Pages: [1] :: one page |
|
Author |
Thread Statistics | Show CCP posts - 0 post(s) |
Finideach
|
Posted - 2008.04.10 17:22:00 -
[1]
I want to keep the insurance discussions going but not derail existing threads with new topics and new types of insurance. "Business interruption" insurance could be valuable in Eve for those companies for whom unexpected downtime after a major patch means the risk of lost isk. I was doodling when I should've been paying attention this morning and came up with one approach to this kind of insurance.
Method: The insurer sells a capped number of shares at a fixed price and payout for each segment of time that they are willing to cover for Eve going *over* the announced downtime. These shares will pay out 10:1 IF the unexpected downtime falls within the segment the share covers.
These shares are bought and sold prior to major scheduled patching events and expected downtimes by companies who want to hedge potentially lost revenue.
For example, using simple numbers, the insurer offers 24,000 shares split into 24 1-hour time segments four weeks before an announced significant downtime that pay 10:1 if the downtime meets their time segment. The shares are arranged as follows:
** H +1 hour: Costs 1,000 isk per share and pays out 10,000 per share if the unexpected downtime is more than H but less than 1 hour. 100 shares available.
** H +2 hour: Costs 1,000isk, payout of 10,000isk for unexpected downtime between 61minutes & 2hours after H. 100 shares available.
....
** H +24 hours: Costs 1,000isk, payout of 10,000 isk for unexpected downtime between 1441 minutes & 25hours after H. 100 shares available.
In a perfect world assuming all shares are sold up front four weeks in advance. If Eve is unexpectedly down between for any time between 1minute and 25hours after the expected downtime the insurer will have to payout 10,000 per share to whomever holds the shares for that time period. If Eve is down for 26 hours - they pay nothing (but if players think that is likely none will buy shares).
Since the insurer took in 2.4m (100*24*1,000) and knows they have to pay out 1m (100*10,000) they have a predictable 1.4m margin and whatever interest they can earn on the entire 2.4m for the period of the month. (3% = 72,000) for a total of 1.472m profit.
If a company decided they really thought Eve would mess things up and bought all 100 shares for the 12th hour, and that's how long it was down, they would get all of the 1m for only putting up 100,000. Or they might buy a spread, 10 shares each in several different segments.
It's a hedge and a bit like options. But the second order effect is that it would allow players to then trade shares amongst themselves after the insurer has sold them - as players became more convinced Eve would take longer the shares of the lower time amounts would become less valuable - perhaps even less than 1,000isk. Conversely the shares of the time periods that are longer would become more valuable and perhaps sell and resell for more than 1,000isk.
I've used simple numbers for illustrative purposes but you know the drill - add in a few zeros and it starts to become real money.
Thoughts?
|
Tasko Pal
Heron Corporation
|
Posted - 2008.04.10 19:52:00 -
[2]
I like this. While we're at it, there's plenty of other things that one could insure against. For example, sovereignty and NPC outpost ownership changes. Major battles in inconvenient locations (as measured by ship loss rates, let's suppose) could disrupt production. Insure your POS's.
|
Finideach
|
Posted - 2008.04.11 01:50:00 -
[3]
The more complex the harder it would be to accurately evaluate - but given the size and cost - might be better to do a custom rating each time every time. CCP running over on downtime is a bit more predictable, and hek, it'd give CCP a free insight into customers confidence of how close they'll get to zero unscheduled extra downtimes on patch days. =)
|
KeeperOf Truth
|
Posted - 2008.04.11 02:42:00 -
[4]
Instead of an insurance more sounds like lets bet on extended downtimes. This is the same model betting houses operate.
|
Finideach
|
Posted - 2008.04.11 14:35:00 -
[5]
Betting and insurance are very similar when it comes down to it - an assessment of risk vs. odds.
|
Tasko Pal
Heron Corporation
|
Posted - 2008.04.11 18:37:00 -
[6]
Edited by: Tasko Pal on 11/04/2008 18:40:28
Originally by: KeeperOf Truth Instead of an insurance more sounds like lets bet on extended downtimes. This is the same model betting houses operate.
