dsengupt wrote: > > Hello, > > I am an undergraduate economics major, writing my thesis on flood > insurance protection for the Grameen Bank's loan portfolio. The > Grameen Bank is a financial institution in Bangladesh that has 2.5 > million borrowers spread across 41,000 villages. The Bank makes > "micro" loans to its borrowers, who use the loans to start their own > small enterprises. Given that Bangladesh is situated on one of the > largest deltaic plains in the world, 20%-33% of the country is subject > floods on a regular basis which adversely affect these borrowers' > enterprises. I am exploring the feasibility and price of insurance > "cover" that would allow the Bank to protect itself against default by > its borrowers after catastrophic floods such as the flood of 1998. > The 1998 flood affected two-thirds of the Bank's borrowers and limited > their ability to repay loans for nearly 10 weeks. > > I have historical data on floods in Bangladesh (1954-1998) and I also > have a basic insurance pricing model. I am looking for suggestions on > how to use Palisade Decision Tools "@Risk" program to fit this > historical data to a probability distribution function, and then use > the function to simulate a set of 5-6,000 events that could be used in > the model. > > Thank you for your help, > > Dev SenGupta > [EMAIL PROTECTED]
I don't have an answer for your immediate question but I do have some advice, which you may not need. Hydrologic events tend to have very heavy tailed distributions. Be sure you fit an appropriate distribution. Then be generous in considering the amount by which even your fitted distribution may underestimate the tail. You have less than 50 years of data, and I imagine you'll need reasonable estimates of at least a 100 year flood. It is difficult to get the latter from the former. U.S. insurance companies have taken some baths in recent years because they underestimated Mother Nature. Also, consider factoring in sea level increase. I believe considerable work has been done along these lines for the case of flooding potential for Venice, Italy. If you're not familiar with it, there might be some useful information to be gleaned from that work. Regards, Russell . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
