A reinsurance treaty example
This section presents one example of a reinsurance treaty as it appears in our data. As reinsurance treaties are very client-specific, this example is not intended to suggest that all reinsurance contracts have this structure. Table A1
shows the structure of a property damage treaty with, for each reinsurance tranche and year, the limit in excess of the priority, the recovery clause (if any), the number of bids sought, the number of reinsurers involved and the number of positive bids. For those reinsurers that offered positive bids, Table A1
also reports the average log of the rate-on-line, the average bid and the reinsurance dispersion.
Example of a reinsurance treaty covering property damage risks
Examining the evolution of the treaty through time, we see that the structure did not change for the first five years, keeping five tranches that have the same priority and limit year after year. Even the recovery conditions remained the same for those five years. However, the size of the reinsurance contract changed. This can be due to the number of individual contracts that are included in the primary insurer’s book of business or merely due to changes in the market value of the insured portfolio. The number of reinsurers that were asked to offer a quote varied by close to 50 per cent over the first five years, although the number of reinsurers that actually offered a strictly positive bid for a given tranche remained relatively steady over the years.
We observe in 2010 that the reinsurance treaty becomes a bit more flexible. Although the reinsurance treaty keeps the five-tranche structure that was present in the years 2005–2009, we see that tranches 1, 3 and 4 are split into two or three different parallel tranches. The parallel tranches differ with respect to the clauses (in particular the recovery clauses),31 which suggests a growing complexity of the reinsurance treaty between the years 2005 and 2010.
The table also presents the number of reinsurers that were contacted. Eighty-six per cent of the contacted reinsurers responded by saying they were interested in having more information about the risk that the primary insurer is looking to cede. Of these interested reinsurers, 70 per cent were willing to reinsure the tranche at some price. In the first two years, more than 85 per cent of the responses were positive. In the last four years, this rate fell to 60 per cent.
The fact that a reinsurer does not respond at all means either that he did not see the request, or that he is not interested to do business with the primary insurer in that year for that line of business or for that tranche of the reinsurance treaty, at whatever price. A reinsurer can refuse to bid for a risk because he may have already used up all his risk-capital or capacity for this type of risk. This is different from a bid of zero that is essentially the result of studying the portfolio of risks that the primary insurer is trying to reinsure and then deciding that the portfolio does not provide the reinsurer enough information to bid on the tranche. A reinsurer who refuses to bid after receiving the information provided by the primary insurer is not the same thing as a reinsurer who was contacted but did not care to even study the primary insurer’s proposal.
Looking at the reinsurance premiums, we observe how bids in each tranche evolve over the years. Firstly, we observe that the average rate-on-line decreases at the seniority of the tranche increases. Interestingly, the average bid seems to decrease as a function of the seniority of the tranche even though more senior tranches are generally larger. We see that the average bids (and a fortiori the average rate-on-line) reach their lowest level for a given treaty for a given year in most senior tranches, thus suggesting that the risks that these senior tranches will ever be used are very low. Take for instance year 2005 in Table A1; tranche 5 has an average log of the rate-on-line of 3.48, which translates into an average Pay Back of approximately 29 (meaning that the average reinsurer believes that losses in excess of 75 million dollars are likely to happen once every 29 years). In contrast, tranche 3 has an average log of the rate-on-line of 4.96, which translates into an average PayBack of approximately six years. In 2009, the same two tranches have lower rate-on-lines and higher payback. This either means that the market’s competitiveness has increased with the years or that, for some reason, it is now believed that the risk has decreased. It is interesting to see that it is in 2008, the year of the financial crisis, that for all the non-working tranches (i.e. for the tranches 3–5), the average bid was the lowest of all the years 2005–2009.
The second interesting bid feature that can be seen in Table A1 is that the dispersion of the bid, as measured by the standard deviation of the bids, generally decreases with the seniority of the tranche. But this reduction in dispersion may be only due to the fact that, on average, bids decrease in tranche seniority. When we normalise the standard deviation of the bids by the average bid, we see that the normalised dispersion of bids is distributed more or less homogeneously across the tranches with no discernible pattern within a given line of business.