Yea, I think you're right. A betting market would be superior. The problem is getting enough liquidity so that you can insure effectively via such a market. Also, there is the slight matter that the market needs to be open when Eve is down, which really isn't a problem since any such market (like virtually all player services) almost surely is going to be outside of Eve.
Incidentally, if anyone is seriously considering building a betting market, I can provide advice for a reasonable fee on market design, available open source software, beta testing, and numerous pitfalls that can cost you profit and customers. I have experience with numerous play and real money betting markets, decent grasp of the theory behind betting markets (and relevant financial mathematics), and have some programming experience.
|
KeeperOf Truth
|
Posted - 2008.04.11 20:04:00 -
[7]
Originally by: Finideach Betting and insurance are very similar when it comes down to it - an assessment of risk vs. odds.
Yes, both have odds involved, but the idea behind them is not the same. With betting you are trying to guess something and make money out of it with insurance you are trying to protect yourself against something and %99 times you are happy if the event never happens. Lets give an example. If I unsure my house against robbery I get paid if my items are stolen, I am protecting myself against such an event and the odds are either my house will get robbed or not. Now if they told me to pick a window and if the robber enters the house from that window I will get pain ten times more than I paid as my insurance premium , it wouldn't be an insurance anymore. I would simply be trying to bet on which window thieves will use.
If you really want it to be an insurance it should be simple as this: If downtime exceeds X hours you will get insurance payment for covering the amount of time you can not play.
There is nothing wrong with your plan, its just not a real insurance.
Good luck |
Luckaro
|
Posted - 2008.04.11 20:43:00 -
[8]
I can not be sure but it seems to me that your insurance is not even remotely actuarly fair (even allowing for administration fees it seems off). My reasoning is mostly because you, as a insurer, will make a profit assuming (perhaps niavely) that each hour is equally demanded. Likely it will be that perhaps the 5 hours or so after downtime is scheduled to end will be in demand and, assuming that those five hours are equally demanded (perhaps a little less niave), you will end up losing money. [X shares per hour * 5 hours * 1000 ISK < X shares paid out * 1 hour * 10000 ISK -----> 5000 ISK hours < 10000 ISK hours]
If you are going to offer insurance, then you need to rethink your concept. Instead, you need to offer insurance on a sliding scale so that insurance for downtime with the hour after it is scheduled to end will be more than insurance for downtime 24 hours after it is scheduled to end. Find your expected value and price according to that maybe with a little extra to cover your time.
|
Finideach
|
Posted - 2008.04.11 21:36:00 -
[9]
Great feedback, thanks!
It sounds like the slotting mechanism might not be desirable - because as a customer I would have to A) predict if downtime would occur and B) pick a time frame I think it will end within. Neither of which has a direct relation to my losses *if* there is unexpected downtime.
From the insurers standpoint however this is a good system because they can limit their payouts and maintain control of loss (by limiting the shares for any available slot). Perhaps too good for the insurer and not enough for the insured. This may have been too close to a hedge or options to be desirable as insurance.
If I'm reading both of you correctly it may be better to have a method wherein the length of time on a sliding scale determines both costs and payout coverages. Is that correct? It might be a little more complicated to put together but I'll stake a stab at it.
As for whether betting and insurance are different I have to disagree with you Keeper of Truth. In both cases you have a provider (insurer or casino) who has calculated the odds of a specific or series of events happening - whether it is cars, rolling the dice or the chance someone will break into your house and steal your stuff. They then apply the law of large numbers to estimate a ratio where no matter how many times they pay out X - the averages will be in their favor to make a profit from all those who either don't win their bet.
This is why catastrophic events that turn every customer into a claimant (hurricanes, city-wide fires or riots) can be nightmares for insurance companies and they try and exclude exposure to those.
