The distributed lags models enable the construction of the lag mean, median
and variance. The estimators of parameters of the distributed lags models
and the autoregressive distributed lags models enable to create the estimat
ors of these basic characteristics. It is obvious that the Values of estima
tors depend on the form of the time series transformation. The methodology
of selection of suitable transformation of time series is based on the prin
ciple of maximization of the likelihood function. The computation of basic
characteristics of the lags in the econometric model is illustrated on the
example of the analysis of relationship between the interest rates on new g
ranted credits and 1R PRIBOR. This analysis revealed some changes in charac
ter of dependency of time series and changes of values of estimators of bas
ic lag characteristics in the period since 1993. In this connection the rec
ursive analysis gave valuable information.