and we have to have your rate change reflected in the ultimate premium. That’s why we have to first to do an exercise to try to bring all the premium to your current premium rate level. And for on loss side we have to do some inflation to bring the losses to the current cost level. After developing the loss to the ultimate and trending, we get the final trend, the ultimate attritional loss ratio. And then we can use this attritional loss ratio to do aggregate loss modeling. We can get different percentiles on the attritional loss ratio, and we can get the expected attritional loss ratio as well as the full utility behind that.

The second component will be the large loss ratio, and here we will utilize your large loss statistics. And we use your large loss statistics to do some modeling on the expected large loss, as well as the volatility. But we are not simply taking a large loss similar to the attritional loss ratio, we have to do some trending as well, especially on the frequency side because the size of your portfolio is moving over time. When your portfolio is growing we have to trend the frequency upward and vice versa, if your portfolio is shrinking we have to trend the frequency downward, to try to reflect as much as possible on your recent situation. And on the severity side, similar to attritional loss ratio, we have to trend the losses with the inflation. This is how we model large loss.

And for the cat losses, normally for modeled perils we can adopt the same approach as the large loss ratio. We use the frequency severity modeling. For the modeled perils, say for example, earthquake, typhoon, we can use the vendor models for pricing analysis and we will utilize your cat exposure info for that kind of analysis.

This is how we arrive at the expected loss for the proportional treaty. But this not the end of the game because this is just one component. For pricing proportional treaty we are not trying to… of course we are not trying to negotiate the price because the premium are shared at a predetermined percentage between the cedent and the reinsurer. But normally when pricing a proportional treaty the reinsurer will try to negotiate the acquisition costs with the cedent. And in simple cases, we just negotiate on a fixed commission but in case, especially for volatile treaties, reinsurers may like to negotiate for a sliding scale commission or even a loss participation. And in this light I’ll try to explain why reinsurer in a volatile situation they will prefer a sliding scale commission.

In this example, again we play 10 different trials, and then we have this loss ratio distribution and expected loss ratio, let’s say is 75%. And then we have two different scenarios here, one is the reinsurer simply pay you 15% fixed commission. And the other scenario is a sliding scale commission. And in this sliding scale actually, the reinsurer will pay you the same commission at the expected loss ratio, but then the reinsurer will get more protection by paying less commission at high loss ratio and they are willing to pay you, if you perform badly they will pay you badly commission.

In this simulation result you can see that actually, the expected commission for sliding scale is lower than fixed one. Despite the fact that they are paying the same commission rate at the expected loss ratio, it’s because this loss ratio distribution is more skewed to the heavier end, that’s why on the average level, the commission to be paid out is less than the fixed commission case. But this is not always the case, it’s just when the loss ratio is skewed to the heavier end, it can also be the other way around. And the main point for sliding scale is that sliding scale can reduce the volatility of the technical combined ratio, or the underwriting result for the reinsurer.

And then we can look at the 80% TVaR for these two scenarios and then you can see that, for the fixed commission case, the TVaR is higher. And for the sliding scale commission case the TVaR is lower because the reinsurers enjoy the protection by paying less commission in adverse scenarios. By having lower TVaR and as I said before, the volatility is affecting the cost of capital loading as well as the profit margin loading. That’s why in the sliding scale case the reinsurer will have lower cost of capital and profit margin loading. And actually if the cedent is willing to enter a sliding scale terms with the reinsurer, they may even pay you more than 15% at 75% loss ratio because they have less loading in other components of the price. That’s why I said reinsurers normally prefer sliding scale in volatile cases.

Then I’ll move on to excess of loss pricing. For excess of loss pricing the data required will be risk profile, showing different sum insured band and the premium for a different band, et cetera, et cetera. And then we’ll need the historical GNPI because we would like to know how the size of your portfolio is moving. The third one again is the rate change, and very similar to the proportional treaty cases, we will need large and cat losses and for long-term treaties we may even require triangulation on these large and cat losses. But for property, normally we just need the latest position. And then the cat exposure info for us to perform cat loading analysis.