You may not realize you're "betting" in the real world when you buy property insurance - but the insurance company surely is and they know the odds. In both cases there is a risk of loss to the customer. In a casino you "bet" your ante and additional chips. In real life you "bet" the *potential* of risk against what you're willing to lose and compare that to monthly coverage.
Business interruption policies are even further removed from property policies in this concept. Movie studios buy insurance for loss of production costs if actors get sick, or that sub they're towing from Russia sinks in high tides. Behind every interaction is an assessment first of the odds by the insurer, the risk they take in offering coverage and the "bet" by the insured that they are willing to make that payment.
Good discussion, thanks!
|
Finideach
|
Posted - 2008.04.11 21:45:00 -
[10]
After further thought the slotting system might be a useful information tool - treated more as options or hedges - to see what the market thought the likelihood of downtime was and how long it would be. That information would then be valuable to the insurer in determining rates of a more traditional insurance policy with those who might lose business.
Also if you aren't going to used a fixed payout ratio - and instead wanted to opt for variable coverage - how much would you figure out a business is worth? If the insurer is going to provide 100m of coverage for every hour of downtime - what would be the methods to determine that the business normally generated 100m without heavy accounting and use of wallets?
|
|
Luckaro
|
Posted - 2008.04.11 21:49:00 -
[11]
A sliding scale would be good but payout would be constant since the loss incurred by the person buying the insurance is likely constant. However, the cost of the insurance would vary according to time elapsed after downtime. It is more likely that downtime will stretch over a hour or two but it is much less likely that downtime will be extended 24 hours. As such, you must assign the cost of insurance according to the probability per unit time of downtime occuring while keeping the payout if downtime occurs the same through all time periods. If one were to lose 10 million ISK per hour through missions or trading then it will not necessarly be related to when the downtime occurs (in the market transaction case it probably would but this needlessly comlicates insurance contracts).
The hard part will be in obtaining the probabilities of downtime. I hardly think that a linear scaling of prices would work except for, perhaps, the first few hours. After that I would scale prices down exponentially. Also, it might help to decrease the time unit on the insurance from one hour to, say, 15 minutes.
|
Finideach
|
Posted - 2008.04.11 22:02:00 -
[12]
The loss of rate per hour may be constant, but the accumulation of loss would be linear. If you lose 10m/hr then at 1hr additional your loss is 10m; at 5hr's it's 50m etc.
Even though the chances of it occuring go way down the losses to cover go way up if the unexpected downtime exceeds a large number of hours.
Would you reccomend capping the downtime hours, i.e. have a sliding scale to 24hours and then it's just 24hours +?
My other question mark is how to mitigate the risk to the insurance in terms of payout. If everyone who buys insurance gets a payout along each hour of unexpected downtime - then if Eve misses an hour you have 100% payout at that point. This is the kind of catastrophic event I was referring to above.
Thinking a bit - one way to do that is to allow the players to cap how much coverage they wish to buy, or at least when it starts. I have no idea on rates or anything right now but the two versions might be:
Capped by Length of Downtime For each X hour of unexpected downtime payout is Y per of coverage until Z downtime is reached.
Coverage determined by Start of Downtime Beginning after X hours of unexpected downtime payout is Y per 10m of coverage until Z downtime is reached.
Combine the two and you might have by way of example:
Customer A: Buying insurance at 10m/hr if unexpected downtime exceeds 3 hours through 10 hours.
Customer B: Buying insurance at 25m/hr if unexpected downtime exceeds 10 hours through 15hours.
Again I think there would be a problem where the insurer would have to carefully monitor who signs up to ensure there were enough funds available for payout. Unlike Geico who wants to sign up as many folks as they possibly can the insurer might want to sign up only a handful (and this would probably be high end insurance anyways).
Is that what you were thinking?
|
Luckaro
|
Posted - 2008.04.11 23:36:00 -
[13]
Both of your insurance plans sound good and each would have its own audience though I would think that the first one ("Capped by Length of Downtime") would be much more popular. One of the problems that you will face is finding a decent estimate of the probabilities of downtime for every 15 minutes. I am not sure if I can help you there as I have not kept track of downtime much.
