On 15 Feb 2003 19:16:32 -0800, [EMAIL PROTECTED] (dsengupt) wrote: >Hello, > > 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 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.
A brief galnce at Palisades' web page suggest their tool may not be best for the exceedances problem you face. For a contrast, see: http://www.math.chalmers.se/Stat/Research/extreme.html (quoting)Extreme events have large impacts throughout the span of engineering, science and economics. Extreme waves, rainfall, and floods are of basic importance in oceanograpy and hydrology, as are high wind speeds and extreme temperatures in meteorology. Much of risk and reliability theory is concerned with extreme occurrences. Methods and problems in statistical process control, and heavy-tailed models for internet traffic are closely related to extreme value theory (end quote) The book, "Rational Descriptions, Decisions and Designs" has a section on exceedances, explaining it in the context of using Bayesian statistics for describing probablistic distributions.. This book was originally published in 1969 by Pergamon Press. It went out of print some time after. A couple of years ago Expira of Sweden issued a reprint. Amazon indicates that a used version may be purchased for $175. Expira sells the reprinted version for less than $50. Write to Hakan Sodersved <[EMAIL PROTECTED]> for detailed information. Lastly, a search using the key prhase "extreme value theory" and hydrology will yield additional material on the subject. John Bailey http://home.rochester.rr.com/jbxroads/mailto.html . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