So pricing non-proportional treaties there are many different methodology but they generally fall into two families. One is experience rating, in which we utilize more on the loss statistics, and the other family is exposure rating, in which we’ll utilize more on your exposure info like the risk profile or the cat exposure. And then in experience rating the most common use is the burning cost. We simply look at the historical statistics for the excess of loss treaties. Of course we don’t do a simple burning cost, just like the proportional cases we have to trend the losses to reflect the current loss level and your current size of your portfolio. And for frequency severity modeling it’s just the same as the proportional cases. The technique is the same.

And for exposure rating we have two different methods. One is first loss curve which we use quite often for risk excess of loss pricing. A first loss curve is actually a probability curve describing the damage ratio of a property, when this happens. And then we combine the risk profile together with the first loss curve. In the risk profile we get the average premium sum insured band and then using a premium we multiply to a select loss ratio. Normally that’s based on your portfolio experience or either based on the market benchmark. We can then arrive at the from ground-up loss cause and then together with the first loss curve we have to allocate this from ground-up loss cause to different layers. How we do that is we look at the risk profile, we get the average sum insured and then the second part we look at limit and deductible for that layer, and then we define the limit or deductible by the average sum insured, by that risk profile band, and then we can get the entry and exit damage ratio for that particular layer. And then for that damage ratio, we then look for the entry and exit probability in the first loss curve, and then we can work out the expected percentage to be allocated to the layer from the from ground-up loss cause. This is how the first loss curve is working.

And the second method is to use vendor model for the cat excess of loss treaties pricing. And it sounds quite confusing to have so many different methodologies for excess of loss pricing, and then in the next slide I’m going to give you some idea how reinsurers picked from different methodologies when they price a layer. First of all the reinsurer will look at the remoteness of the layer. Normally for a lower layer, because there are more losses hitting the layer, that’s why the loss experience is more credible. In general for a lower layer, reinsurers tend to use more experience rating. And the other way around for high layers because there is maybe no losses at all, that’s why reinsurers will rely more on exposure rating.

Even for a cat excess of loss treaties, there are different vendor models in the market and then how should we pick the appropriate vendor model from the vendor models appear in the market? And the first factor we will consider is for each region we would do some homework, we try to validate the model result again the historical event losses on a market wide basis.

When we pick a vendor model for cat excess of loss pricing, we will first validate our model against actual historical losses by picking the events from the vendor models event catalog, and then we try to compare either from a cedent perspective or from a market wide perspective to see which vendor model performs better on the historical losses. And sometimes one model may work better at the lower end, and sometimes the other model may work better at the higher end. That’s why we may take different models for different layers as well.

And this is one part, the second part is we will try to observe the general market perception on different vendor models. And we try not to deviate ourselves too much from the market because it would put us into a very awkward situation just because we pick a different model from the market. And this is how we pick the vendor model.

And in worst case, from the cedent we don’t see any credible data we will also adopt benchmark pricing. How we do benchmark pricing, say for example we may pick frequency severity parameters from your peers or we may use an industry cat loss curve and then apply your market share to try to price the layer. And I think this is the worst case for both the reinsurer and the cedent because from a cedent perspective you may be penalizing yourself by having… you may be having above average risk, but then when you don’t give us data, we only treat you as an average cedent in the market. Or even after all the pricing exercise you may find the price reasonable but then without credible data, normally I think underwriters will be reluctant to because the parameter risk for the pricing is too much.

And I think it comes to the end of my section, but before I hand over to Andrew, perhaps we just go back a few slides. And in this slide I tried to explain the relationship between diversification and the capital requirement. And if we expand this concept a little bit further, it actually applies to the reinsurer and cedent relationship, especially for mutual companies because normally you just focus on a particular region. That’s why you don’t employ diversification. But then when you enter a reinsurance agreement, actually you are entering a game of capital arbitrage because your insurer can do the same business with less capital. That’s why by doing reinsurance actually both parties are enjoying a lower cost. So let’s head over to Andrew on explaining how that works. Andrew, over to you.

Andrew McGuinness:

Thank you. Nice segue. I’m going to talk about something related to, but slightly different to reinsurance pricing, which Chi-Hang’s just expertly described. I want to discuss the benefits of geographic diversification and that’s a key to understanding why reinsurance can offer competitive price cat XLs, which probably compare well to alternative solutions available to insurance companies. Diversification allows insurers to hold less cats to cover the same risks with the same certainty. It enables capital to go further in terms of regulatory and rating benefits. And diversifying the portfolio enables insurers to ensure that they’re not solely exposed to one major peril making them more sustainable, resilient.

