Description: This paper develops a new technique, which allows the analyst to maximally use all thte historical interest rate information available in forecasting interest rates. Interest rates' daily change do not follow a stationary random process over long time periods. Instead, the process is comprised of stationary periods, each a few years long, with relatively short transitional periods between different stationary periods.

Description: In this paper we give new general criteria for the weak convergence of one-dimensional distributions of generalized risk processes and describe the class of possible limit laws under an infinite growth of stochastic portfolio size and stochastic intensity of insurance payment.