You need to make the probabilities scuh that on average you should break eaven though that might mean the first few times you will lose a lot of money in this. To cover your losses, you might need at least one reinsurer depending on how much insurance you sell. This way, if you must payout, then you have sufficient funds to do so. The trick might be in getting a reinsurer.
Ideally, you have something of this nature. . .
[Person X] wants [X amount] of coverage for [X time]. [Person X] pays: [X amount] * [P{downtime in the first 15 minutes)] [X amount] * [P{downtime in the second 15 minutes)] + [X amount] * [P{downtime in the third 15 minutes)] + . . . + [X amount] * [P{downtime in the last 15 minutes)]
Note that [X amount] is the loss the person wants to be insured against per 15 minutes and [X time] is the number of 15 minute periods a person wants to be insured for. Now, the the insurance agent will weight the amount to be insured against the probability of downtime during that 15 minute period. The consumer agent will pay the sum of the resulting amounts. If the consumer agent buys [X time] of coverage but the downtime lasts for [Y time] of coverage (such that [X time] < [Y time]--that is, the downtime lasts for longer than the amount of time purchased), then the consumer agent only recieves up to the amount of time that was purchased (that is, [X time] and not [Y time]).
Let me know if you do not understand this. I will try to rephrase.
|
Finideach
|
Posted - 2008.04.12 00:32:00 -
[14]
I think in excel so I'll take a shot at this using numbers in a worksheet, then let you know if I have questions.
You've asked the great "which comes first, the chicken or the egg" question with reinsurers. My guess is that no one will step up to offer reinsurance until there are stable player run insurance companies active. Then again many insurance companies may not even start without a reinsurer program backing them from catastrophic loss. =)
|
KeeperOf Truth
|
Posted - 2008.04.12 01:14:00 -
[15]
How about starting simple? Simplifying your example (adjust numbers as you need) : If downtime exceed 3 hours insurance pay-out is 10m . One plan, one pay-out. It will make all calculations much easier for you. Then depending on customer demands you can adjust the policy or add different ones.
Complex plans are not only more difficult to manage for you, but also harder to explain potential customers at the beginning.
Keeper Of Truth is the alternative persona of Priest Amarr . He exist for company management and casual forum chatting. |
Finideach
|
Posted - 2008.04.12 18:48:00 -
[16]
Simple is good. =) Would you, as a customer, rather see a program that was only offered 100 of these policies (to limit risk to the insurer)?
Another question - what would be a fair way to determine when downtime ended? Does CCP post an official "back up"downtime and how would you qualify access to the game?
|
KeeperOf Truth
|
Posted - 2008.04.12 19:38:00 -
[17]
Edited by: KeeperOf Truth on 12/04/2008 19:40:57
I think it depends on what you mean by 100 policies. You can limit the number of clients you can have at any given time. I dont think anybody would blame you for not accepting more customers than you can handle but they would blame you if you accepted them and didnt pay.
Do you mean size of the policy per person? My understanding is you will start it as a small plot project at first, so I dont think it would be a problem if pay out amounts and premiums were low at the beginning. When/if you decide to launch this service premiums must be affordable and coverage amount should make sense. If total coverage will be 5 million isk , I am not sure if having an insurance for that amount worths time and effort. On the other hand if you start offering 50-100 million you might end up risking too much as your customer base grows.
Think this as a kind of insurance where you make a pay out soon or late. Like health insurance or life insurance. The event will occur at one point and you will end up making payments, the question is how you can make more money before event happens.
These are my thoughts but I hope more people contribute to this and we can examine different angles together.
And one question: It is good to have a solid example to talk on, but why downtime insurance? Do you think its a real need in game? I am interested in this because I want to see if player run insurances have any practical chance in Eve, but otherwise I am not sure if I need downtime insurance.. There might be other people who needs it though. Do you think if you can figure out how not to lose money on it, there is a customer demand for this policy?