But it’s not always possible for insurance companies to achieve meaningful geographic diversification in their product portfolios especially on a global scale. I want to explain the advantage of seeking this diversification and show that accessing reinsurers is an efficient and attractive means for the insurer, both to control the volatility but also use this to reduce their capital costs.

Firstly a couple of examples. I thought it’d be nice to give real world examples of the ability of international, global reinsurance to provide large sums of contingent capital at the time of need in a specific region. This is obviously the first and most obvious benefit of cat XL, it supplies the money needed to rebuild properties damaged in a catastrophic event. The most extreme example we show here is Chile, where the direct economic loss according to the world bank is 8.8 billion, which is pretty huge, 9% of the country’s GDP. But most significantly for my purposes there, 95% of that picked up the international reinsurance market. The point in there I want to make is not just that reinsurance supplied the money after the loss, but the existence of this reinsurance capital of 8.8 billion, 9% of GDP, or more than that, had to be set aside by Chile in anticipation of a potential future earthquake before it happened.

That’s looking at it at a country level, but maybe talking to lots of reinsurers, it might be of greater interest to consider the individual company level, which I show below. I think the most compelling one here is that 2005, during the… the hurricane year, when we had Katrina, Rita, and Wilma, 12% of U.S. insurers received reinsurance recoverables that exceeded their entire share equity. So without reinsurers, they would have been going bust just paying their clients. 23%, which is still going to be… 23% of them received recoverables that exceeded a third of their equity capital, which is not quite as extreme as the example above, but is still going to really test the viability and resilience and future solvency of these companies. As it actually happened, the insurance industry absorbed the losses and emerged in pretty robust shape because cat reinsurance operated as intended.

A simple example of the statistical principle, which underpins the benefit of diversification. The first graph on the top left shows an imagined insurance portfolio containing 10 insurance policies. That means there are 11 possible outcomes for this portfolio. With those being zero losses on any of the 10 policies, one policy having a loss, two policies have a loss, all the way to all 10 policies having a loss. Claims are assumed to be uncorrelated with each other, so that that means that the probability of each outcome is equal and one in 11.

So if all of the policies could only have a loss of $100 or zero, that means the company must hold $900 of capital to meet this potential claims obligation in 90% of the potential outcomes with just that single extra 10th loss being that 10% set on top.

Now move on to the bottom right graph. Here a scenario where there are two portfolios identical to the one I just described. Both have 11 possible outcomes as described, but the combined portfolio would have 121 outcomes, which is 11 times 11 each, with a probability of about 0.38%. When you sum up all these outcomes with the total number of losses, it shows that the average loss, shown in the middle there, of 10 losses, is the most likely outcome. And the likelihood of experiencing the most extreme outcomes, good and bad, is significantly reduced. Because the loss cost follows this normal distribution shown on the orange bars.

There’s only a 0.83% chance of over zero losses or the worst case scenario out of all 20 policies having a loss, compared to a 9% chance of the average 10 losses occurring. The green line, again, is showing the cumulative probability, which if you’ve got very good eyes, you might be able to read from. To achieve the same certainty of 90% probability that you’d be able to meet all losses in a given year, you now only need to cover 16… the cost of 16 losses. If you can read up from that.

So using the same assumption of $100 per loss, on that combined diversified portfolio you’d now need to hold $1,600 as opposed to two sets of $900 on the individual portfolios without that diversification benefit. Hopefully this simple example demonstrate that diversification can reduce volatility and consequently reduces the insurers capital requirement. In reality the scale of the reduction can be much larger than the one that I show here in this example.

So diversification of risk is therefore normally a desirable goal because of that principle. The next step is to consider how you achieve better diversification for your portfolio. So here I show the risk which could potentially used, to be paired with yours and everyone else’s portfolio, to reduce that capital burden. For this I’ve used AIR’s model of expected loss globally, using their industry exposure database. AIR sorry are modeling, and this is their estimate of what the loss would be for the whole world, insured and insurable, both. On average per year and annually. And then a one in 100 year aggregate exceedance probability basis. So basically a 1% chance of having a loss or larger next year.

