According to the Koyck distribution in the lag distribution model:
the βq coefficients with the lag values of the explained variable are defined by the decreasing geometric progression:
, where α < 1.
Therefore, the distributed lag model looks as follows:
Multiply the equation calculated for the previous period of time t -1 to α coefficient:
Subtract the obtained equation from the original equation. As a result, obtain:
With this transformation, the equation with an infinite number of lags is transformed into an autoregression equation, for which only three coefficients are to be evaluated.
See also:
Library of Methods and Models | Distributed Lag Models | ISmLinearRegress.GDLTerms