* Keeper Of Truth is the alternative persona of Priest Amarr . He exist for company management and casual forum chatting * |
Finideach
|
Posted - 2008.04.13 00:15:00 -
[18]
There have been several discussions that have now rolled off the main page keeper.
The insurance discussion began when a few of us hijacked a thread on "why does CCP insurance act as it does" here.
Then some of us began a thread specifically on different types of "property" insurance - i.e. ships, fittings and modules. That's located here.
What I wanted to do with this thread is explore an entirely different type of insurance. Generally in insurance there's several main types:
* Property * Casualty * Liability * Business Loss * Life * Business Interruption
I started this thread to specifically talk about the last type. I am definitely interested in player run insurance companies - but I don't think I'm going to be the only one to give it a try so my hope is that by discussing different types we can work out some of the kinks and players might think "hey I might give that a try" or spark a thought on a different tangent.
Right now in the above threads and this one, given Eve's unique circumstances, I'm focusing on a very narrow scope of activities that are easier to prove actually happened. For example with a killmail (or using API to confirm it wasn't modified) one can confirm that a ship, fittings even cargo was actually destroyed. Not sure on implants - how can you prove you lost implants? Even with that "simplicity" it's proving to be pretty tricky to come up with good robust systems that prevent scamming on either side.
When it comes to business interruption there's obviously a lot more advanced applications out there. Loss of access to POS because of wardec, other interruptions, market manipulations, theft etc.; and not to the property itself - but to the loss of business resulting from the loss of access to or the property itself. For instance - maybe in some far distant land IPO's or Bond's would have to prove they not only had taken Property Insurance riders on their key assets - but also Business Interruption insurance so that investors had greater security that a hiccup (loss of POS, real life interruption, account suspended whatever) wouldn't derail the whole darn thing and cost them their returns.
The trick is - downtime is FAR easier to track and understand. Which is why I started with it.
But both property and business interruption kinds of policies have second and third order effect
An insurance company offering insurance to a new IPO is not going to work with a known, suspected or even whiffed at thief/scammer - and they'll have methods to ensure as much as possible before that. Likewise they're not going to insure against business interruptions unless there's a well thought out set of contingency plans.
In essence, in the ideal anarcho-capitalist society: viable insurers, banks and reinsures act in the capacity of government regulation - except because it's private, you have competition *within* the regulatory environment. |
Finideach
|
Posted - 2008.04.13 00:19:00 -
[19]
And yes - downtime insurance would definitely "pay out" eventually. Most insurance isn't predicated on never having to pay out - but rather using the funds to generate interest and investment income in excess of the eventual payouts. |
Finideach
|
Posted - 2008.04.15 00:37:00 -
[20]
So we have an opportunity with tomorrow's downtime to test drive this concept.
We'll use Keeper of Truth's simplified formula and use PLAY MONEY. Here's the deal.
Please post on this forum if you are ponying up 1m of NOT REAL MONEY as a "premium".
If Eve is down for longer than expected 4 hour downtime + three hours, i.e. it is down for 7 hours or longer you will receive 10m of NOT REAL MONEY.
This is an experiment only. I can't stress enough that this is NOT REAL MONEY. But we'll see how many pony up fake money and how much is owed to pay out after tomorrow's patch. (We still don't have a good method of determining when "back up" is but we'll work on it).
|
|
KeeperOf Truth
|
Posted - 2008.04.15 07:44:00 -
[21]
I am becoming your customer for this test , although I believe 10:1 is too risky for you. It is almost certain downtime will exceed announced hours in one of next ten patches. How about something like 10:3 (3 million premium for 10m policy) ?
Downtime results are easier to see compared to other type of insurances and less subject to scam, I agree , but my only worry is this type of insurance will be too risky for you either you will end up losing money or charging very high premium.
Still lets wait and see what will happen tomorrow.
* Keeper Of Truth is the alternative persona of Priest Amarr . He exist for company management and casual forum chatting * |
|
|
|
Pages: [1] :: one page |
First page | Previous page | Next page | Last page |