That database is pretty strong, it includes data about risk counts, structure attributes, replacement values, insurance policy terms and conditions, and then they use the hazard models, perils like earthquake and windstorm, to figure out what they think the loss to that the portfolio would be. Here I’m basically using the one in 100 year aggregate exceedance probability. The loss is a proxy for property catastrophe insurance risk globally. You can see on the right, North America’s contributing over half, which is going to be driven by Gulf hurricane and California quake. Japanese earthquake and typhoon have always been significant contributors but Asia’s share of the global cat risk total has been growing over the years and is at 17% now. Countries like India and China and those in Southeast Asia are growing. Insurable values increased in tandem as they build new and more expensive houses, commercial and industrial risks. And they’re often in geographies exposed to typhoon, floods, or earthquake, given the nature of the region.

Developed countries in Europe less cat exposed but lots of insurable value, they’ve still got windstorms, still got flood, still got quake in the South. And then in Latin America you’ve hurricane risk in the North, you’ve got quake risk down the West Coast. Why have I gone to the trouble of listing all these perils and their contribution to global property insurance risk? Sorry, can you guys hear me? It’s giving a warning. Okay. It’s because these are the risks that are available to you to offer geographic diversification to your insurance portfolio and so achieve the reduced capital requirements described in previous slides.

Some insurers can do a pretty good job of this by themselves. Companies like AXA or Allianz, or Generali, but for most companies, accessing good business from around the globe is going to impossible. Complying with the local regulation, accessing the right distribution channels, employing the right people to underwrite the risks around the world is an expensive and daunting prospect. But reinsurance companies offer a means to access this geographic diversification benefit worldwide. We’re able to achieve a greater capital efficiency than is possible on a more geographically concentrated portfolio, and so we reduce the capital needed by our cedents that we partner with. This should provide a win/win, low expense solution to both parties.

So now we get to what this might mean financially for an individual insurance company. We thought it’d be instructive to show a comparison of the differing costs of buying a cat treaty, an alternative of an insurer raising and holding enough capital to deal with cat risks by themselves with reinsurance protection. Slightly busy slide, sorry, but I’ll try and pick out the more important bits of this annual result for the two scenarios. They’re numbers for a fictional insurance company but are typical and realistic representations of reinsurance purchase. It shows a company that’s seeking to protect against the one or two in 200 year event, which is a typical regulatory requirement, basically a really big cat. The one in 200 for our fictional portfolio is $250 million, so Peak Re have proposed meeting this need with a property cat XL treaty of $225 million excess of $25 million. We think a typical market price might be three and a half, which is a cost of 7.9 million reinsurance premium.

And with the help of my friends in our reserving team, because I’m not expert in what I’m about to say, we then estimate that the capital that’d be required by regulator, think about attritional losses, cat risk, premium risk, and diversification credit, be 41 million if you were to buy the cat treaty, or a much bigger 232 million if you don’t have that reinsurance protection driven by that one in 200 year event number. Now normally insurers aren’t happy with barely meeting capital requirements imposed by regulators and often achieve a solvency ratio of about 200%. So double both these numbers to more closely mimic the real world. Meaning that’d be 82 million needed with a cat treaty and 465 million capital if you don’t buy the cat treaty.

The benchmarked RMS are another catastrophe model. Modeled loss estimates to do realistic stress tests of company result should there be a cat loss in the coming year of a one in 10, one in 25, and one in 200 year event. So that’s quite a small loss, a slightly bigger loss, but not unusual, and a really big loss. You can see in a loss-free year or in small event years, the P&L’s better without the cat treaty. But as soon as a slightly larger event occurs the reinsurance shows its worth. So one in 25 years, you can see that there would be a 36 million loss without the cat treaty. One in 200, a 260 million loss. Whereas with the cat treaty you’re limited your downsides significantly to just maximum loss of 16 million.

