The importance of modelling and forecasting of time series volatility for the control of economic processes

It can be observed that the volatility of high frequency time series (daily and higher frequency) has in certain time periods different level. The classical time series linear modelling methodology is based on condition of constant volatility. There are two conceptions of solution of the problem of the classical linear model. The first conception is based on idea that the problem is in the type of model of time series. Therefore R. Engle developed the class of ARCH models. The second conception is based on idea that the assumption of normality in models is not real in practise. In transmission of regulative intervention from one time series to another, the level of their volatility plays very important role. It is empirically analyzed the problem of intervention transmission from REPO interest rate to PRIBOR 1T, 1M, 3M and 1R rates. The GARCH models of volatility enables to construct the confidence intervals for conditional means and to define the effective intervention to the development of time series. The non effective intervention to PRIBOR 1T series does not have chance to be transmitted to another time series. Only the effective intervention can be transmitted. The realization of transmission depends on many factors among them, the expectations of changes of REPO interest rate play very important role.