So I’m going to move on as we’re running out of time. Maybe the benefit of the cat treaty on annual result should be obvious, given that they’re designed to restrict volatility. But the resilience of balance sheet is also a fundamental concern. This slide continues to use information from our fictional company and repeats stress test to assess the impacts on the balance sheet. Here you can see the expected profits, losses from underwriting, profit from investment income, expenses, et cetera, et cetera. And then it looks at the capital solvency ratio at the end of each of our scenarios. The most important numbers are those at the bottom of the screen. The cat XL treaty insures the solvency ratio won’t fall below 200% in any of these loss scenarios, which is the target that’s here.

Profits are slightly reduced throughout the cat treaty in low-loss years. But if they were trying to hold capital sufficient to cover these losses themselves, you can see in large cat years there’s a severe impairment on their solvency, with dropping as low as 107% in a really big cat year, and even can impacted quite significantly, being dragged down to 184% in a relatively small year…

The final slide, so hopefully I’ll keep to my time limit. It compares the costs, benefits of holding capital or buying this property cat treaty. Access to capital isn’t free, as Chi-Hang discussed before, that attracts interest payments, shareholders with dividends, mutual members need bonuses. So looking at light green line on the top left graph is the average cost of capital for reinsurers over the past 15 years, which in recent years has hovered about 7%. I’m taking a slightly more optimistic assumption for my fictional company that it’ll cost them 5% to raise capital. We can benchmark their cost of capital given all the assumptions from previous slides against the cost of the cat XL. And again, the cat XL comes out as the preferable option with a total cost of 7 million for cat XL plus 4 million in finance costing being 48% or lower if you adjust the finance rate in the capital directly yourselves.

So I’ve skimmed over the end a bit there, sorry, to meet time requirements. But to summarize, basically diversification reduces the amount of capital required to ensure insurance risk. Reinsurance treaties transfer that risk to the reinsurer and off the reinsurance balance sheet, bringing up the insurer’s capital. Reinsurers are typically active globally and therefore enjoy the greatest geographic diversification benefit, reducing the cost of capital for the reinsurer and the insurer. And as a result, the cost of reinsurance treaty is often pretty competitive when benchmarked against the alternative solutions available to insurance companies.

Andy Souter:

So thanks Chi-Hang, thanks Andrew. Thankfully we had a few questions in, but in the interest of time, we’ll just pick out one question, and we’d be quite happy to answer the others offline. So this is a question which I’ll direct towards Chi-Hang, but Chi-Hang, how do you take into account BI, so business interruption, when you’re pricing property exposure? And Andrew also, if you want to take any view on BI as well.

Chi-Hang Wong:

Let’s take a cat treaty, actually when we model property risk. Normally we will ask cedents to provide their exposure for both building content as well as BI. That’s why in cat model it’s obviously taken into account. And for pricing, let’s say it’s proportional risk and normally our cedents will include their BI losses in their statistics as well. So I think the BI components are well covered from the quantitative side. And as well, I think normally in South Asian Pac we will have the largest exposure from BI point of view and then underwriters will take a view from the qualitative side as well. So I think that’s how reinsurers take BI into account when they do the pricing or underwriting view.

Andrew McGuinness:

Yeah, presuming the original coverage for BI doesn’t change year on year, thinking about Chi-Hang’s two fundamental approaches to pricing, experience based and exposure based, there should be an experience pricing already, so there should be enormous data and accounted for that way. If we were exposure rating, think if we’re using a cat model, BI is captured as a separate element in the sum insured, so it’s important that the data accuracy is good and that it’s entered on the right basis. So if there’s any… a year’s worth of potential BI, don’t accidentally enter overly conservative high numbers, it should be an accurate estimate of what the potential BI is.

Andy Souter:

Good. Thanks everyone.

Mike Ashurst:

Yeah, thanks. Thanks Andy, thanks Chi-Hang, and thanks Andrew for a great presentation. As Andy said, we will try to get the answers to those questions that we didn’t have time to answer right now.

The above text has been produced by machine transcription from the webinar recording. ICMIF has made every effort to ensure that transcriptions are as accurate as possible, however, in some cases some text may be incomplete or inaccurate due to inaudible passages or transcription errors. Listening to or watching the webinar recording will allow you to hear the full text as delivered during the webinar but this is available in English only. Our transcriptions are provided to enable members to select the language of their choosing using the dropdown menu above.